inequality.stanford.edu · in this distribution over time (marshall, swift, and roberts 1997). as...

29
I ntergenerational mobility and socioeconom- ic inequality are two well-established topics in sociology. Although closely linked, these two aspects of the social distribution of resources and rewards are conceptually distinct (Hout 2004). Whereas inequality describes the distri- bution of resources at a particular point in time, mobility measures how individuals move with- in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari- ance of a particular distribution, and mobility refers to the intertemporal correlation (Gottschalk and Danzinger 1997). Thus, even if related, there is no necessary association between inequality and mobility. Furthermore, Unequal But Fluid: Social Mobility in Chile in Comparative Perspective Florencia Torche Queens College CUNY Columbia University A major finding in comparative mobility research is the high similarity across countries and the lack of association between mobility and other national attributes, with one exception: higher inequality seems to be associated with lower mobility. Evidence for the mobility–inequality link is, however, inconclusive, largely because most mobility studies have been conducted in advanced countries with relatively similar levels of inequality. This article introduces Chile to the comparative project. As the 10th most unequal country in the world, Chile is an adjudicative case. If high inequality results in lower mobility, Chile should be significantly more rigid than its industrialized peers. This hypothesis is disproved by the analysis. Despite vast economic inequality, Chile is as fluid, if not more so, than the much more equal industrialized nations. Furthermore, there is no evidence of a decline in mobility as the result of the increase in inequality during the market-oriented transformation of the country in the 1970s and 1980s. Study of the specific mobility flows in Chile indicates a significant barrier to long-range downward mobility from the elite (signaling high “elite closure”), but very low barriers across nonelite classes. This particular mobility regime is explained by the pattern, not the level, of Chilean inequality—high concentration in the top income decile, but significantly less inequality across the rest of the class structure. The high Chilean mobility is, however, largely inconsequential, because it takes place among classes that share similar positions in the social hierarchy of resources and rewards. The article concludes by redefining the link between inequality and mobility. AMERICAN SOCIOLOGICAL REVIEW, 2005, VOL. 70 (June:422–450) Direct all correspondence to Florencia Torche, Center for the Study of Wealth and Inequality, Columbia University, IAB, 420W 118th Street, Rm. 805B, New York, NY 10027 (fmt9@ columbia.edu). The author thanks Peter Bearman, Richard Breen, Nicole Marwell, Seymour Spilerman, Donald Treiman, the ASR editors, and four anonymous ASR reviewers for valuable advice. The author also thanks Paul Luettinger for pro- viding the CASMIN dataset, and Meir Yaish for providing the Israeli mobility table. This research was supported by the Ford Foundation Grant 1040–1239 to the Center for the Study of Wealth and Inequality, Columbia University, and by the Chilean National Center for Science and Technology FONDECYT Grant #1010474 to Guillermo Wormald, Department of Sociology Universidad Catolica de Chile. Delivered by Ingenta to Columbia University (cid 66000309) IP : 127.0.0.1 Mon, 12 Sep 2005 12:04:11

Upload: others

Post on 16-Mar-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

Intergenerational mobility and socioeconom-ic inequality are two well-established topics

in sociology Although closely linked these twoaspects of the social distribution of resourcesand rewards are conceptually distinct (Hout2004) Whereas inequality describes the distri-bution of resources at a particular point in timemobility measures how individuals move with-

in this distribution over time (Marshall Swiftand Roberts 1997) As expressed by standardstatistical concepts inequality refers to the vari-ance of a particular distribution and mobilityrefers to the intertemporal correlation(Gottschalk and Danzinger 1997) Thus even ifrelated there is no necessary associationbetween inequality and mobility Furthermore

UUnneeqquuaall BBuutt FFlluuiidd SSoocciiaall MMoobbiilliittyy iinn CChhiillee iinn CCoommppaarraattiivvee PPeerrssppeeccttiivvee

Florencia TorcheQueens College CUNY

Columbia University

A major finding in comparative mobility research is the high similarity across countries

and the lack of association between mobility and other national attributes with one

exception higher inequality seems to be associated with lower mobility Evidence for the

mobilityndashinequality link is however inconclusive largely because most mobility studies

have been conducted in advanced countries with relatively similar levels of inequality

This article introduces Chile to the comparative project As the 10th most unequal

country in the world Chile is an adjudicative case If high inequality results in lower

mobility Chile should be significantly more rigid than its industrialized peers This

hypothesis is disproved by the analysis Despite vast economic inequality Chile is as

fluid if not more so than the much more equal industrialized nations Furthermore

there is no evidence of a decline in mobility as the result of the increase in inequality

during the market-oriented transformation of the country in the 1970s and 1980s Study

of the specific mobility flows in Chile indicates a significant barrier to long-range

downward mobility from the elite (signaling high ldquoelite closurerdquo) but very low barriers

across nonelite classes This particular mobility regime is explained by the pattern not

the level of Chilean inequalitymdashhigh concentration in the top income decile but

significantly less inequality across the rest of the class structure The high Chilean

mobility is however largely inconsequential because it takes place among classes that

share similar positions in the social hierarchy of resources and rewards The article

concludes by redefining the link between inequality and mobility

AAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW 22000055 VVOOLL 7700 ((JJuunnee442222ndashndash445500))

2322-ASR 703 filename70304-torche

Direct all correspondence to Florencia TorcheCenter for the Study of Wealth and InequalityColumbia University IAB 420W 118th StreetRm 805B New York NY 10027 (fmt9columbiaedu) The author thanks Peter BearmanRichard Breen Nicole Marwell SeymourSpilerman Donald Treiman the ASR editors andfour anonymous ASR reviewers for valuable adviceThe author also thanks Paul Luettinger for pro-

viding the CASMIN dataset and Meir Yaish forproviding the Israeli mobility table This researchwas supported by the Ford Foundation Grant1040ndash1239 to the Center for the Study of Wealthand Inequality Columbia University and by theChilean National Center for Science andTechnology FONDECYT Grant 1010474 toGuillermo Wormald Department of SociologyUniversidad Catolica de Chile

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

some analysts have argued that growing inequal-ity can be offset by a rise in mobility tacitlyassuming that these two distributional phe-nomena can move in opposite directions(Friedman 1962)

At the theoretical level approaches linkingmobility and inequality tend to focus on micro-level mechanismsmdashthe motivations andresources affecting individual decision mak-ingmdashconnecting these macrostructural phe-nomena Two perspectives can be distinguishedA ldquoresource perspectiverdquo indicates that highinequality will result in decreased mobilitybecause the uneven distribution of resourceswill benefit those most advantaged in the com-petition for success In contrast an ldquoincentiveperspectiverdquo argues that inequality will raisethe stakes in the competition thereby inducinghigher mobility

In the end the question about the link betweenintergenerational mobility and socioeconomicinequality is empirical and currently the empir-ical evidence is thin and inconclusive Part ofthe problem is that mobility has been studied ina small pool of mostly industrialized countriesall of which share relatively similar levels ofinequality To explore the link between mobil-ity and inequality at the empirical and theoret-ical levels this article introduces Chile to thecomparative mobility project

Chile is a middle-income country that hasundergone significant political and economicchange in the past few decades In the midndash20thcentury the Chilean development strategy wasdefined by import-substitutive industrializa-tion The economy of Chile was closed to inter-national markets and the state had a pivotaleconomic and productive role Mounting socialproblems associated with unequal developmentand urban migration led to two progressiveadministrations in the 1960s and early 1970sThese administrations conducted major redis-tributive reforms including the nationalizationof enterprises an educational reform and anagrarian reform This progressive path was vio-lently halted by a military coup in 1973 Led byGeneral Pinochet the military took power andretained it until 1990

Once in power the authoritarian regime con-ducted a deep and fast market-oriented trans-formation consisting of the now standardpackage of macroeconomic stabilization pri-vatization of enterprises and social services

and liberalization of prices and markets As aresult Chile transformed from a closed econo-my with heavy state intervention into one of themost if not the most open and market-basedeconomy of the world

After a deep recession in the early 1980sChile has experienced substantial and sustainedeconomic growth since the late 1980s whichconcurs with the reestablishment of democrat-ic rule The period of redemocratization andgrowth during the 1990s has led to significant-ly improved living standards for the Chileanpopulation The dark side of this success storyis the persistent economic inequality in thecountry Inequality historically very high grewduring the military government and Chile cur-rently ranks as the 10th most unequal countryin the world The Chilean pattern of inequalitycan be characterized by ldquoconcentration at thetoprdquo Chile is highly unequal because thewealthiest segment of the society receives avery large portion of the national incomewhereas the differences between the poor andmiddle-income sectors are much less pro-nounced lower even than in some industrializednations Although inequality is by definitionassociated with concentration the Chilean caseis extreme as compared with the industrializedworld and even with other Latin Americannations

These features render Chile an adjudicativecase If mobility and inequality are relatedChile should show mobility rates significantlydifferent from those of its industrialized peersTo explore these issues my analysis includesthree components

First I analyze the Chilean mobility regimein a comparative perspective Mobility is here-in defined as the association between class oforigin (fatherrsquos class position) and class of des-tination (current class position) net of thechanges in the class structure over time (struc-tural mobility) Also known as ldquorelative mobil-ityrdquo or ldquosocial fluidityrdquo mobility represents thelevel of openness or degree of equality of oppor-tunity in a society I place Chile in the compar-ative context by fitting the ldquocore model of socialfluidityrdquo to the Chilean intergenerational mobil-ity table This model claims to represent thebasic similarity in mobility across countriesbut it has been tested in only a few mostlyindustrialized countries (see for exampleErikson and Goldthorpe 1992a 1992b) By fit-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442233

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ting the core model to the Chilean table I deter-mine to what extent Chile departs if it doesfrom the assumed international homogeneityin mobility regimes and what the sources of itspossible departure are Even if the core modelhas become a milestone in comparative analy-sis it has a number of undesirable propertiesdiscussed later in the Methods section Thus asa modeling alternative I use a hybrid model thatcombines association parameters assuming ahierarchical conception of the mobility structurewith patterns accounting for class immobilityInterpretation of these two models provides adetailed description of the Chilean mobilityregime and an evaluation of the main factorsmdashhierarchical differences across classes classinheritance barriers across sectors of the econ-omymdashdriving Chilean mobility dynamics

Second I compare the level of mobility inChile with that of other countries using pri-mary data from France England ScotlandIreland Sweden the United States and IsraelAll these countries have a much more egalitar-ian income distribution than Chile Thus ifmobility is associated with inequality the inter-national comparison should show that mobili-ty opportunities in Chile are signif icantlydifferent from the opportunities in these indus-trialized nations

Third my analysis moves from an interna-tional to a temporal comparison To explore theassociation between mobility and inequalityfurther I study Chilean mobility rates over timeUsing a cohort analysis I examine whether thesignificant increase in inequality associatedwith the market transformation of the 1970sand 1980s had any noticeable effect on Chileanmobility rates

Advancing some of the major results thefindings present an interesting paradox On theone hand the Chilean mobility regime is foundto be driven almost exclusively by the hierar-chical distance between classes which is deter-mined in turn by the level of inequality in thecountry This finding is consistent with the neg-ative relationship between mobility and inequal-ity posed by the ldquoresource perspectiverdquo On theother hand the international comparison indi-cates that Chile is impressively fluid despite itshigh economic inequality and the temporalanalysis shows no change in Chilean mobilityover time despite the growing inequality duringthe military regime Therefore the internation-

al and temporal comparisons seem to contradictthe negative relationship between mobility andinequality and to suggest that high inequalitymay in fact induce mobility

Solving this apparent paradox I arguerequires switching the analytical focus from thelevel to the pattern of both mobility and inequal-ity When the pattern of these two distributivephenomena is considered the contradiction dis-appears and the mechanisms driving the mobil-ityndashinequality association can be examined

The argument is organized as follows Section2 presents the empirical and theoretical evi-dence concerning the association of mobilitywith inequality Section 3 describes the Chileancontext It discusses the most salient character-istics of the Chilean socioeconomic structureand presents a brief historical description ofthe major changes in the national political econ-omy over the past few decades Section 4 intro-duces the data and analytical approach Section5 presents the comparative analysis of mobili-ty in Chile including the description of theChilean mobility regime the international com-parison of the fluidity level in Chile and theanalysis of change over time Section 6 presentsa summary of the findings the conclusionsand the implications of the Chilean case for thecomparative study of inequality

AASSSSOOCCIIAATTIIOONN BBEETTWWEEEENN IINNEEQQUUAALLIITTYYAANNDD MMOOBBIILLIITTYY EEMMPPIIRRIICCAALL AANNDDTTHHEEOORREETTIICCAALL EEVVIIDDEENNCCEE

Although mobility and inequality ldquogo togetherintuitivelyrdquo (Hout 2004969) they are differentdimensions of the social distribution of advan-tage The former refers to the cross-sectionaldimension which determines the individualposition in a social hierarchy The latter refersto its intergenerational dimension specificallyto differential access to these positions as deter-mined by the position of origin (Marshall andSwift 1999243 Marshall et al 199713)

At the conceptual level the differencesbetween these two distributional phenomenaare clear Inequality describes the distributionof resources at any particular point in timeMobility describes inequality of opportunitythe chances that someone with a particular socialorigin will attain a more rather than a less advan-taged destination regardless of the socioeco-nomic distance between these destinations This

442244mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

is why ldquoindividuals are anonymous for inequal-ity but not for mobilityrdquo (Behrman 199972) Infact mobility analysis assumes path dependencein the reproduction of the social structure acrossgenerations and studies the extent and mor-phology of such intergenerational dependence

The conceptual difference may have impor-tant practical implications As argued byFriedman (1962) and echoed more recently bysome researchers and policymakers a givenextent of income inequality under conditions ofgreat mobility and change may be less a causefor concern than the same degree of inequalityin a rigid system wherein the position of ldquopar-ticular families in the income distribution doesnot vary widely over timerdquo (p 171)

The key question is whether these two dis-tributional phenomena are independent Twotheoretical approaches address this questionBoth focus on the micro-level mechanisms thatpurportedly link these macrophenomena theindividual decision-making processes Theldquoincentive approachrdquo claims that the motiva-tion to pursue mobility is proportional to theamount of cross-sectional inequality If inequal-ity approaches zero so does the payoff formobility Conversely great inequality increas-es both the inducement to pursue mobility forthose who are initially disadvantaged and theincentive to resist mobility for those who are ini-tially advantaged In other words inequalityraises the stakes of mobility (Hout 2004970 seealso Tahlin 2004)

The ldquoresource approachrdquo contends thatmobility depends critically on resources ratherthan on incentives The higher the inequality thegreater the distance in terms of human finan-cial cultural and social resources across dif-ferent social origins In a highly unequal societythe stakes of mobility will be high across theboard but the crucial resources controlled byindividuals will be so unevenly distributed thatcompetition certainly will benefit those moreadvantaged (see Goldthorpe 2000254 Stephens197954 and Tawney 1965 for a seminal elab-oration of this argument) The outcome of highcross-sectional inequality will be the rigid repro-duction of the class structure over time ielow social fluidity

Because resources and incentives work indifferent directions the effect of inequality onmobility can be thought of as an additive cal-culation If the impact of resources outweighs

that of incentives we should expect a negativeassociation between inequality and mobilityConversely the association should be positiveif incentives are more relevant than resources inthe competition for advantaged positions Inwhat follows I argue that consideration ofresources and incentives affecting mobility at themicro-level should be preceded by a macro-level understanding of both mobility andinequality This understanding should considerthe pattern of these two distributional phenom-ena not only their overall level and shouldfocus on the location and depth of the majorcleavages in the social structure

TTHHEE EEMMPPIIRRIICCAALL EEVVIIDDEENNCCEE

Empirical evidence concerning themobilityndashinequality association is scarceAlthough high-quality data on inequality existfor a large number of countries (eg Deiningerand Squire 1996) comparable data on mobili-ty are available only for a few mostly industri-alized nations The most important source ofcomparable mobility data is the ComparativeAnalysis of Social Mobility in IndustrializedCountries (CASMIN) project which includes 15industrialized nations1 The findings of theCASMIN project are highly relevant for theassociation between mobility and inequalityThe CASMIN project concluded that whereasabsolute mobility varies across countries andover time because of differences in nationaloccupational structures relative mobilitymdashtheassociation between class of origin and desti-nation net of differences in marginal occupa-tional distributionsmdashis extremely homogeneousacross countries and over time These findingswere expressed in the ldquocommonrdquo and ldquocon-stantrdquo social fluidity hypotheses respectivelyThe basic temporal and international similari-ty was systematized into a ldquocore modelrdquo ofsocial fluidity (Erikson and Goldthorpe 1987a1987b 1992a see Featherman Jones andHauser 1975 and Hauser et al 1975a 1975b for

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442255

2322-ASR 703 filename70304-torche

1 These include 12 European countries (EnglandFrance the former Federal Republic of GermanyHungary Ireland Northern Ireland Poland ScotlandSweden the former Czechoslovakia Italy and theNetherlands) and three industrial non-European coun-tries (the United States Australia and Japan)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

the original hypotheses) Common social flu-idity is claimed to hold (at least) in industrialsocieties with a ldquomarket economy and nuclearfamilyrdquo (Featherman et al 1975340) and itwould be explained by ldquothe substantial uni-formity in the economic resources and desir-ability of occupationsrdquo (Grusky and Hauser1984)

The significant international similarity inmobility patterns has been confirmed largelyin pairwise or three-country comparisons in afew industrialized countries EriksonGoldthorpe and Portocarero (1979) Eriksonet al (1982) and Hauser (1984a 1984b) forEngland France and Sweden McRoberts andSelbec (1981) for the United States andCanada and Kerckhoff et al (1985) forEngland and the United States As more indus-trialized countries were added to the compar-ative template it was shown that Sweden andthe Netherlands are more fluid (Erikson et al1982 Ganzeboom and De Graaf 1984) andthat Germany Japan and Ireland are morerigid (Hout and Jackson 1986 Ishida 1993Muumlller 1986) However the most striking find-ing was the significant similarity across coun-tries despite the different historicalbackgrounds and institutional arrangements

On the basis of the CASMIN project a rel-ative consensus has emerged favoring thehypothesis of substantial international simi-larity in mobility patterns Similarity does notmean complete homogeneity however Therecertainly exist some national deviations butit is argued they can be explained by highlyspecific historically based institutional factorsrather than by systematic relationship to othervariable characteristics of national societiesThere seems to be only one exception to thisldquounsystematic fluctuationrdquo Comparison of theCASMIN countries found that a small but sig-nificant portion of the international variationin mobility is related to economic inequalitywith higher inequality leading to less fluidity(Erikson and Goldthorpe 1992a chapter 12also see Tyree Semyonov and Hodge 1979for an earlier analysis)

This finding supports the resource approachlinking the two distributive phenomenaAdditional evidence consistent with theresources approach was presented by Jonssonand Mills (1993) who compared Sweden andEngland and found higher fluidity in the more

equal Swedish society by Bjorklund and Jantti(1997) who compared earnings mobility inthe United States and Sweden and found mobil-ity to be higher in Sweden and by analyses ofintergenerational earnings mobility in severalOrganization for Economic Cooperation andDevelopment (OECD) countries which sug-gest higher mobility in more equal nations(Solon 2002)2

However a more recent comparative analy-sis of mobility trends between 1970 and 1990in 11 industrialized countries does not findevidence of a relationship between inequalityand fluidity (Breen and Luijkx 2004a396)This supports an earlier indication of no asso-ciation between the two distributional phe-nomena (Grusky and Hauser 1984)

In sum evidence about a potential mobili-ty-inequality association based on cross-coun-try comparisons is inconclusive not onlybecause of divergent f indings but alsobecause of the small number of countriesincluded and the potentially high collinearitybetween inequality and other explanatory fac-tors

Empirical analysis of mobility trends with-in countries has not provided a conclusiveanswer either The CASMIN finding of ldquocon-stant social fluidityrdquo over time has been recent-ly tested in the aforementioned comparativeanalysis of mobility in 11 countries Thisanalysis including 10 European nations andIsrael (Breen 2004) finds growing fluidity insome countries but null or slight temporalchange in others While Britain Israel andless conclusively Germany display ldquoconstantfluidityrdquo some indication of growing open-ness is detected in France Hungary IrelandItaly the Netherlands Norway Poland andSweden However in France Hungary Polandand Sweden all change occurred between the1970s and 1980s and stability has prevailedsince then (Breen and Luijkx 2004b54)Changes in Ireland and Italy are quite minor(Layte and Whelan 2004 Pisati andSchizzerotto 2004) and only the Netherlandsdisplays a sustained increase in mobility overthe entire period considered (Ganzeboom andLuijkx 2004) Mobility has also been found to

442266mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

2 To my knowledge there is no empirical evidencesupporting the incentive approach

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

increase over time in the United Statesbetween the early 1970s and the mid 1980s(Hout 1988) However these increases in flu-idity are not univocally preceded by a reduc-tion of inequality thus casting doubts on apotential association between inequality andmobility

A careful analysis of mobility trends inRussia shows a significant decline in mobili-ty associated with an increase in inequalityafter the market transformation of the early1990s (Gerber and Hout 2004) which pro-vides additional support for the resourceapproach The advantage of the Russian studyis that ldquothe market transition in Russia ||altered so many fundamental economic insti-tutions so rapidly that we can confidentlyascribe changes in social mobility || to thissource rather than to cultural change or indus-trializationrdquo (Gerber and Hout 2004678) It isnot clear however whether it is growth ininequality recession some other change asso-ciated with the radical liberalization of theeconomy or a combination of these factorsthat triggered a decline in fluidity in contem-porary Russia

In this context Chile presents an ideal casefor an examination of the association betweeninequality and mobility Given the extremeeconomic inequality in the country if theunequal distribution of resources or incentiveshas an impact on mobility opportunities theChilean level of fluidity should be signifi-cantly different from that of industrializednations Additionally given the increase ininequality associated with the market reformof the 1970s and 1980s analysis of mobilitybefore and after the reform would provide sup-plementary evidence concerning the potentialmobilityndashinequality relationship Naturally asingle case study will not supply a definitiveanswer to these questions However by com-bining an examination of the Chilean mobili-ty regime in the context of historicaltransformations in the country with an inter-national comparative analysis and with anassessment of mobility trends over time thisarticle provides important insights into theexistence of a link between mobility andinequality and the mechanisms driving it

TTHHEE CCHHIILLEEAANN CCOONNTTEEXXTT IINNCCOOMMPPAARRAATTIIVVEE PPEERRSSPPEECCTTIIVVEE

During the second half of the 20th centuryChile transformed from an agrarian semifeu-dal society into an urban service-based oneBetween 1950 and 2000 the rural populationdeclined from almost 40 to 17 percent (Braunet al 2000 INE 2002) This defines Chile asa mostly urban country with an 83 percentrate of urbanization larger than the 78 percentrate in the United States In tandem with urban-ization Chile experienced a reallocation ofemployment from the agricultural to the terti-ary sector of the economy The share of agri-culture in total employment declined from 38percent in 1950 to 17 percent in 2000 Whereasthe share of manufacturing remained constantat about 18 percent the share of the servicesector rose from 42 to 65 percent (Braun et al2000 INE 2002)

Urbanization and tertiarization are process-es that virtually all countries experienced dur-ing the 20th century Within this secular trendthe Chilean political economy is marked byspecific institutional developments that shapeits stratification structure From the 1940s tothe 1970s the Chilean economic landscape asthat of its Latin American neighbors wasdefined by import-substitutive industrialization(ISI) Emerging as a reaction to the collapse ofinternational trade caused by the GreatDepression (Ellworth 1945) and based on theldquodeterioration of the terms of traderdquo theory(Prebisch 1950) ISI was based on two typesof policies The first was oriented to closing theChilean economy to international markets andthe second was oriented to promoting nation-al industrialization The Chilean state becamethe leading productive agent supporting indus-try through credit investments and technicalassistance and taking a direct productive rolethrough the creation of public enterprises(Stallings 1978) After a sanguine beginningwith industrial production growing at almost7 percent per year between 1940 and 1950(Mamalakis 1976 Munoz 1968) ISI started tofail economic growth stagnated and socialturmoil resulting from massive urban migra-tion and vast social inequalities increased

In 1964 a progressive administration tookpower and adopted as its mandate the correc-tion of extreme inequality in the country by a

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442277

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ldquoRevolution in Freedomrdquo (Gazmuri 2000)This progressive government launched redis-tributive policies including an agrarian reformand an educational reform The redistributiveagenda boosted by a socialist administrationthat came to power in 1970 was abruptly halt-ed by a military coup in 1973 The militarytook power and retained it until 1990 Duringthese 17 years the military regime conduct-ed a deep market-oriented transformationNow turned into the ldquoWashington consensusrdquoparadigm (Williamson 1990) this reformincluded macroeconomic stabilization dereg-ulation of prices and markets and the priva-tization of enterprises and social services Ittransformed Chile from a closed economywith heavy state intervention into one of themost open market-based economies in theworld with the productive and welfare role ofthe state reduced to a minimum (Edwards andCox-Edwards 1991 Martinez and Diaz 1999Meller 1996 Velasco 1994)

The depth of the market reform coupledwith a world recession led in the late 1970sand early 1980s to the deepest economic cri-sis since the Great Depression A third of thelabor force was unemployed and povertyafflicted nearly half of Chilean households(Meller 1991) The post-crisis recovery start-ing in the late 1980s was substantial and sus-tained and coincided with theredemocratization of the country The grossdomestic product (GDP) per capita grew morethan 6 percent annually for 15 years a novelrate for Chile only comparable in recent timesto the ldquoEast Asian Miraclerdquo (World Bank1993) The sharp economic growth has trans-formed Chile from one of the poorest coun-tries in Latin America (Hofman 2000) into aldquomiddle-income economyrdquo (World Bank2003a) In the year 2000 the income per capi-ta was approximately US$5000 much lowerthan the United States average of US$31910but the highest in Latin America (World Bank2003a) As a consequence of the sharp eco-nomic growth Chileans have reached levels ofconsumption unthinkable two decades agoand the poverty rate fell from 45 to 21 percentbetween 1985 and 2000 (MIDEPLAN 2000Raczynski 2000)

TTHHEE CCHHIILLEEAANN LLEEVVEELL AANNDD PPAATTTTEERRNN OOFF

IINNEEQQUUAALLIITTYY

The dark side of this ldquosuccess storyrdquo is eco-nomic inequality Structurally rooted in a feu-dal agrarian structure the institutional legacyof the colonial period and in the slow expan-sion of education (Engerman and Sokoloff1997 World Bank 2003b) inequality hasremained persistently high during the 20thcentury The Gini coefficient reached 58 in the1990s which compares with a much lowerGini of 34 among the industrialized coun-tries and is large even in the highly unequalLatin American context with its average Giniof 49 (Deininger and Squire 1996 Marceland Solimano 1994) As Figure 1 indicatesChilean inequality is almost twice that in mostindustrialized countries and 15 times that inthe United States the most unequal nation ofthe industrialized world

Not only the level but also the pattern ofinequality in Chile significantly departs fromthat of the industrialized world With thewealthiest Chilean decile receiving 423 per-cent of the total national income (MIDEPLAN2001) the Chilean pattern of inequality ischaracterized by high ldquoconcentration at thetoprdquo Although inequality is by definition relat-ed to concentration the Chilean case isextreme as compared with the industrializedworld and even with other Latin Americancountries A comparison between the incomeof each decile and the income of the preced-ing (poorer) decile illustrates the point Theratio between the wealthiest and the secondwealthiest decile is twice as large in Chile asin the United States and England and one ofthe largest in Latin America depicting highelite concentration In contrast the ratiobetween the second poorest and the poorestdeciles in Chile is half that of the United Statesand England indicating that inequality at thebottom of the income distribution is muchlower in Chile than in these industrializednations (Szekely and Hilgert 1999)

In fact as Figure 2 indicates Chile is thefourth most unequal country in the mostunequal region of the world However if thewealthiest decile is excluded Chilean inequal-ity is dramatically reduced and Chile becomesthe most equal Latin American country evenmore equal than the United States (Inter-American Development Bank 1999)

442288mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442299

2322-ASR 703 filename70304-torche

Figure 1 Income Inequality in Chile and Industrialized Countries Circa 2000

Note Jap = Japan Swe = Sweden Ita = Italy Ger = Germany Neth = Netherlands Fra = France Ire = Republicof Ireland Eng = England Source World Bank 2001

Figure 2 Gini Coefficient for Total Population and Excluding Wealthiest Decile Chile other Latin AmericanCountries and Unites States in 1998

Note Par = Paraguay Bra = Brazil Ecu = Ecuador Arg = Argentina Bol = Bolivia Ven = Venezuela Uru =Uruguay Source Inter-American Development Bank 1999

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Therefore Chile is unequal largely because theelite concentrates an extremely high proportionof the national income Across the noneliteclasses the distribution of resources is muchmore uniform

Inequality is not a new development inChilean society but rather has deep historicalroots Figure 3 presents the longest availableseries on inequality in Chile depicting earningsdistribution since 1957 in Santiago3 and show-ing persistently high inequality over the lasthalf century However there is significant vari-ation over time a short decline of inequality dur-ing the progressive administrations of themid-1960s and early 1970s an increase since themilitary regime took power in 1973 a peak in1984 and a small decline after the democratictransition to levels still higher than those thatpreceded the military regime

LLAATTEE IINNDDUUSSTTRRIIAALLIIZZAATTIIOONN MMAARRKKEETTRREEFFOORRMM AANNDD TTHHEE CCHHIILLEEAANN CCLLAASSSSSSTTRRUUCCTTUURREE

The ldquocommon social fluidityrdquo hypothesis assertsthat all international variation in mobility pat-terns is attributable to highly specific histori-cally formed national characteristics In theChilean case these factors may be related to themarket reform of the 1970s and 1980s and therapid economic growth since the late 1980sNational mobility studies suggest that coun-tries experiencing rapid industrialization may becharacterized by weaker sector barriers to mobil-ity specif ically barriers between the self-employed and employees and betweenagricultural and nonagricultural classes(Goldthorpe Yaish and Kraus 1997 and Yaish2004 for Israel Ishida Goldthorpe and Erikson1991 for Japan Park 2004 for Korea and Costa-Ribeiro 2003 for Brazil)

In addition two components of the Chileanmarket transformation may have contributed tothe weakness of mobility barriers across sectorsof the economy The first component is theagrarian reform and subsequent counterreformundertaken by the military regime The secondone is the opening of the economy to interna-tional trade and the retrenchment of the statewhich led to a decline in formal employment

443300mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Figure 3 Earnings Inequality in Greatest Santiago 1957ndash2000Source Larranaga 1999 and calculations by the author

3 Santiago the Chilean capital comprises aboutone-third of the countryrsquos population Assessmentby Chilean experts suggests that Santiago trendspresent an unbiased picture of trends at the nationallevel Although earnings inequality is not as com-prehensive a measure as total income inequality it isan adequate much more accurately measured proxy(Galbraith 2002)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

and to the growth and segmentation of the self-employed sector

The main objective of the Chilean agrarianreform (1962ndash1973) was to reduce the extremeland concentration in the country The reformwas very successful In 1955 the top 7 percentof landowners held 65 percent of the land andthe bottom 37 percent held only 1 percent By1973 the last year of the reform 43 percent ofthe land had been expropriated (Kay 2002 Scott1996) An agrarian counterreform was launchedby the military regime as a way of returning theland to its former owners It did not howeverrestore the traditional hacienda order (Gomezand Echenique 1988 Rivera 1988) In fact themilitary government returned only one-third ofthe plots to their old owners sold another thirdand adjudicated the remaining third to the smallproprietors benefited by the agrarian reformUnable to compete with the now cheap foodimports many small proprietors opted to selltheir plots As a consequence as much as two-thirds of the Chilean agricultural land came upfor sale creating an active land market Themain beneficiaries of the ldquomarketization of thecountrysiderdquo were a new group of export-ori-ented entrepreneurs and international investorswho bought large tracts of land (Gwynne 1996Kay 2002) Thus the reform and counterreformtriggered a massive ldquochange of handsrdquo in theChilean countryside likely altering the patternsof land inheritance

The second major effect of the market reformon the class structure was the reduction of for-mal employment caused by the decline of theindustrial working class and by the shrinkage ofthe state Incapable of competing with the nowcheap imports the working class plummetedfrom 34 percent of total employment in theearly 1970s to 20 percent in 1980 while pub-lic employment dropped from 14 to 84 per-cent in the same period (Leon and Martinez2000 Schkolnick 2000 Velasquez 1990) Thedecline in industrial and public employmentled to an increase in self-employment fromabout 15 percent in 1970 to about 28 percent in1980 (Thomas 1996) followed by a declinethrough the economic recovery before stabi-lization at about 22 percent of the total employ-ment in 2000 (MIDEPLAN 2001) In contrastto industrialized countries in which the self-employment rate usually is within the one-digitrange and involves capital ownership in Chile

it encompasses more than one-fifth of the laborforce Furthermore self-employment hadbecome a survival strategy for a large numberof former industrial and public employeesappropriately labeled ldquoforced entrepreneurial-ismrdquo (Infante and Klein 1995 Portes Castellsand Benton 1989) As a consequence theChilean self-employed sector is voluminousand segmented It includes a small segment ofentrepreneurs oriented to capital accumulationand able to hire employees and a large seg-ment of self-employed workers mostly engagedin ldquosurvival activitiesrdquo and unable to hire thelabor of others Differences in economic well-being between these two groups are massivewith the former group earning on averagethree times more than the latter (MIDEPLAN2001)

To be sure a large self-employed sector is nota result of the market transformation Self-employment has been historically high in LatinAmerica because according to some scholarsthe region has experienced ldquodependent inte-grationrdquo into the world capitalistic system(Hopkins and Wallerstein 1982 Luxembourg1951) However the internal heterogeneity ofthe self-employed sector increased during themarket reform in a pattern that foretold whathappened in the rest of Latin America during the1990s (Klein and Tokman 2000 Portes andHoffman 2003) and what may be current devel-opments in the industrialized world (Arum andMuumlller 2004 Noorderhaven et al 2003) Thesecharacteristics of the Chilean class structureare considered in the comparative analysis ofChilean mobility

DDAATTAA AANNDD AANNAALLYYTTIICCAALL AAPPPPRROOAACCHH

This study uses the 2001 Chilean MobilitySurvey (CMS) The CMS is a nationally repre-sentative multistage stratified sample of maleheads of household ages 24 to 69 The samplingstrategy includes the following stages First 87primary sampling units (PSUs) (counties) areselected Then blocks within the PSUs areselected and finally households within blocksare chosen Counties are stratified according tosize (fewer than 20000 20000 to 100000100000 to 200000 and more than 200000inhabitants) and by geographic zone (NorthCenter South) All PSUs in the large size stra-tum are included in the sample to increase effi-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443311

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ciency The fieldwork conducted between Apriland June 2001 consists of face-to-face inter-views in the respondentrsquos household carried outby trained personnel The survey excludesnonndashhead-of-household males which represent17 percent of the male population of the rele-vant age (MIDEPLAN 2001) Among thoseexcluded 865 percent are the sons of the headof the household 124 percent are another rel-ative of the head and 11 percent are other non-related males Their occupational distributionwith control used for age is almost identical tothat of the heads of household The small pro-portion represented by this group and their sim-ilar occupational distribution suggest that theirinclusion would not significantly alter the find-ings presented in this article Excluding thehouseholds not eligible for the survey theresponse rate is 63 percent Although nonre-sponse rates usually are not reported in Chileansurveys exchange with Chilean experts indi-cates that the nonresponse rate is about 20 to 25percent for face-to-face household surveys Thehigher nonresponse rate of the CMS is likelyattributable to the difficulty contacting maleheads of household4 The total sample size is3544 I exclude individuals outside the agerange of 25 to 64 years which is conventional-ly used in comparative mobility research andcases with unusable data After this exclusionthe usable sample size is 3002

CCLLAASSSS SSCCHHEEMMAA

This study uses a class perspective to describethe Chilean social structure It utilizes the seven-category version of the classification devel-oped by the CASMIN project Thisclassification is widely used in internationalcomparative research and describes the basic

stratification of advanced industrial societiesbased on ldquoemployment relationshipsrdquo (Eriksonand Goldthorpe 1992a35ndash47 Goldthorpe andHeath 1992) The seven classes included in theschema are I+II (service class) III (routine non-manual workers) IVab (self-employed work-ers) V+VI (manual supervisors and skilledmanual workers) VIIa (unskilled manual work-ers) IVc (farmers) and VIIb (farm workers)

Use of the 7-class categorization instead ofan alternative detailed 12-class version pro-duced by the CASMIN project (Erikson andGoldthorpe 1992b) induces some loss of infor-mation about the originndashdestination associa-tion A global test of all five aggregationsobtained from the ratio of L2 tests for the inde-pendence model in the collapsed and full tablesindicates that the seven-class schema masks 29percent of the association shown by the full setof classes ([L2 full table ndash L2 collapsed table]L2 full table = [108644 ndash 77191]108644 =29) This proportion is high compared withthat of the CASMIN countries and is surpassedonly by Sweden with 31 percent (Hout andHauser 1992 Table 2) A collapsed sevenfoldclass schema was however preferred because ofthe moderate sample size of the Chilean surveyand because it grants international comparabil-ity of the findings5

Operationalization of this class classifica-tion is based on detailed information concern-ing job title (recoded into the standard ISCO-88classification [ILO 1990]) industry occupa-tional status and supervisory status of workersBecause the CMS is specifically designed toassess class mobility it includes all the infor-mation necessary to produce the CASMIN classschema The CASMIN project has not produceda standard algorithm to generate the class clas-sification but other researchers have producedsuch an algorithm which I use in this analysis(Ganzeboom and Treiman 2003) Table 1 pres-ents the basic intergenerational mobility tableto be used in this analysis including the countsand marginal percentage distribution for originand destination classes

443322mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

4 The nonresponse rate can yield bias if those whowere unreachable or refused to participate differ fromthose included in the sample To estimate the mag-nitude of this bias I compared the distribution of keyvariables with the CASEN 2000 survey CASEN isa large survey (n = 252595) conducted by the Chileangovernment and has a refusal rate lower than 10percent The comparison (available upon request)suggests that the CMS slightly underrepresents agri-cultural workers and the upper class Overall how-ever there is no indication of major nonresponsebias

5 Analyses were replicated using the detailed 12-class table and no difference in substantive find-ings was found

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

MMEETTHHOODDSS

I fit two alternative models to the Chilean mobil-ity table The first is the ldquocore model of social flu-idityrdquo which claims to capture basic internationalsimilarity in mobility Although the core modelhas become a benchmark for mobility researchand currently is the only model that provides acomparative framework for mobility analysisanalysts have drawn attention to a number of itslimitations The model has been criticized forinadequately representing hierarchical mobilityeffects for being tailored to a few industrializedcountries for the ad hoc nature of some effects itincludes (specifically the affinity effects) andfor lacking a criterion to determine how muchdivergence is necessary for it to be rejected(Ganzeboom Luijkx and Treiman 1989 Houtand Hauser 1992 Sorensen 1992 Yamaguchi1987) I therefore also test an alternative ldquohybridmodelrdquo that combines the rowndashcolumn (II) asso-ciation model (Goodman 1979) with parametersaccounting for class immobility The internationaland temporal comparison of Chilean mobilitypatterns uses the uniform-difference (UNIDIFF)model also known as the multiplicative layereffect model (Erikson and Goldthorpe 1992aXie 1992) Model comparison and selection isbased on the Bayesian Information Criterion(BIC) statistic (Raftery 1995)

AANNAALLYYSSIISS OOFF CCHHIILLEEAANN MMOOBBIILLIITTYY IINNCCOOMMPPAARRAATTIIVVEE PPEERRSSPPEECCTTIIVVEE

TTHHEE CCOORREE MMOODDEELL OOFF SSOOCCIIAALL FFLLUUIIDDIITTYY

I start by fitting the core model to the Chileanmobility table The core model represents the

basic international commonality in social flu-idity using a topological formulation In contrastto standard topological models constructed froma single allocation of cells (Hauser 1978) thecore model uses eight matrices each of whichis designed to capture a particular effect thatenhances or reduces mobility between specificclasses (Erikson and Goldthorpe 1987a 1992achapter 4) In this way the model permits thesources of commonality across nations to beidentified clearly and likewise the departuresfrom it The effects are of four types hierarchyinheritance sector and affinity6

HIERARCHY EFFECTS Hierarchy effects reflectthe impact of status distances between classeson fluidity between them To estimate theseeffects the mobility table is divided into threestrata reflecting differences in resources andrewards across classes The strata are the fol-lowing upper stratum (class I+II) middle stra-tum (classes III IVab IVc and V-VI) and lowerstratum (classes VIIa and VIIb) Because thereare three strata two hierarchy effects (HI1 andHI2) can be identified each representing thecrossing of one additional hierarchical barrierThe association between each pair of classes isexpressed as an inverse function of the numberof hierarchical strata crossed

INHERITANCE EFFECTS Inheritance effects aredesigned to capture the propensity for class

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443333

2322-ASR 703 filename70304-torche

Table 1 Seven-Class Intergenerational Mobility Table for Chilean Men

Class of Destination

I-II III IVab IVc V-VI VIIa VIIb N

I-II Service Class 229 49 81 2 37 30 2 430 14III Routine Nonmanual 61 15 34 0 24 27 1 162 5IVab Self-employed 123 37 173 28 102 90 20 573 19IVc Farmers 42 16 68 54 58 54 24 316 11V-VI Skilled Manual 91 33 116 13 155 135 36 579 19VIIa Semiskilled and Unskilled Manual 49 43 115 13 111 123 44 498 17VIIb Farm Workers 28 16 78 17 89 101 115 444 15N 623 209 665 127 576 560 242 3002 21 7 22 4 19 19 8

Note Cell entries represent counts

6 Design matrices in Erikson and Goldthorpe(1992a124ndash129)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

immobility They include three matrices IN1identifies immobility across all classes alike IN2identifies the higher immobility of the serviceclass (I+II) the self-employed group (IVab) andfarmers (IVc) and IN3 accounts for the highestimmobility of farmers (IVc)

SECTOR EFFECTS Sector effects are intended toidentify the difficulty of moving between theagricultural and nonagricultural sectors of theeconomy (SE1)

AFFINITY EFFECTS Affinity effects are intend-ed to capture specific discontinuities (negativeaffinities) or linkages (positive affinities) betweenclasses which either reinforce or offset the over-all effects of hierarchy and sector There are twoaffinity effects The first AF1 identifies the dif-ficulty of moving between the service class (I+II)and the class of farm workers (VIIb) which addsto their hierarchical distance In contrast AF2identifies instances in which mobility is morefrequent than accounted for by hierarchy and sec-tor effects and includes affinities within the non-manual sector (I+II and III) and within the manualsector (V+VI and VIIa) a symmetrical affinitybased on capital possession (IVab and IVc and Iand IVab) and an asymmetrical link between agri-cultural classes of origin (IVc and VIIb) and theunskilled manual class of destination (VIIa)7

TTHHEE MMOOBBIILLIITTYY PPAATTTTEERRNN IINN CCHHIILLEECCOONNVVEERRGGEENNCCEE OORR DDIIVVEERRGGEENNCCEE

Table 2 presents the fit of several models appliedto the Chilean data and Table 3 displays theparameter estimates for each of these modelsModel 1 in Table 2 indicates that the fit of theoriginal core model to the Chilean table is unsat-isfactory under standard statistical criteria butit explains a significant 775 percent of associ-ation under independence To assess the strengthof the different factors driving mobility oppor-tunities I compare the Chilean coefficients withthose obtained for the CASMIN countriesCoefficients from the original core model fit-ted to the Chilean table are reported in row 1 ofTable 3 CASMIN coefficients are reported inrow 2 and differences in coefficientsrsquo magni-tude and the statistical significance of the dif-ferences are reported in row 38

All Chilean coefficients are significant andhave the same sign as for the CASMIN coun-tries Comparison of the magnitude of the coef-ficients however shows significant departuresfrom the core for five effects which revealinteresting differences between Chile and theCASMIN countries The hierarchy coefficientsindicate that whereas the short-range hierar-chical barriers are somewhat weaker in Chile

443344mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

8 CASMIN coefficients were obtained from fittingthe core model to the combined table of France andEngland standardizing both tables to have 10000cases as recommended by Erikson and Goldthorpe(1992a121) The coefficients obtained are virtuallyidentical to those presented by Erikson andGoldthorpe (1992a Table 44)

Table 2 Fit statistics selected models to the Chilean Mobility Table

AssociationL2 BIC df p Explained

Core Modelmdash1 Core Model Original 1375 285 28 000 775mdash2 Core Model EHE 1081 ndash08 28 000 860mdash3 Chilean version of Core Model ([Model 2] + Chilean Parameters) 724 ndash326 27 000 906Association Models 000mdash4 Heterogeneous Quasi-RC(II) Model 734 ndash161 23 000 905mdash5 Homogeneous Quasi-RC(II) Model 803 ndash287 28 000 896mdash6 Quasi-Linear by Linear Association (SES row and column ranking)a 1563 279 33 000 798

Note EHE = empirical hierarchical effects SES = eocioeconomic statusa Linear-by-linear association model using socioeconomic status scores as presented in Appendix Figure A1 torank origin and destination classes

7 For a detailed account of the empirical derivationof the model see Erikson and Goldthorpe (1987a1987b and 1992a chapter 4)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443355

2322-ASR 703 filename70304-torche

Tab

le 3

M

odel

Par

amet

er E

stim

ates

for

the

Chi

lean

Mob

ilit

y Ta

ble

Eff

ect

1 C

ore

mod

el o

rigi

nala

mdash 2 C

ore

mod

el C

AS

MIN

b

mdash 3 D

iffe

renc

e C

ASM

INmdash

Chi

lec

mdash 4 C

ore

mod

el E

HE

d

mdash 5 C

hile

an v

ersi

on o

f co

ree

mdash Not

eD

ata

show

n ar

e m

odel

par

amet

er e

stim

ates

wit

h st

anda

rd e

rror

s in

par

enth

eses

EH

E =

em

piri

cal h

iera

rchi

cal e

ffec

ts

aM

odel

1 f

rom

Tab

le 2

b

Ori

gina

l Cor

e M

odel

fit

ted

to th

e E

ngli

sh a

nd F

renc

h m

obil

ity

tabl

es S

ee te

xt f

or d

etai

ls

cD

iffe

renc

e in

par

amet

er m

agni

tude

bet

wee

n or

igin

al c

ore

mod

el f

itte

d to

the

Chi

lean

tabl

e (r

ow 1

) an

d to

the

CA

SM

IN ta

ble

(row

2)

dM

odel

2 f

rom

Tab

le 2

e

Mod

el 3

fro

m T

able

2

dagger p

lt 1

p

lt 0

5

p

lt 0

1

p

lt 0

01 (

two-

tail

ed)

HI1

ndash12

(

05)

ndash22

(02

)ndash

10 dagger

(06

)ndash

03(

07)

ndash09

dagger(

06)

HI2

ndash65

(11

)ndash

45

(

04)

20

dagger(

12)

ndash74

(09

)ndash

47

(

10)

IN1

41

(

08)

43

(

03)

Insi

g

37

(

07)

34

(

07)

IN2

39

(

10)

84

(

04)

45

(

11)

31

(13

)mdash

IN3

68

(

24)

101

(11

)In

sig

61

(24

)4

8(

23)

SE

1

ndash38

(07

)ndash1

06

(

05)

ndash68

(08

)ndash

42

(

07)

ndash20

(

08)

AF

1

ndash76

(21

)ndash

83

(

13)

Insi

g

ndash84

(20

)ndash

63

(21

)

AF

2

33

(

05)

45

(

02)

ndash12

(

06)

20

(

06)

26

(

06)

AF

1ndashC

mdash mdash mdash mdash

ndash14

9

(46

)

IN2ndash

C

mdash mdash mdash mdash 71

(

14)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(HI1) long-range hierarchical mobility is moredifficult in this country (HI2) This suggestssubstantial inequality between the extremes ofthe class hierarchy combined with smaller dif-ferentiation in the middle of the distribution Asto the inheritance effects the service class theself-employed group and the farmers (classesI+II IVab and IVc respectively) have a sig-nificantly lower propensity to immobility inChile than in the CASMIN countries suggest-ing that class positions usually associated withindependent work are less able to reproducetheir class status intergenerationally than depict-ed by the core Most impressive the barrierbetween the agricultural and nonagriculturalsectors of the economy (SE1) is much weakerin Chile as is the affinity effect linking non-manual classes manual classes and the class-es that own capital among themselves (AF2)

Interestingly all the significant differences incoefficients with the single exception of the bar-rier to long-range hierarchical mobility sug-gest that Chile is more fluid than the core modeldepicts Higher fluidity seems to be driven bythe weakness of horizontal barriers separatingthe agricultural manual and self-employed sec-tors of the economy This is consistent withfindings of weaker sector effects in countriesthat have experienced rapid industrializationsuch as Korea (Park 2004) Brazil (Costa-Ribeiro 2003) Israel (Yaish 2004 Goldthorpeet al 1997) and Japan (Ishida et al 1991)

Additionally in the Chilean case the weak-ness of sector barriers may have been intensi-fied by the deep market transformation of the1970s and 1980s On the one hand the ldquochangeof handsrdquo in the agricultural sector induced bythe agrarian reform and counterreform likelyincreased fluidity between the agricultural andurban classes On the other hand the role of self-employment as an ephemeral refuge against(industrial and public sector) unemploymentmay have reduced barriers between the self-employed group and the rest of the social struc-ture Added to the larger difficulty of long-rangehierarchical mobility these features suggestweak sector cleavages in the Chilean fluiditypattern and high salience of the hierarchicaldimension of mobility

For the countries in which the fit of the coremodel is not adequate Erikson and Goldthorpe(1992a145) suggest adding parameters thataccount for country-specific deviations from

the model I therefore introduce adjustments toreflect salient characteristics of the Chilean flu-idity pattern producing a ldquoChilean versionrdquo ofthe core model

First I modify the hierarchical effects Fittingof the core model suggests that hierarchicalbarriers are a crucial determinant of mobility inChile This conclusion is however obscuredby the fact that the hierarchical strata distin-guished by the core model are not empiricallyobtained and may not accurately represent theChilean ranking of classes in terms of socioe-conomic status (SES) To model hierarchicaleffects in Chile adequately I rank classes usingthe unweighted average of their schooling andearning levels as a proxy for SES I then collapsethe seven classes into three hierarchical stratausing cluster analysis These three empiricallyobtained strata are the following Upper stratum(service class I+II) middle stratum (routinenonmanual and self-employed group III andIVab) and lower stratum (manual and agricul-tural classes V-VI VIIa IVc and VIIb) Detailson the ranking of classes cluster procedures andthe comparison of the CASMIN and empiricalChilean class ranking can be found in theAppendix Comparison of the CASMIN and theempirical class ranking suggests higher differ-entiation at the top than at the bottom of thesocial structure in the Chilean case

Model 2 in Table 2 uses the empiricallyobtained hierarchical strata (EHE) Theimprovement in fit is massive (the difference inBIC is 293 without using any degrees of free-dom) The model now accounts for 86 percentof the association under independence

The large improvement in fit when empiricalhierarchical strata are used challenges theassumption of an internationally homogeneoushierarchical class ranking At the empiricallevel there is no proof that the hierarchicalranking of classes assumed by the core modelfits all CASMIN countries9 Moreover someevidence points to significant cross-country dif-

443366mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

9 Erikson and Goldthorpe(1992a Table 22) pres-ent the average ranking of classes based on standardstatus measures but they do not show the empiricalranking of classes for each country If the occupa-tional composition of classes varies across countriesthere may be significant international differences inthe hierarchical ranking of classes

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ferences in the ranking of classes (Hout andHauser 1992 Sorensen 1992) At the theoreti-cal level the creators of the core model high-light the relevance of highly specific historicallyformed national attributes to explain interna-tional variation in mobility These national par-ticularities may very well result in differenthierarchical positions of classes across coun-tries10 Furthermore even if the homogeneousranking of classes imposed by the CASMINresearchers applied to most CASMIN coun-tries significant departures may be found inthe late industrialized countries currently beingadded to the comparative mobility project TheChilean findings suggest that instead of impos-ing international homogeneity to the hierarchi-cal dimension of mobility the ranking of classesshould be treated as nationally specific param-eters to be estimated as mobility researchersroutinely treat row and column marginals ofthe mobility table The issue is importantbecause the hierarchical dimension is the mostrelevant dimension of mobility (Hout andHauser 1992 Wong 1992) and inadequate mod-eling can bias assessment of not only hierar-chical effects of mobility but also other effects

Two other changes were introduced toaccount for Chilean-specific deviations fromthe core model First the IN2 effect wasrearranged to exclude the self-employed class(IVab) and to include the class of farm workers(VIIb) This rearrangement accounts for thefact that the Chilean self-employed classappeared less likely than depicted by the coremodel to reproduce its class position acrossgenerations and the inheritance level of Chileanfarm workers is higher than in the CASMINcountries These changes are expressed in theChilean-specific inheritance parameter IN2-CSecond an affinity parameter (AF1-C) wasadded to capture the difficulty of long-rangedownward elite mobility This negative asym-metrical affinity links flows from the service

class to the agricultural classes and from theroutine nonmanual class to the class of farmworkers Note that the AF1-C adds to the hier-archical distance between the two extremes ofthe class structure and to the negative affinitycaptured by AF1 to depict the extreme difficultyof long-range downward mobility from theChilean elite The fit of this lsquoChilean versionrsquoof the core model is reported in Table 2 model3 The improvement in BIC from model 2 is sub-stantial Furthermore even if the model fallsshort in terms of standard statistical criteria itaccounts for a large 91 of the associationunder independence thereby supporting theclaim that it adequately represents the Chileanmobility pattern The parameter estimates for theChilean version of the core are reported in row5 Table 3

Because of the core modelrsquos undesirable prop-erties I also use an alternative approach toexamine Chilean mobility patterns Thisapproach uses the now standard associationrowndashcolumn (II) (RC[II]) model (Goodman1979 see also Hout 1983 and Wong 1992) Inthis model both origin and destination classesare scaled so that the association can beexpressed as a linear-by-linear interaction by asingle parameter The multiplicative form ofthe model can be expressed as follows

Fij = iO j

D exp( microi νj)

where i indexes class of origin j indexes classof destination Fij is the expected frequency inthe (ij) cell is the grand mean i

O pertainsto the class of origin marginal effect j

D pertainsto the class of destination marginal effect isa global association parameter microi is the scalevalue for the ith class of origin and νj is the scalescore for the jth class of destination subject tothe following normalization constraints Σmicroi =Σνj = 0 (normalization of the location) and Σmicroi2

= Σνj2 = 1 (normalization of the scale)The class scale scores reflect a latent con-

tinuous variable made manifest by the class cat-egories Empirically obtained from the datathese ldquodistancerdquo scores produce an optimalranking of classes for the purpose of mobilityanalysis The RC(II) model is appropriate formodeling the association of any two ordinalvariables but mobility tables are distinctivebecause of the correspondence between class oforigin and class of destination (Gerber and Hout2004691)

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443377

2322-ASR 703 filename70304-torche

10 For instance during the socialist period inHungary the status of the proletariat may have beenindistinguishable from the status of the routine non-manual class (Szelenyi 199862) In Scotland givento early industrialization the hierarchical barrierbetween the skilled and unskilled working classes thatcharacterizes other CASMIN countries may not exist(Erikson and Goldthorpe 1992a163)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I introduce two adjustments to account forthis correspondence transforming the formu-lation into a ldquohybrid modelrdquo First topologicalparameters will account for the large counts insome diagonal cells reflecting immobility(inheritance) effects These parameters distin-guish two inheritance levels ldquohigh inheritancerdquofor the service class (I+II) farmers (IVc) andfarm workers (VIIb) and ldquolow inheritancerdquo forall other classes with the exception of the rou-tine nonmanual workers

Model 4 in Table 2 reports the fit statistic forthis quasi-RC(II) model The BIC statisticrejects it and prefers the more parsimoniousChilean version of the core model ThereforeI introduce a second adjustment to account forthe correspondence between rows and columnsin mobility tables Origin and destination scoresare constrained to be the same assuming thatthe scaling of classes has not significantlychanged over time Model 5 in Table 2 reportsfit statistics for this homogeneous quasi-RC(II)model On the basis of the BIC statistic the fitof the model is as good as that of the Chileanversion of the core model (Wong [1994] showsthat BIC differences of less than 5 points shouldbe considered indeterminate) Thus under stan-dard statistical criteria the Chilean core modeland the homogeneous quasi-RC(II) model areindistinguishable Given that the Chilean versionof the core model was tailored to fit the Chileantable the fit of model 5 is impressive and it indi-cates that a unidimensional scale can adequatelycapture mobility distances between classes

The logical next question is what this scalerepresents and specifically whether it repre-sents the hierarchical ranking of classes in termsof SES If a close correspondence is foundbetween the mobility distances among classesand their distances in terms of SES this find-ing would suggest that the intragenerationaldistribution of resources and rewards expressedin the SES ranking of classes drives the processof intergenerational mobility In other words thisfinding would support the resource approachrsquosclaim that differences in resources associatedwith diverse class positions determine mobili-ty opportunities across generations

To examine this possibility I compare theclass scores obtained from the homogeneousquasi-RC(II) model with the empiricallyobtained SES ranking of classes based on edu-cation and earnings (Appendix) Figure 4 pres-

ents the comparison between the standardizedvalues of both sets of scores The correspon-dence is striking The mobility distancesbetween classes closely reproduce their dis-tances in terms of SES This close isomorphismsuggests that at least in Chile the intragener-ational inequality in the distribution of socialresources and rewards across classes largelydrives the intergenerational mobility process Totest the isomorphism further I use the SESclass scores as row and column rankings to esti-mate a quasindashlinear-by-linear association model(model 6 Table 2) The fit of model 6 is sig-nificantly worse than that of model 5 almost aspoor as that of the original core model This indi-cates that although the inter- and intragenera-tional dimensions of inequality are very similarin Chile they have a somewhat different struc-ture

In summary the central characteristic of theChilean mobility regime is the predominance oflong-range hierarchical barriers and the weak-ness of sector cleavages separating classes in themiddle and at the lower end of the hierarchicalstructure Findings from the ldquoChilean versionrdquoof the core model and the quasi-RC(II) hybridmodels are consistent depicting a social struc-ture characterized by signif icant barriersbetween the top echelon and the rest of the classstructure The good fit of the homogeneousquasi-RC(II) model provides preliminary sup-port for the resource approach linking inequal-ity and mobility Mobility dynamics seem to bedriven largely by the hierarchical distancesbetween classes and overall Chile seems not tobe less fluid than the CASMIN countries

CCOOMMPPAARRIISSOONN OOFF TTHHEE OORRIIGGIINNAALL AANNDD CCHHIILLEEAANN

VVEERRSSIIOONNSS OOFF TTHHEE CCOORREE MMOODDEELL

Although Chilean mobility dynamics seem to bedriven by inequality across classes in a veryunequal society the fitting of the core modelprovides no indication that Chile is significantlyless fluid than depicted by the core model Asa preliminary analysis of the Chilean level of flu-idity I compare the predicted propensities formobility and immobility between specific pairsof classes in Chile with those in the CASMINcountries Propensities are obtained from theoriginal and Chilean versions of the core modelrespectively (models 2 and 5 of Table 3)

443388mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I begin with the immobility of those at the topof the class structure namely the service class(I+II) The service classrsquo propensity for immo-bility in the original core model is captured bytwo parameters IN1 and IN2 This propensityis more than three times what it would be in theabsence of these effects (e43+84 = 356) InChile the propensity for immobility of class I+IIis captured by the same two parameters and itis lower than what the core model expresses286 times larger than it would be in the absenceof these effects (e34+71) What about the immo-bility of the two agricultural classes In theCASMIN countries immobility of the IVc andVIIb classes is respectively 785 and 154 timeswhat it would be in the absence of these param-eters (e43+84+101ndash22 and e43 respectively) InChile the comparable values are 422 and 286(e34+71+48ndash09 and e43+71 respectively) indi-cating the relatively low inheritance associatedwith land ownership in the Latin Americannation Thus if attention is focused on theimmobility in the upper and lower ends of thesocial hierarchy the comparison does not indi-cate less fluidity in the highly unequal Chileansociety

What about the mobility rates between thetwo most distant classes in the hierarchical rank-ing namely the service class (I+II) and farmworkers (VIIb) In the case of the CASMINcountries the mobility chances between theseclasses are captured by four parameters (HI1HI2 SE1 and AF1) and is 077 times smallerthan it would be in the absence of these effects(endash22 ndash45ndash106ndash83) In Chile the distance fromclass I to class VIIb is expressed by these samefour parameters in addition to the asymmetri-cal disaffinity AF1-C This yields mobilitychances 056 times what would be obtained inthe absence of these effects (endash09 ndash47ndash20ndash63ndash149)Thus the barrier to long-range downwardmobility from the elite is more significant inChile than in the CASMIN countries The sameoccurs with long-range downward mobility fromclass III

If we measure the reciprocal flow (ieupward mobility from class VIIb to class I+II)we obtain the same value for the CASMINcountries because all parameters are symmet-rical in the original core model For Chile how-ever this calculation removes the asymmetricalbarrier to elite downward mobility which rais-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443399

2322-ASR 703 filename70304-torche

Figure 4 Class Scores Obtained from the Homogeneous Quasi-RC(II) Model and from Socioeconomic Status(SES) Ranking

Note (1) = Class standardized socioeconomic status scores based on schooling and earnings (2) = Class stan-dardized scores based on the Homogeneous Quasi-RC(II) model (see text for details)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 2: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

some analysts have argued that growing inequal-ity can be offset by a rise in mobility tacitlyassuming that these two distributional phe-nomena can move in opposite directions(Friedman 1962)

At the theoretical level approaches linkingmobility and inequality tend to focus on micro-level mechanismsmdashthe motivations andresources affecting individual decision mak-ingmdashconnecting these macrostructural phe-nomena Two perspectives can be distinguishedA ldquoresource perspectiverdquo indicates that highinequality will result in decreased mobilitybecause the uneven distribution of resourceswill benefit those most advantaged in the com-petition for success In contrast an ldquoincentiveperspectiverdquo argues that inequality will raisethe stakes in the competition thereby inducinghigher mobility

In the end the question about the link betweenintergenerational mobility and socioeconomicinequality is empirical and currently the empir-ical evidence is thin and inconclusive Part ofthe problem is that mobility has been studied ina small pool of mostly industrialized countriesall of which share relatively similar levels ofinequality To explore the link between mobil-ity and inequality at the empirical and theoret-ical levels this article introduces Chile to thecomparative mobility project

Chile is a middle-income country that hasundergone significant political and economicchange in the past few decades In the midndash20thcentury the Chilean development strategy wasdefined by import-substitutive industrializa-tion The economy of Chile was closed to inter-national markets and the state had a pivotaleconomic and productive role Mounting socialproblems associated with unequal developmentand urban migration led to two progressiveadministrations in the 1960s and early 1970sThese administrations conducted major redis-tributive reforms including the nationalizationof enterprises an educational reform and anagrarian reform This progressive path was vio-lently halted by a military coup in 1973 Led byGeneral Pinochet the military took power andretained it until 1990

Once in power the authoritarian regime con-ducted a deep and fast market-oriented trans-formation consisting of the now standardpackage of macroeconomic stabilization pri-vatization of enterprises and social services

and liberalization of prices and markets As aresult Chile transformed from a closed econo-my with heavy state intervention into one of themost if not the most open and market-basedeconomy of the world

After a deep recession in the early 1980sChile has experienced substantial and sustainedeconomic growth since the late 1980s whichconcurs with the reestablishment of democrat-ic rule The period of redemocratization andgrowth during the 1990s has led to significant-ly improved living standards for the Chileanpopulation The dark side of this success storyis the persistent economic inequality in thecountry Inequality historically very high grewduring the military government and Chile cur-rently ranks as the 10th most unequal countryin the world The Chilean pattern of inequalitycan be characterized by ldquoconcentration at thetoprdquo Chile is highly unequal because thewealthiest segment of the society receives avery large portion of the national incomewhereas the differences between the poor andmiddle-income sectors are much less pro-nounced lower even than in some industrializednations Although inequality is by definitionassociated with concentration the Chilean caseis extreme as compared with the industrializedworld and even with other Latin Americannations

These features render Chile an adjudicativecase If mobility and inequality are relatedChile should show mobility rates significantlydifferent from those of its industrialized peersTo explore these issues my analysis includesthree components

First I analyze the Chilean mobility regimein a comparative perspective Mobility is here-in defined as the association between class oforigin (fatherrsquos class position) and class of des-tination (current class position) net of thechanges in the class structure over time (struc-tural mobility) Also known as ldquorelative mobil-ityrdquo or ldquosocial fluidityrdquo mobility represents thelevel of openness or degree of equality of oppor-tunity in a society I place Chile in the compar-ative context by fitting the ldquocore model of socialfluidityrdquo to the Chilean intergenerational mobil-ity table This model claims to represent thebasic similarity in mobility across countriesbut it has been tested in only a few mostlyindustrialized countries (see for exampleErikson and Goldthorpe 1992a 1992b) By fit-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442233

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ting the core model to the Chilean table I deter-mine to what extent Chile departs if it doesfrom the assumed international homogeneityin mobility regimes and what the sources of itspossible departure are Even if the core modelhas become a milestone in comparative analy-sis it has a number of undesirable propertiesdiscussed later in the Methods section Thus asa modeling alternative I use a hybrid model thatcombines association parameters assuming ahierarchical conception of the mobility structurewith patterns accounting for class immobilityInterpretation of these two models provides adetailed description of the Chilean mobilityregime and an evaluation of the main factorsmdashhierarchical differences across classes classinheritance barriers across sectors of the econ-omymdashdriving Chilean mobility dynamics

Second I compare the level of mobility inChile with that of other countries using pri-mary data from France England ScotlandIreland Sweden the United States and IsraelAll these countries have a much more egalitar-ian income distribution than Chile Thus ifmobility is associated with inequality the inter-national comparison should show that mobili-ty opportunities in Chile are signif icantlydifferent from the opportunities in these indus-trialized nations

Third my analysis moves from an interna-tional to a temporal comparison To explore theassociation between mobility and inequalityfurther I study Chilean mobility rates over timeUsing a cohort analysis I examine whether thesignificant increase in inequality associatedwith the market transformation of the 1970sand 1980s had any noticeable effect on Chileanmobility rates

Advancing some of the major results thefindings present an interesting paradox On theone hand the Chilean mobility regime is foundto be driven almost exclusively by the hierar-chical distance between classes which is deter-mined in turn by the level of inequality in thecountry This finding is consistent with the neg-ative relationship between mobility and inequal-ity posed by the ldquoresource perspectiverdquo On theother hand the international comparison indi-cates that Chile is impressively fluid despite itshigh economic inequality and the temporalanalysis shows no change in Chilean mobilityover time despite the growing inequality duringthe military regime Therefore the internation-

al and temporal comparisons seem to contradictthe negative relationship between mobility andinequality and to suggest that high inequalitymay in fact induce mobility

Solving this apparent paradox I arguerequires switching the analytical focus from thelevel to the pattern of both mobility and inequal-ity When the pattern of these two distributivephenomena is considered the contradiction dis-appears and the mechanisms driving the mobil-ityndashinequality association can be examined

The argument is organized as follows Section2 presents the empirical and theoretical evi-dence concerning the association of mobilitywith inequality Section 3 describes the Chileancontext It discusses the most salient character-istics of the Chilean socioeconomic structureand presents a brief historical description ofthe major changes in the national political econ-omy over the past few decades Section 4 intro-duces the data and analytical approach Section5 presents the comparative analysis of mobili-ty in Chile including the description of theChilean mobility regime the international com-parison of the fluidity level in Chile and theanalysis of change over time Section 6 presentsa summary of the findings the conclusionsand the implications of the Chilean case for thecomparative study of inequality

AASSSSOOCCIIAATTIIOONN BBEETTWWEEEENN IINNEEQQUUAALLIITTYYAANNDD MMOOBBIILLIITTYY EEMMPPIIRRIICCAALL AANNDDTTHHEEOORREETTIICCAALL EEVVIIDDEENNCCEE

Although mobility and inequality ldquogo togetherintuitivelyrdquo (Hout 2004969) they are differentdimensions of the social distribution of advan-tage The former refers to the cross-sectionaldimension which determines the individualposition in a social hierarchy The latter refersto its intergenerational dimension specificallyto differential access to these positions as deter-mined by the position of origin (Marshall andSwift 1999243 Marshall et al 199713)

At the conceptual level the differencesbetween these two distributional phenomenaare clear Inequality describes the distributionof resources at any particular point in timeMobility describes inequality of opportunitythe chances that someone with a particular socialorigin will attain a more rather than a less advan-taged destination regardless of the socioeco-nomic distance between these destinations This

442244mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

is why ldquoindividuals are anonymous for inequal-ity but not for mobilityrdquo (Behrman 199972) Infact mobility analysis assumes path dependencein the reproduction of the social structure acrossgenerations and studies the extent and mor-phology of such intergenerational dependence

The conceptual difference may have impor-tant practical implications As argued byFriedman (1962) and echoed more recently bysome researchers and policymakers a givenextent of income inequality under conditions ofgreat mobility and change may be less a causefor concern than the same degree of inequalityin a rigid system wherein the position of ldquopar-ticular families in the income distribution doesnot vary widely over timerdquo (p 171)

The key question is whether these two dis-tributional phenomena are independent Twotheoretical approaches address this questionBoth focus on the micro-level mechanisms thatpurportedly link these macrophenomena theindividual decision-making processes Theldquoincentive approachrdquo claims that the motiva-tion to pursue mobility is proportional to theamount of cross-sectional inequality If inequal-ity approaches zero so does the payoff formobility Conversely great inequality increas-es both the inducement to pursue mobility forthose who are initially disadvantaged and theincentive to resist mobility for those who are ini-tially advantaged In other words inequalityraises the stakes of mobility (Hout 2004970 seealso Tahlin 2004)

The ldquoresource approachrdquo contends thatmobility depends critically on resources ratherthan on incentives The higher the inequality thegreater the distance in terms of human finan-cial cultural and social resources across dif-ferent social origins In a highly unequal societythe stakes of mobility will be high across theboard but the crucial resources controlled byindividuals will be so unevenly distributed thatcompetition certainly will benefit those moreadvantaged (see Goldthorpe 2000254 Stephens197954 and Tawney 1965 for a seminal elab-oration of this argument) The outcome of highcross-sectional inequality will be the rigid repro-duction of the class structure over time ielow social fluidity

Because resources and incentives work indifferent directions the effect of inequality onmobility can be thought of as an additive cal-culation If the impact of resources outweighs

that of incentives we should expect a negativeassociation between inequality and mobilityConversely the association should be positiveif incentives are more relevant than resources inthe competition for advantaged positions Inwhat follows I argue that consideration ofresources and incentives affecting mobility at themicro-level should be preceded by a macro-level understanding of both mobility andinequality This understanding should considerthe pattern of these two distributional phenom-ena not only their overall level and shouldfocus on the location and depth of the majorcleavages in the social structure

TTHHEE EEMMPPIIRRIICCAALL EEVVIIDDEENNCCEE

Empirical evidence concerning themobilityndashinequality association is scarceAlthough high-quality data on inequality existfor a large number of countries (eg Deiningerand Squire 1996) comparable data on mobili-ty are available only for a few mostly industri-alized nations The most important source ofcomparable mobility data is the ComparativeAnalysis of Social Mobility in IndustrializedCountries (CASMIN) project which includes 15industrialized nations1 The findings of theCASMIN project are highly relevant for theassociation between mobility and inequalityThe CASMIN project concluded that whereasabsolute mobility varies across countries andover time because of differences in nationaloccupational structures relative mobilitymdashtheassociation between class of origin and desti-nation net of differences in marginal occupa-tional distributionsmdashis extremely homogeneousacross countries and over time These findingswere expressed in the ldquocommonrdquo and ldquocon-stantrdquo social fluidity hypotheses respectivelyThe basic temporal and international similari-ty was systematized into a ldquocore modelrdquo ofsocial fluidity (Erikson and Goldthorpe 1987a1987b 1992a see Featherman Jones andHauser 1975 and Hauser et al 1975a 1975b for

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442255

2322-ASR 703 filename70304-torche

1 These include 12 European countries (EnglandFrance the former Federal Republic of GermanyHungary Ireland Northern Ireland Poland ScotlandSweden the former Czechoslovakia Italy and theNetherlands) and three industrial non-European coun-tries (the United States Australia and Japan)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

the original hypotheses) Common social flu-idity is claimed to hold (at least) in industrialsocieties with a ldquomarket economy and nuclearfamilyrdquo (Featherman et al 1975340) and itwould be explained by ldquothe substantial uni-formity in the economic resources and desir-ability of occupationsrdquo (Grusky and Hauser1984)

The significant international similarity inmobility patterns has been confirmed largelyin pairwise or three-country comparisons in afew industrialized countries EriksonGoldthorpe and Portocarero (1979) Eriksonet al (1982) and Hauser (1984a 1984b) forEngland France and Sweden McRoberts andSelbec (1981) for the United States andCanada and Kerckhoff et al (1985) forEngland and the United States As more indus-trialized countries were added to the compar-ative template it was shown that Sweden andthe Netherlands are more fluid (Erikson et al1982 Ganzeboom and De Graaf 1984) andthat Germany Japan and Ireland are morerigid (Hout and Jackson 1986 Ishida 1993Muumlller 1986) However the most striking find-ing was the significant similarity across coun-tries despite the different historicalbackgrounds and institutional arrangements

On the basis of the CASMIN project a rel-ative consensus has emerged favoring thehypothesis of substantial international simi-larity in mobility patterns Similarity does notmean complete homogeneity however Therecertainly exist some national deviations butit is argued they can be explained by highlyspecific historically based institutional factorsrather than by systematic relationship to othervariable characteristics of national societiesThere seems to be only one exception to thisldquounsystematic fluctuationrdquo Comparison of theCASMIN countries found that a small but sig-nificant portion of the international variationin mobility is related to economic inequalitywith higher inequality leading to less fluidity(Erikson and Goldthorpe 1992a chapter 12also see Tyree Semyonov and Hodge 1979for an earlier analysis)

This finding supports the resource approachlinking the two distributive phenomenaAdditional evidence consistent with theresources approach was presented by Jonssonand Mills (1993) who compared Sweden andEngland and found higher fluidity in the more

equal Swedish society by Bjorklund and Jantti(1997) who compared earnings mobility inthe United States and Sweden and found mobil-ity to be higher in Sweden and by analyses ofintergenerational earnings mobility in severalOrganization for Economic Cooperation andDevelopment (OECD) countries which sug-gest higher mobility in more equal nations(Solon 2002)2

However a more recent comparative analy-sis of mobility trends between 1970 and 1990in 11 industrialized countries does not findevidence of a relationship between inequalityand fluidity (Breen and Luijkx 2004a396)This supports an earlier indication of no asso-ciation between the two distributional phe-nomena (Grusky and Hauser 1984)

In sum evidence about a potential mobili-ty-inequality association based on cross-coun-try comparisons is inconclusive not onlybecause of divergent f indings but alsobecause of the small number of countriesincluded and the potentially high collinearitybetween inequality and other explanatory fac-tors

Empirical analysis of mobility trends with-in countries has not provided a conclusiveanswer either The CASMIN finding of ldquocon-stant social fluidityrdquo over time has been recent-ly tested in the aforementioned comparativeanalysis of mobility in 11 countries Thisanalysis including 10 European nations andIsrael (Breen 2004) finds growing fluidity insome countries but null or slight temporalchange in others While Britain Israel andless conclusively Germany display ldquoconstantfluidityrdquo some indication of growing open-ness is detected in France Hungary IrelandItaly the Netherlands Norway Poland andSweden However in France Hungary Polandand Sweden all change occurred between the1970s and 1980s and stability has prevailedsince then (Breen and Luijkx 2004b54)Changes in Ireland and Italy are quite minor(Layte and Whelan 2004 Pisati andSchizzerotto 2004) and only the Netherlandsdisplays a sustained increase in mobility overthe entire period considered (Ganzeboom andLuijkx 2004) Mobility has also been found to

442266mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

2 To my knowledge there is no empirical evidencesupporting the incentive approach

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

increase over time in the United Statesbetween the early 1970s and the mid 1980s(Hout 1988) However these increases in flu-idity are not univocally preceded by a reduc-tion of inequality thus casting doubts on apotential association between inequality andmobility

A careful analysis of mobility trends inRussia shows a significant decline in mobili-ty associated with an increase in inequalityafter the market transformation of the early1990s (Gerber and Hout 2004) which pro-vides additional support for the resourceapproach The advantage of the Russian studyis that ldquothe market transition in Russia ||altered so many fundamental economic insti-tutions so rapidly that we can confidentlyascribe changes in social mobility || to thissource rather than to cultural change or indus-trializationrdquo (Gerber and Hout 2004678) It isnot clear however whether it is growth ininequality recession some other change asso-ciated with the radical liberalization of theeconomy or a combination of these factorsthat triggered a decline in fluidity in contem-porary Russia

In this context Chile presents an ideal casefor an examination of the association betweeninequality and mobility Given the extremeeconomic inequality in the country if theunequal distribution of resources or incentiveshas an impact on mobility opportunities theChilean level of fluidity should be signifi-cantly different from that of industrializednations Additionally given the increase ininequality associated with the market reformof the 1970s and 1980s analysis of mobilitybefore and after the reform would provide sup-plementary evidence concerning the potentialmobilityndashinequality relationship Naturally asingle case study will not supply a definitiveanswer to these questions However by com-bining an examination of the Chilean mobili-ty regime in the context of historicaltransformations in the country with an inter-national comparative analysis and with anassessment of mobility trends over time thisarticle provides important insights into theexistence of a link between mobility andinequality and the mechanisms driving it

TTHHEE CCHHIILLEEAANN CCOONNTTEEXXTT IINNCCOOMMPPAARRAATTIIVVEE PPEERRSSPPEECCTTIIVVEE

During the second half of the 20th centuryChile transformed from an agrarian semifeu-dal society into an urban service-based oneBetween 1950 and 2000 the rural populationdeclined from almost 40 to 17 percent (Braunet al 2000 INE 2002) This defines Chile asa mostly urban country with an 83 percentrate of urbanization larger than the 78 percentrate in the United States In tandem with urban-ization Chile experienced a reallocation ofemployment from the agricultural to the terti-ary sector of the economy The share of agri-culture in total employment declined from 38percent in 1950 to 17 percent in 2000 Whereasthe share of manufacturing remained constantat about 18 percent the share of the servicesector rose from 42 to 65 percent (Braun et al2000 INE 2002)

Urbanization and tertiarization are process-es that virtually all countries experienced dur-ing the 20th century Within this secular trendthe Chilean political economy is marked byspecific institutional developments that shapeits stratification structure From the 1940s tothe 1970s the Chilean economic landscape asthat of its Latin American neighbors wasdefined by import-substitutive industrialization(ISI) Emerging as a reaction to the collapse ofinternational trade caused by the GreatDepression (Ellworth 1945) and based on theldquodeterioration of the terms of traderdquo theory(Prebisch 1950) ISI was based on two typesof policies The first was oriented to closing theChilean economy to international markets andthe second was oriented to promoting nation-al industrialization The Chilean state becamethe leading productive agent supporting indus-try through credit investments and technicalassistance and taking a direct productive rolethrough the creation of public enterprises(Stallings 1978) After a sanguine beginningwith industrial production growing at almost7 percent per year between 1940 and 1950(Mamalakis 1976 Munoz 1968) ISI started tofail economic growth stagnated and socialturmoil resulting from massive urban migra-tion and vast social inequalities increased

In 1964 a progressive administration tookpower and adopted as its mandate the correc-tion of extreme inequality in the country by a

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442277

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ldquoRevolution in Freedomrdquo (Gazmuri 2000)This progressive government launched redis-tributive policies including an agrarian reformand an educational reform The redistributiveagenda boosted by a socialist administrationthat came to power in 1970 was abruptly halt-ed by a military coup in 1973 The militarytook power and retained it until 1990 Duringthese 17 years the military regime conduct-ed a deep market-oriented transformationNow turned into the ldquoWashington consensusrdquoparadigm (Williamson 1990) this reformincluded macroeconomic stabilization dereg-ulation of prices and markets and the priva-tization of enterprises and social services Ittransformed Chile from a closed economywith heavy state intervention into one of themost open market-based economies in theworld with the productive and welfare role ofthe state reduced to a minimum (Edwards andCox-Edwards 1991 Martinez and Diaz 1999Meller 1996 Velasco 1994)

The depth of the market reform coupledwith a world recession led in the late 1970sand early 1980s to the deepest economic cri-sis since the Great Depression A third of thelabor force was unemployed and povertyafflicted nearly half of Chilean households(Meller 1991) The post-crisis recovery start-ing in the late 1980s was substantial and sus-tained and coincided with theredemocratization of the country The grossdomestic product (GDP) per capita grew morethan 6 percent annually for 15 years a novelrate for Chile only comparable in recent timesto the ldquoEast Asian Miraclerdquo (World Bank1993) The sharp economic growth has trans-formed Chile from one of the poorest coun-tries in Latin America (Hofman 2000) into aldquomiddle-income economyrdquo (World Bank2003a) In the year 2000 the income per capi-ta was approximately US$5000 much lowerthan the United States average of US$31910but the highest in Latin America (World Bank2003a) As a consequence of the sharp eco-nomic growth Chileans have reached levels ofconsumption unthinkable two decades agoand the poverty rate fell from 45 to 21 percentbetween 1985 and 2000 (MIDEPLAN 2000Raczynski 2000)

TTHHEE CCHHIILLEEAANN LLEEVVEELL AANNDD PPAATTTTEERRNN OOFF

IINNEEQQUUAALLIITTYY

The dark side of this ldquosuccess storyrdquo is eco-nomic inequality Structurally rooted in a feu-dal agrarian structure the institutional legacyof the colonial period and in the slow expan-sion of education (Engerman and Sokoloff1997 World Bank 2003b) inequality hasremained persistently high during the 20thcentury The Gini coefficient reached 58 in the1990s which compares with a much lowerGini of 34 among the industrialized coun-tries and is large even in the highly unequalLatin American context with its average Giniof 49 (Deininger and Squire 1996 Marceland Solimano 1994) As Figure 1 indicatesChilean inequality is almost twice that in mostindustrialized countries and 15 times that inthe United States the most unequal nation ofthe industrialized world

Not only the level but also the pattern ofinequality in Chile significantly departs fromthat of the industrialized world With thewealthiest Chilean decile receiving 423 per-cent of the total national income (MIDEPLAN2001) the Chilean pattern of inequality ischaracterized by high ldquoconcentration at thetoprdquo Although inequality is by definition relat-ed to concentration the Chilean case isextreme as compared with the industrializedworld and even with other Latin Americancountries A comparison between the incomeof each decile and the income of the preced-ing (poorer) decile illustrates the point Theratio between the wealthiest and the secondwealthiest decile is twice as large in Chile asin the United States and England and one ofthe largest in Latin America depicting highelite concentration In contrast the ratiobetween the second poorest and the poorestdeciles in Chile is half that of the United Statesand England indicating that inequality at thebottom of the income distribution is muchlower in Chile than in these industrializednations (Szekely and Hilgert 1999)

In fact as Figure 2 indicates Chile is thefourth most unequal country in the mostunequal region of the world However if thewealthiest decile is excluded Chilean inequal-ity is dramatically reduced and Chile becomesthe most equal Latin American country evenmore equal than the United States (Inter-American Development Bank 1999)

442288mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442299

2322-ASR 703 filename70304-torche

Figure 1 Income Inequality in Chile and Industrialized Countries Circa 2000

Note Jap = Japan Swe = Sweden Ita = Italy Ger = Germany Neth = Netherlands Fra = France Ire = Republicof Ireland Eng = England Source World Bank 2001

Figure 2 Gini Coefficient for Total Population and Excluding Wealthiest Decile Chile other Latin AmericanCountries and Unites States in 1998

Note Par = Paraguay Bra = Brazil Ecu = Ecuador Arg = Argentina Bol = Bolivia Ven = Venezuela Uru =Uruguay Source Inter-American Development Bank 1999

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Therefore Chile is unequal largely because theelite concentrates an extremely high proportionof the national income Across the noneliteclasses the distribution of resources is muchmore uniform

Inequality is not a new development inChilean society but rather has deep historicalroots Figure 3 presents the longest availableseries on inequality in Chile depicting earningsdistribution since 1957 in Santiago3 and show-ing persistently high inequality over the lasthalf century However there is significant vari-ation over time a short decline of inequality dur-ing the progressive administrations of themid-1960s and early 1970s an increase since themilitary regime took power in 1973 a peak in1984 and a small decline after the democratictransition to levels still higher than those thatpreceded the military regime

LLAATTEE IINNDDUUSSTTRRIIAALLIIZZAATTIIOONN MMAARRKKEETTRREEFFOORRMM AANNDD TTHHEE CCHHIILLEEAANN CCLLAASSSSSSTTRRUUCCTTUURREE

The ldquocommon social fluidityrdquo hypothesis assertsthat all international variation in mobility pat-terns is attributable to highly specific histori-cally formed national characteristics In theChilean case these factors may be related to themarket reform of the 1970s and 1980s and therapid economic growth since the late 1980sNational mobility studies suggest that coun-tries experiencing rapid industrialization may becharacterized by weaker sector barriers to mobil-ity specif ically barriers between the self-employed and employees and betweenagricultural and nonagricultural classes(Goldthorpe Yaish and Kraus 1997 and Yaish2004 for Israel Ishida Goldthorpe and Erikson1991 for Japan Park 2004 for Korea and Costa-Ribeiro 2003 for Brazil)

In addition two components of the Chileanmarket transformation may have contributed tothe weakness of mobility barriers across sectorsof the economy The first component is theagrarian reform and subsequent counterreformundertaken by the military regime The secondone is the opening of the economy to interna-tional trade and the retrenchment of the statewhich led to a decline in formal employment

443300mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Figure 3 Earnings Inequality in Greatest Santiago 1957ndash2000Source Larranaga 1999 and calculations by the author

3 Santiago the Chilean capital comprises aboutone-third of the countryrsquos population Assessmentby Chilean experts suggests that Santiago trendspresent an unbiased picture of trends at the nationallevel Although earnings inequality is not as com-prehensive a measure as total income inequality it isan adequate much more accurately measured proxy(Galbraith 2002)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

and to the growth and segmentation of the self-employed sector

The main objective of the Chilean agrarianreform (1962ndash1973) was to reduce the extremeland concentration in the country The reformwas very successful In 1955 the top 7 percentof landowners held 65 percent of the land andthe bottom 37 percent held only 1 percent By1973 the last year of the reform 43 percent ofthe land had been expropriated (Kay 2002 Scott1996) An agrarian counterreform was launchedby the military regime as a way of returning theland to its former owners It did not howeverrestore the traditional hacienda order (Gomezand Echenique 1988 Rivera 1988) In fact themilitary government returned only one-third ofthe plots to their old owners sold another thirdand adjudicated the remaining third to the smallproprietors benefited by the agrarian reformUnable to compete with the now cheap foodimports many small proprietors opted to selltheir plots As a consequence as much as two-thirds of the Chilean agricultural land came upfor sale creating an active land market Themain beneficiaries of the ldquomarketization of thecountrysiderdquo were a new group of export-ori-ented entrepreneurs and international investorswho bought large tracts of land (Gwynne 1996Kay 2002) Thus the reform and counterreformtriggered a massive ldquochange of handsrdquo in theChilean countryside likely altering the patternsof land inheritance

The second major effect of the market reformon the class structure was the reduction of for-mal employment caused by the decline of theindustrial working class and by the shrinkage ofthe state Incapable of competing with the nowcheap imports the working class plummetedfrom 34 percent of total employment in theearly 1970s to 20 percent in 1980 while pub-lic employment dropped from 14 to 84 per-cent in the same period (Leon and Martinez2000 Schkolnick 2000 Velasquez 1990) Thedecline in industrial and public employmentled to an increase in self-employment fromabout 15 percent in 1970 to about 28 percent in1980 (Thomas 1996) followed by a declinethrough the economic recovery before stabi-lization at about 22 percent of the total employ-ment in 2000 (MIDEPLAN 2001) In contrastto industrialized countries in which the self-employment rate usually is within the one-digitrange and involves capital ownership in Chile

it encompasses more than one-fifth of the laborforce Furthermore self-employment hadbecome a survival strategy for a large numberof former industrial and public employeesappropriately labeled ldquoforced entrepreneurial-ismrdquo (Infante and Klein 1995 Portes Castellsand Benton 1989) As a consequence theChilean self-employed sector is voluminousand segmented It includes a small segment ofentrepreneurs oriented to capital accumulationand able to hire employees and a large seg-ment of self-employed workers mostly engagedin ldquosurvival activitiesrdquo and unable to hire thelabor of others Differences in economic well-being between these two groups are massivewith the former group earning on averagethree times more than the latter (MIDEPLAN2001)

To be sure a large self-employed sector is nota result of the market transformation Self-employment has been historically high in LatinAmerica because according to some scholarsthe region has experienced ldquodependent inte-grationrdquo into the world capitalistic system(Hopkins and Wallerstein 1982 Luxembourg1951) However the internal heterogeneity ofthe self-employed sector increased during themarket reform in a pattern that foretold whathappened in the rest of Latin America during the1990s (Klein and Tokman 2000 Portes andHoffman 2003) and what may be current devel-opments in the industrialized world (Arum andMuumlller 2004 Noorderhaven et al 2003) Thesecharacteristics of the Chilean class structureare considered in the comparative analysis ofChilean mobility

DDAATTAA AANNDD AANNAALLYYTTIICCAALL AAPPPPRROOAACCHH

This study uses the 2001 Chilean MobilitySurvey (CMS) The CMS is a nationally repre-sentative multistage stratified sample of maleheads of household ages 24 to 69 The samplingstrategy includes the following stages First 87primary sampling units (PSUs) (counties) areselected Then blocks within the PSUs areselected and finally households within blocksare chosen Counties are stratified according tosize (fewer than 20000 20000 to 100000100000 to 200000 and more than 200000inhabitants) and by geographic zone (NorthCenter South) All PSUs in the large size stra-tum are included in the sample to increase effi-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443311

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ciency The fieldwork conducted between Apriland June 2001 consists of face-to-face inter-views in the respondentrsquos household carried outby trained personnel The survey excludesnonndashhead-of-household males which represent17 percent of the male population of the rele-vant age (MIDEPLAN 2001) Among thoseexcluded 865 percent are the sons of the headof the household 124 percent are another rel-ative of the head and 11 percent are other non-related males Their occupational distributionwith control used for age is almost identical tothat of the heads of household The small pro-portion represented by this group and their sim-ilar occupational distribution suggest that theirinclusion would not significantly alter the find-ings presented in this article Excluding thehouseholds not eligible for the survey theresponse rate is 63 percent Although nonre-sponse rates usually are not reported in Chileansurveys exchange with Chilean experts indi-cates that the nonresponse rate is about 20 to 25percent for face-to-face household surveys Thehigher nonresponse rate of the CMS is likelyattributable to the difficulty contacting maleheads of household4 The total sample size is3544 I exclude individuals outside the agerange of 25 to 64 years which is conventional-ly used in comparative mobility research andcases with unusable data After this exclusionthe usable sample size is 3002

CCLLAASSSS SSCCHHEEMMAA

This study uses a class perspective to describethe Chilean social structure It utilizes the seven-category version of the classification devel-oped by the CASMIN project Thisclassification is widely used in internationalcomparative research and describes the basic

stratification of advanced industrial societiesbased on ldquoemployment relationshipsrdquo (Eriksonand Goldthorpe 1992a35ndash47 Goldthorpe andHeath 1992) The seven classes included in theschema are I+II (service class) III (routine non-manual workers) IVab (self-employed work-ers) V+VI (manual supervisors and skilledmanual workers) VIIa (unskilled manual work-ers) IVc (farmers) and VIIb (farm workers)

Use of the 7-class categorization instead ofan alternative detailed 12-class version pro-duced by the CASMIN project (Erikson andGoldthorpe 1992b) induces some loss of infor-mation about the originndashdestination associa-tion A global test of all five aggregationsobtained from the ratio of L2 tests for the inde-pendence model in the collapsed and full tablesindicates that the seven-class schema masks 29percent of the association shown by the full setof classes ([L2 full table ndash L2 collapsed table]L2 full table = [108644 ndash 77191]108644 =29) This proportion is high compared withthat of the CASMIN countries and is surpassedonly by Sweden with 31 percent (Hout andHauser 1992 Table 2) A collapsed sevenfoldclass schema was however preferred because ofthe moderate sample size of the Chilean surveyand because it grants international comparabil-ity of the findings5

Operationalization of this class classifica-tion is based on detailed information concern-ing job title (recoded into the standard ISCO-88classification [ILO 1990]) industry occupa-tional status and supervisory status of workersBecause the CMS is specifically designed toassess class mobility it includes all the infor-mation necessary to produce the CASMIN classschema The CASMIN project has not produceda standard algorithm to generate the class clas-sification but other researchers have producedsuch an algorithm which I use in this analysis(Ganzeboom and Treiman 2003) Table 1 pres-ents the basic intergenerational mobility tableto be used in this analysis including the countsand marginal percentage distribution for originand destination classes

443322mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

4 The nonresponse rate can yield bias if those whowere unreachable or refused to participate differ fromthose included in the sample To estimate the mag-nitude of this bias I compared the distribution of keyvariables with the CASEN 2000 survey CASEN isa large survey (n = 252595) conducted by the Chileangovernment and has a refusal rate lower than 10percent The comparison (available upon request)suggests that the CMS slightly underrepresents agri-cultural workers and the upper class Overall how-ever there is no indication of major nonresponsebias

5 Analyses were replicated using the detailed 12-class table and no difference in substantive find-ings was found

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

MMEETTHHOODDSS

I fit two alternative models to the Chilean mobil-ity table The first is the ldquocore model of social flu-idityrdquo which claims to capture basic internationalsimilarity in mobility Although the core modelhas become a benchmark for mobility researchand currently is the only model that provides acomparative framework for mobility analysisanalysts have drawn attention to a number of itslimitations The model has been criticized forinadequately representing hierarchical mobilityeffects for being tailored to a few industrializedcountries for the ad hoc nature of some effects itincludes (specifically the affinity effects) andfor lacking a criterion to determine how muchdivergence is necessary for it to be rejected(Ganzeboom Luijkx and Treiman 1989 Houtand Hauser 1992 Sorensen 1992 Yamaguchi1987) I therefore also test an alternative ldquohybridmodelrdquo that combines the rowndashcolumn (II) asso-ciation model (Goodman 1979) with parametersaccounting for class immobility The internationaland temporal comparison of Chilean mobilitypatterns uses the uniform-difference (UNIDIFF)model also known as the multiplicative layereffect model (Erikson and Goldthorpe 1992aXie 1992) Model comparison and selection isbased on the Bayesian Information Criterion(BIC) statistic (Raftery 1995)

AANNAALLYYSSIISS OOFF CCHHIILLEEAANN MMOOBBIILLIITTYY IINNCCOOMMPPAARRAATTIIVVEE PPEERRSSPPEECCTTIIVVEE

TTHHEE CCOORREE MMOODDEELL OOFF SSOOCCIIAALL FFLLUUIIDDIITTYY

I start by fitting the core model to the Chileanmobility table The core model represents the

basic international commonality in social flu-idity using a topological formulation In contrastto standard topological models constructed froma single allocation of cells (Hauser 1978) thecore model uses eight matrices each of whichis designed to capture a particular effect thatenhances or reduces mobility between specificclasses (Erikson and Goldthorpe 1987a 1992achapter 4) In this way the model permits thesources of commonality across nations to beidentified clearly and likewise the departuresfrom it The effects are of four types hierarchyinheritance sector and affinity6

HIERARCHY EFFECTS Hierarchy effects reflectthe impact of status distances between classeson fluidity between them To estimate theseeffects the mobility table is divided into threestrata reflecting differences in resources andrewards across classes The strata are the fol-lowing upper stratum (class I+II) middle stra-tum (classes III IVab IVc and V-VI) and lowerstratum (classes VIIa and VIIb) Because thereare three strata two hierarchy effects (HI1 andHI2) can be identified each representing thecrossing of one additional hierarchical barrierThe association between each pair of classes isexpressed as an inverse function of the numberof hierarchical strata crossed

INHERITANCE EFFECTS Inheritance effects aredesigned to capture the propensity for class

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443333

2322-ASR 703 filename70304-torche

Table 1 Seven-Class Intergenerational Mobility Table for Chilean Men

Class of Destination

I-II III IVab IVc V-VI VIIa VIIb N

I-II Service Class 229 49 81 2 37 30 2 430 14III Routine Nonmanual 61 15 34 0 24 27 1 162 5IVab Self-employed 123 37 173 28 102 90 20 573 19IVc Farmers 42 16 68 54 58 54 24 316 11V-VI Skilled Manual 91 33 116 13 155 135 36 579 19VIIa Semiskilled and Unskilled Manual 49 43 115 13 111 123 44 498 17VIIb Farm Workers 28 16 78 17 89 101 115 444 15N 623 209 665 127 576 560 242 3002 21 7 22 4 19 19 8

Note Cell entries represent counts

6 Design matrices in Erikson and Goldthorpe(1992a124ndash129)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

immobility They include three matrices IN1identifies immobility across all classes alike IN2identifies the higher immobility of the serviceclass (I+II) the self-employed group (IVab) andfarmers (IVc) and IN3 accounts for the highestimmobility of farmers (IVc)

SECTOR EFFECTS Sector effects are intended toidentify the difficulty of moving between theagricultural and nonagricultural sectors of theeconomy (SE1)

AFFINITY EFFECTS Affinity effects are intend-ed to capture specific discontinuities (negativeaffinities) or linkages (positive affinities) betweenclasses which either reinforce or offset the over-all effects of hierarchy and sector There are twoaffinity effects The first AF1 identifies the dif-ficulty of moving between the service class (I+II)and the class of farm workers (VIIb) which addsto their hierarchical distance In contrast AF2identifies instances in which mobility is morefrequent than accounted for by hierarchy and sec-tor effects and includes affinities within the non-manual sector (I+II and III) and within the manualsector (V+VI and VIIa) a symmetrical affinitybased on capital possession (IVab and IVc and Iand IVab) and an asymmetrical link between agri-cultural classes of origin (IVc and VIIb) and theunskilled manual class of destination (VIIa)7

TTHHEE MMOOBBIILLIITTYY PPAATTTTEERRNN IINN CCHHIILLEECCOONNVVEERRGGEENNCCEE OORR DDIIVVEERRGGEENNCCEE

Table 2 presents the fit of several models appliedto the Chilean data and Table 3 displays theparameter estimates for each of these modelsModel 1 in Table 2 indicates that the fit of theoriginal core model to the Chilean table is unsat-isfactory under standard statistical criteria butit explains a significant 775 percent of associ-ation under independence To assess the strengthof the different factors driving mobility oppor-tunities I compare the Chilean coefficients withthose obtained for the CASMIN countriesCoefficients from the original core model fit-ted to the Chilean table are reported in row 1 ofTable 3 CASMIN coefficients are reported inrow 2 and differences in coefficientsrsquo magni-tude and the statistical significance of the dif-ferences are reported in row 38

All Chilean coefficients are significant andhave the same sign as for the CASMIN coun-tries Comparison of the magnitude of the coef-ficients however shows significant departuresfrom the core for five effects which revealinteresting differences between Chile and theCASMIN countries The hierarchy coefficientsindicate that whereas the short-range hierar-chical barriers are somewhat weaker in Chile

443344mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

8 CASMIN coefficients were obtained from fittingthe core model to the combined table of France andEngland standardizing both tables to have 10000cases as recommended by Erikson and Goldthorpe(1992a121) The coefficients obtained are virtuallyidentical to those presented by Erikson andGoldthorpe (1992a Table 44)

Table 2 Fit statistics selected models to the Chilean Mobility Table

AssociationL2 BIC df p Explained

Core Modelmdash1 Core Model Original 1375 285 28 000 775mdash2 Core Model EHE 1081 ndash08 28 000 860mdash3 Chilean version of Core Model ([Model 2] + Chilean Parameters) 724 ndash326 27 000 906Association Models 000mdash4 Heterogeneous Quasi-RC(II) Model 734 ndash161 23 000 905mdash5 Homogeneous Quasi-RC(II) Model 803 ndash287 28 000 896mdash6 Quasi-Linear by Linear Association (SES row and column ranking)a 1563 279 33 000 798

Note EHE = empirical hierarchical effects SES = eocioeconomic statusa Linear-by-linear association model using socioeconomic status scores as presented in Appendix Figure A1 torank origin and destination classes

7 For a detailed account of the empirical derivationof the model see Erikson and Goldthorpe (1987a1987b and 1992a chapter 4)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443355

2322-ASR 703 filename70304-torche

Tab

le 3

M

odel

Par

amet

er E

stim

ates

for

the

Chi

lean

Mob

ilit

y Ta

ble

Eff

ect

1 C

ore

mod

el o

rigi

nala

mdash 2 C

ore

mod

el C

AS

MIN

b

mdash 3 D

iffe

renc

e C

ASM

INmdash

Chi

lec

mdash 4 C

ore

mod

el E

HE

d

mdash 5 C

hile

an v

ersi

on o

f co

ree

mdash Not

eD

ata

show

n ar

e m

odel

par

amet

er e

stim

ates

wit

h st

anda

rd e

rror

s in

par

enth

eses

EH

E =

em

piri

cal h

iera

rchi

cal e

ffec

ts

aM

odel

1 f

rom

Tab

le 2

b

Ori

gina

l Cor

e M

odel

fit

ted

to th

e E

ngli

sh a

nd F

renc

h m

obil

ity

tabl

es S

ee te

xt f

or d

etai

ls

cD

iffe

renc

e in

par

amet

er m

agni

tude

bet

wee

n or

igin

al c

ore

mod

el f

itte

d to

the

Chi

lean

tabl

e (r

ow 1

) an

d to

the

CA

SM

IN ta

ble

(row

2)

dM

odel

2 f

rom

Tab

le 2

e

Mod

el 3

fro

m T

able

2

dagger p

lt 1

p

lt 0

5

p

lt 0

1

p

lt 0

01 (

two-

tail

ed)

HI1

ndash12

(

05)

ndash22

(02

)ndash

10 dagger

(06

)ndash

03(

07)

ndash09

dagger(

06)

HI2

ndash65

(11

)ndash

45

(

04)

20

dagger(

12)

ndash74

(09

)ndash

47

(

10)

IN1

41

(

08)

43

(

03)

Insi

g

37

(

07)

34

(

07)

IN2

39

(

10)

84

(

04)

45

(

11)

31

(13

)mdash

IN3

68

(

24)

101

(11

)In

sig

61

(24

)4

8(

23)

SE

1

ndash38

(07

)ndash1

06

(

05)

ndash68

(08

)ndash

42

(

07)

ndash20

(

08)

AF

1

ndash76

(21

)ndash

83

(

13)

Insi

g

ndash84

(20

)ndash

63

(21

)

AF

2

33

(

05)

45

(

02)

ndash12

(

06)

20

(

06)

26

(

06)

AF

1ndashC

mdash mdash mdash mdash

ndash14

9

(46

)

IN2ndash

C

mdash mdash mdash mdash 71

(

14)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(HI1) long-range hierarchical mobility is moredifficult in this country (HI2) This suggestssubstantial inequality between the extremes ofthe class hierarchy combined with smaller dif-ferentiation in the middle of the distribution Asto the inheritance effects the service class theself-employed group and the farmers (classesI+II IVab and IVc respectively) have a sig-nificantly lower propensity to immobility inChile than in the CASMIN countries suggest-ing that class positions usually associated withindependent work are less able to reproducetheir class status intergenerationally than depict-ed by the core Most impressive the barrierbetween the agricultural and nonagriculturalsectors of the economy (SE1) is much weakerin Chile as is the affinity effect linking non-manual classes manual classes and the class-es that own capital among themselves (AF2)

Interestingly all the significant differences incoefficients with the single exception of the bar-rier to long-range hierarchical mobility sug-gest that Chile is more fluid than the core modeldepicts Higher fluidity seems to be driven bythe weakness of horizontal barriers separatingthe agricultural manual and self-employed sec-tors of the economy This is consistent withfindings of weaker sector effects in countriesthat have experienced rapid industrializationsuch as Korea (Park 2004) Brazil (Costa-Ribeiro 2003) Israel (Yaish 2004 Goldthorpeet al 1997) and Japan (Ishida et al 1991)

Additionally in the Chilean case the weak-ness of sector barriers may have been intensi-fied by the deep market transformation of the1970s and 1980s On the one hand the ldquochangeof handsrdquo in the agricultural sector induced bythe agrarian reform and counterreform likelyincreased fluidity between the agricultural andurban classes On the other hand the role of self-employment as an ephemeral refuge against(industrial and public sector) unemploymentmay have reduced barriers between the self-employed group and the rest of the social struc-ture Added to the larger difficulty of long-rangehierarchical mobility these features suggestweak sector cleavages in the Chilean fluiditypattern and high salience of the hierarchicaldimension of mobility

For the countries in which the fit of the coremodel is not adequate Erikson and Goldthorpe(1992a145) suggest adding parameters thataccount for country-specific deviations from

the model I therefore introduce adjustments toreflect salient characteristics of the Chilean flu-idity pattern producing a ldquoChilean versionrdquo ofthe core model

First I modify the hierarchical effects Fittingof the core model suggests that hierarchicalbarriers are a crucial determinant of mobility inChile This conclusion is however obscuredby the fact that the hierarchical strata distin-guished by the core model are not empiricallyobtained and may not accurately represent theChilean ranking of classes in terms of socioe-conomic status (SES) To model hierarchicaleffects in Chile adequately I rank classes usingthe unweighted average of their schooling andearning levels as a proxy for SES I then collapsethe seven classes into three hierarchical stratausing cluster analysis These three empiricallyobtained strata are the following Upper stratum(service class I+II) middle stratum (routinenonmanual and self-employed group III andIVab) and lower stratum (manual and agricul-tural classes V-VI VIIa IVc and VIIb) Detailson the ranking of classes cluster procedures andthe comparison of the CASMIN and empiricalChilean class ranking can be found in theAppendix Comparison of the CASMIN and theempirical class ranking suggests higher differ-entiation at the top than at the bottom of thesocial structure in the Chilean case

Model 2 in Table 2 uses the empiricallyobtained hierarchical strata (EHE) Theimprovement in fit is massive (the difference inBIC is 293 without using any degrees of free-dom) The model now accounts for 86 percentof the association under independence

The large improvement in fit when empiricalhierarchical strata are used challenges theassumption of an internationally homogeneoushierarchical class ranking At the empiricallevel there is no proof that the hierarchicalranking of classes assumed by the core modelfits all CASMIN countries9 Moreover someevidence points to significant cross-country dif-

443366mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

9 Erikson and Goldthorpe(1992a Table 22) pres-ent the average ranking of classes based on standardstatus measures but they do not show the empiricalranking of classes for each country If the occupa-tional composition of classes varies across countriesthere may be significant international differences inthe hierarchical ranking of classes

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ferences in the ranking of classes (Hout andHauser 1992 Sorensen 1992) At the theoreti-cal level the creators of the core model high-light the relevance of highly specific historicallyformed national attributes to explain interna-tional variation in mobility These national par-ticularities may very well result in differenthierarchical positions of classes across coun-tries10 Furthermore even if the homogeneousranking of classes imposed by the CASMINresearchers applied to most CASMIN coun-tries significant departures may be found inthe late industrialized countries currently beingadded to the comparative mobility project TheChilean findings suggest that instead of impos-ing international homogeneity to the hierarchi-cal dimension of mobility the ranking of classesshould be treated as nationally specific param-eters to be estimated as mobility researchersroutinely treat row and column marginals ofthe mobility table The issue is importantbecause the hierarchical dimension is the mostrelevant dimension of mobility (Hout andHauser 1992 Wong 1992) and inadequate mod-eling can bias assessment of not only hierar-chical effects of mobility but also other effects

Two other changes were introduced toaccount for Chilean-specific deviations fromthe core model First the IN2 effect wasrearranged to exclude the self-employed class(IVab) and to include the class of farm workers(VIIb) This rearrangement accounts for thefact that the Chilean self-employed classappeared less likely than depicted by the coremodel to reproduce its class position acrossgenerations and the inheritance level of Chileanfarm workers is higher than in the CASMINcountries These changes are expressed in theChilean-specific inheritance parameter IN2-CSecond an affinity parameter (AF1-C) wasadded to capture the difficulty of long-rangedownward elite mobility This negative asym-metrical affinity links flows from the service

class to the agricultural classes and from theroutine nonmanual class to the class of farmworkers Note that the AF1-C adds to the hier-archical distance between the two extremes ofthe class structure and to the negative affinitycaptured by AF1 to depict the extreme difficultyof long-range downward mobility from theChilean elite The fit of this lsquoChilean versionrsquoof the core model is reported in Table 2 model3 The improvement in BIC from model 2 is sub-stantial Furthermore even if the model fallsshort in terms of standard statistical criteria itaccounts for a large 91 of the associationunder independence thereby supporting theclaim that it adequately represents the Chileanmobility pattern The parameter estimates for theChilean version of the core are reported in row5 Table 3

Because of the core modelrsquos undesirable prop-erties I also use an alternative approach toexamine Chilean mobility patterns Thisapproach uses the now standard associationrowndashcolumn (II) (RC[II]) model (Goodman1979 see also Hout 1983 and Wong 1992) Inthis model both origin and destination classesare scaled so that the association can beexpressed as a linear-by-linear interaction by asingle parameter The multiplicative form ofthe model can be expressed as follows

Fij = iO j

D exp( microi νj)

where i indexes class of origin j indexes classof destination Fij is the expected frequency inthe (ij) cell is the grand mean i

O pertainsto the class of origin marginal effect j

D pertainsto the class of destination marginal effect isa global association parameter microi is the scalevalue for the ith class of origin and νj is the scalescore for the jth class of destination subject tothe following normalization constraints Σmicroi =Σνj = 0 (normalization of the location) and Σmicroi2

= Σνj2 = 1 (normalization of the scale)The class scale scores reflect a latent con-

tinuous variable made manifest by the class cat-egories Empirically obtained from the datathese ldquodistancerdquo scores produce an optimalranking of classes for the purpose of mobilityanalysis The RC(II) model is appropriate formodeling the association of any two ordinalvariables but mobility tables are distinctivebecause of the correspondence between class oforigin and class of destination (Gerber and Hout2004691)

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443377

2322-ASR 703 filename70304-torche

10 For instance during the socialist period inHungary the status of the proletariat may have beenindistinguishable from the status of the routine non-manual class (Szelenyi 199862) In Scotland givento early industrialization the hierarchical barrierbetween the skilled and unskilled working classes thatcharacterizes other CASMIN countries may not exist(Erikson and Goldthorpe 1992a163)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I introduce two adjustments to account forthis correspondence transforming the formu-lation into a ldquohybrid modelrdquo First topologicalparameters will account for the large counts insome diagonal cells reflecting immobility(inheritance) effects These parameters distin-guish two inheritance levels ldquohigh inheritancerdquofor the service class (I+II) farmers (IVc) andfarm workers (VIIb) and ldquolow inheritancerdquo forall other classes with the exception of the rou-tine nonmanual workers

Model 4 in Table 2 reports the fit statistic forthis quasi-RC(II) model The BIC statisticrejects it and prefers the more parsimoniousChilean version of the core model ThereforeI introduce a second adjustment to account forthe correspondence between rows and columnsin mobility tables Origin and destination scoresare constrained to be the same assuming thatthe scaling of classes has not significantlychanged over time Model 5 in Table 2 reportsfit statistics for this homogeneous quasi-RC(II)model On the basis of the BIC statistic the fitof the model is as good as that of the Chileanversion of the core model (Wong [1994] showsthat BIC differences of less than 5 points shouldbe considered indeterminate) Thus under stan-dard statistical criteria the Chilean core modeland the homogeneous quasi-RC(II) model areindistinguishable Given that the Chilean versionof the core model was tailored to fit the Chileantable the fit of model 5 is impressive and it indi-cates that a unidimensional scale can adequatelycapture mobility distances between classes

The logical next question is what this scalerepresents and specifically whether it repre-sents the hierarchical ranking of classes in termsof SES If a close correspondence is foundbetween the mobility distances among classesand their distances in terms of SES this find-ing would suggest that the intragenerationaldistribution of resources and rewards expressedin the SES ranking of classes drives the processof intergenerational mobility In other words thisfinding would support the resource approachrsquosclaim that differences in resources associatedwith diverse class positions determine mobili-ty opportunities across generations

To examine this possibility I compare theclass scores obtained from the homogeneousquasi-RC(II) model with the empiricallyobtained SES ranking of classes based on edu-cation and earnings (Appendix) Figure 4 pres-

ents the comparison between the standardizedvalues of both sets of scores The correspon-dence is striking The mobility distancesbetween classes closely reproduce their dis-tances in terms of SES This close isomorphismsuggests that at least in Chile the intragener-ational inequality in the distribution of socialresources and rewards across classes largelydrives the intergenerational mobility process Totest the isomorphism further I use the SESclass scores as row and column rankings to esti-mate a quasindashlinear-by-linear association model(model 6 Table 2) The fit of model 6 is sig-nificantly worse than that of model 5 almost aspoor as that of the original core model This indi-cates that although the inter- and intragenera-tional dimensions of inequality are very similarin Chile they have a somewhat different struc-ture

In summary the central characteristic of theChilean mobility regime is the predominance oflong-range hierarchical barriers and the weak-ness of sector cleavages separating classes in themiddle and at the lower end of the hierarchicalstructure Findings from the ldquoChilean versionrdquoof the core model and the quasi-RC(II) hybridmodels are consistent depicting a social struc-ture characterized by signif icant barriersbetween the top echelon and the rest of the classstructure The good fit of the homogeneousquasi-RC(II) model provides preliminary sup-port for the resource approach linking inequal-ity and mobility Mobility dynamics seem to bedriven largely by the hierarchical distancesbetween classes and overall Chile seems not tobe less fluid than the CASMIN countries

CCOOMMPPAARRIISSOONN OOFF TTHHEE OORRIIGGIINNAALL AANNDD CCHHIILLEEAANN

VVEERRSSIIOONNSS OOFF TTHHEE CCOORREE MMOODDEELL

Although Chilean mobility dynamics seem to bedriven by inequality across classes in a veryunequal society the fitting of the core modelprovides no indication that Chile is significantlyless fluid than depicted by the core model Asa preliminary analysis of the Chilean level of flu-idity I compare the predicted propensities formobility and immobility between specific pairsof classes in Chile with those in the CASMINcountries Propensities are obtained from theoriginal and Chilean versions of the core modelrespectively (models 2 and 5 of Table 3)

443388mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I begin with the immobility of those at the topof the class structure namely the service class(I+II) The service classrsquo propensity for immo-bility in the original core model is captured bytwo parameters IN1 and IN2 This propensityis more than three times what it would be in theabsence of these effects (e43+84 = 356) InChile the propensity for immobility of class I+IIis captured by the same two parameters and itis lower than what the core model expresses286 times larger than it would be in the absenceof these effects (e34+71) What about the immo-bility of the two agricultural classes In theCASMIN countries immobility of the IVc andVIIb classes is respectively 785 and 154 timeswhat it would be in the absence of these param-eters (e43+84+101ndash22 and e43 respectively) InChile the comparable values are 422 and 286(e34+71+48ndash09 and e43+71 respectively) indi-cating the relatively low inheritance associatedwith land ownership in the Latin Americannation Thus if attention is focused on theimmobility in the upper and lower ends of thesocial hierarchy the comparison does not indi-cate less fluidity in the highly unequal Chileansociety

What about the mobility rates between thetwo most distant classes in the hierarchical rank-ing namely the service class (I+II) and farmworkers (VIIb) In the case of the CASMINcountries the mobility chances between theseclasses are captured by four parameters (HI1HI2 SE1 and AF1) and is 077 times smallerthan it would be in the absence of these effects(endash22 ndash45ndash106ndash83) In Chile the distance fromclass I to class VIIb is expressed by these samefour parameters in addition to the asymmetri-cal disaffinity AF1-C This yields mobilitychances 056 times what would be obtained inthe absence of these effects (endash09 ndash47ndash20ndash63ndash149)Thus the barrier to long-range downwardmobility from the elite is more significant inChile than in the CASMIN countries The sameoccurs with long-range downward mobility fromclass III

If we measure the reciprocal flow (ieupward mobility from class VIIb to class I+II)we obtain the same value for the CASMINcountries because all parameters are symmet-rical in the original core model For Chile how-ever this calculation removes the asymmetricalbarrier to elite downward mobility which rais-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443399

2322-ASR 703 filename70304-torche

Figure 4 Class Scores Obtained from the Homogeneous Quasi-RC(II) Model and from Socioeconomic Status(SES) Ranking

Note (1) = Class standardized socioeconomic status scores based on schooling and earnings (2) = Class stan-dardized scores based on the Homogeneous Quasi-RC(II) model (see text for details)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 3: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

ting the core model to the Chilean table I deter-mine to what extent Chile departs if it doesfrom the assumed international homogeneityin mobility regimes and what the sources of itspossible departure are Even if the core modelhas become a milestone in comparative analy-sis it has a number of undesirable propertiesdiscussed later in the Methods section Thus asa modeling alternative I use a hybrid model thatcombines association parameters assuming ahierarchical conception of the mobility structurewith patterns accounting for class immobilityInterpretation of these two models provides adetailed description of the Chilean mobilityregime and an evaluation of the main factorsmdashhierarchical differences across classes classinheritance barriers across sectors of the econ-omymdashdriving Chilean mobility dynamics

Second I compare the level of mobility inChile with that of other countries using pri-mary data from France England ScotlandIreland Sweden the United States and IsraelAll these countries have a much more egalitar-ian income distribution than Chile Thus ifmobility is associated with inequality the inter-national comparison should show that mobili-ty opportunities in Chile are signif icantlydifferent from the opportunities in these indus-trialized nations

Third my analysis moves from an interna-tional to a temporal comparison To explore theassociation between mobility and inequalityfurther I study Chilean mobility rates over timeUsing a cohort analysis I examine whether thesignificant increase in inequality associatedwith the market transformation of the 1970sand 1980s had any noticeable effect on Chileanmobility rates

Advancing some of the major results thefindings present an interesting paradox On theone hand the Chilean mobility regime is foundto be driven almost exclusively by the hierar-chical distance between classes which is deter-mined in turn by the level of inequality in thecountry This finding is consistent with the neg-ative relationship between mobility and inequal-ity posed by the ldquoresource perspectiverdquo On theother hand the international comparison indi-cates that Chile is impressively fluid despite itshigh economic inequality and the temporalanalysis shows no change in Chilean mobilityover time despite the growing inequality duringthe military regime Therefore the internation-

al and temporal comparisons seem to contradictthe negative relationship between mobility andinequality and to suggest that high inequalitymay in fact induce mobility

Solving this apparent paradox I arguerequires switching the analytical focus from thelevel to the pattern of both mobility and inequal-ity When the pattern of these two distributivephenomena is considered the contradiction dis-appears and the mechanisms driving the mobil-ityndashinequality association can be examined

The argument is organized as follows Section2 presents the empirical and theoretical evi-dence concerning the association of mobilitywith inequality Section 3 describes the Chileancontext It discusses the most salient character-istics of the Chilean socioeconomic structureand presents a brief historical description ofthe major changes in the national political econ-omy over the past few decades Section 4 intro-duces the data and analytical approach Section5 presents the comparative analysis of mobili-ty in Chile including the description of theChilean mobility regime the international com-parison of the fluidity level in Chile and theanalysis of change over time Section 6 presentsa summary of the findings the conclusionsand the implications of the Chilean case for thecomparative study of inequality

AASSSSOOCCIIAATTIIOONN BBEETTWWEEEENN IINNEEQQUUAALLIITTYYAANNDD MMOOBBIILLIITTYY EEMMPPIIRRIICCAALL AANNDDTTHHEEOORREETTIICCAALL EEVVIIDDEENNCCEE

Although mobility and inequality ldquogo togetherintuitivelyrdquo (Hout 2004969) they are differentdimensions of the social distribution of advan-tage The former refers to the cross-sectionaldimension which determines the individualposition in a social hierarchy The latter refersto its intergenerational dimension specificallyto differential access to these positions as deter-mined by the position of origin (Marshall andSwift 1999243 Marshall et al 199713)

At the conceptual level the differencesbetween these two distributional phenomenaare clear Inequality describes the distributionof resources at any particular point in timeMobility describes inequality of opportunitythe chances that someone with a particular socialorigin will attain a more rather than a less advan-taged destination regardless of the socioeco-nomic distance between these destinations This

442244mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

is why ldquoindividuals are anonymous for inequal-ity but not for mobilityrdquo (Behrman 199972) Infact mobility analysis assumes path dependencein the reproduction of the social structure acrossgenerations and studies the extent and mor-phology of such intergenerational dependence

The conceptual difference may have impor-tant practical implications As argued byFriedman (1962) and echoed more recently bysome researchers and policymakers a givenextent of income inequality under conditions ofgreat mobility and change may be less a causefor concern than the same degree of inequalityin a rigid system wherein the position of ldquopar-ticular families in the income distribution doesnot vary widely over timerdquo (p 171)

The key question is whether these two dis-tributional phenomena are independent Twotheoretical approaches address this questionBoth focus on the micro-level mechanisms thatpurportedly link these macrophenomena theindividual decision-making processes Theldquoincentive approachrdquo claims that the motiva-tion to pursue mobility is proportional to theamount of cross-sectional inequality If inequal-ity approaches zero so does the payoff formobility Conversely great inequality increas-es both the inducement to pursue mobility forthose who are initially disadvantaged and theincentive to resist mobility for those who are ini-tially advantaged In other words inequalityraises the stakes of mobility (Hout 2004970 seealso Tahlin 2004)

The ldquoresource approachrdquo contends thatmobility depends critically on resources ratherthan on incentives The higher the inequality thegreater the distance in terms of human finan-cial cultural and social resources across dif-ferent social origins In a highly unequal societythe stakes of mobility will be high across theboard but the crucial resources controlled byindividuals will be so unevenly distributed thatcompetition certainly will benefit those moreadvantaged (see Goldthorpe 2000254 Stephens197954 and Tawney 1965 for a seminal elab-oration of this argument) The outcome of highcross-sectional inequality will be the rigid repro-duction of the class structure over time ielow social fluidity

Because resources and incentives work indifferent directions the effect of inequality onmobility can be thought of as an additive cal-culation If the impact of resources outweighs

that of incentives we should expect a negativeassociation between inequality and mobilityConversely the association should be positiveif incentives are more relevant than resources inthe competition for advantaged positions Inwhat follows I argue that consideration ofresources and incentives affecting mobility at themicro-level should be preceded by a macro-level understanding of both mobility andinequality This understanding should considerthe pattern of these two distributional phenom-ena not only their overall level and shouldfocus on the location and depth of the majorcleavages in the social structure

TTHHEE EEMMPPIIRRIICCAALL EEVVIIDDEENNCCEE

Empirical evidence concerning themobilityndashinequality association is scarceAlthough high-quality data on inequality existfor a large number of countries (eg Deiningerand Squire 1996) comparable data on mobili-ty are available only for a few mostly industri-alized nations The most important source ofcomparable mobility data is the ComparativeAnalysis of Social Mobility in IndustrializedCountries (CASMIN) project which includes 15industrialized nations1 The findings of theCASMIN project are highly relevant for theassociation between mobility and inequalityThe CASMIN project concluded that whereasabsolute mobility varies across countries andover time because of differences in nationaloccupational structures relative mobilitymdashtheassociation between class of origin and desti-nation net of differences in marginal occupa-tional distributionsmdashis extremely homogeneousacross countries and over time These findingswere expressed in the ldquocommonrdquo and ldquocon-stantrdquo social fluidity hypotheses respectivelyThe basic temporal and international similari-ty was systematized into a ldquocore modelrdquo ofsocial fluidity (Erikson and Goldthorpe 1987a1987b 1992a see Featherman Jones andHauser 1975 and Hauser et al 1975a 1975b for

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442255

2322-ASR 703 filename70304-torche

1 These include 12 European countries (EnglandFrance the former Federal Republic of GermanyHungary Ireland Northern Ireland Poland ScotlandSweden the former Czechoslovakia Italy and theNetherlands) and three industrial non-European coun-tries (the United States Australia and Japan)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

the original hypotheses) Common social flu-idity is claimed to hold (at least) in industrialsocieties with a ldquomarket economy and nuclearfamilyrdquo (Featherman et al 1975340) and itwould be explained by ldquothe substantial uni-formity in the economic resources and desir-ability of occupationsrdquo (Grusky and Hauser1984)

The significant international similarity inmobility patterns has been confirmed largelyin pairwise or three-country comparisons in afew industrialized countries EriksonGoldthorpe and Portocarero (1979) Eriksonet al (1982) and Hauser (1984a 1984b) forEngland France and Sweden McRoberts andSelbec (1981) for the United States andCanada and Kerckhoff et al (1985) forEngland and the United States As more indus-trialized countries were added to the compar-ative template it was shown that Sweden andthe Netherlands are more fluid (Erikson et al1982 Ganzeboom and De Graaf 1984) andthat Germany Japan and Ireland are morerigid (Hout and Jackson 1986 Ishida 1993Muumlller 1986) However the most striking find-ing was the significant similarity across coun-tries despite the different historicalbackgrounds and institutional arrangements

On the basis of the CASMIN project a rel-ative consensus has emerged favoring thehypothesis of substantial international simi-larity in mobility patterns Similarity does notmean complete homogeneity however Therecertainly exist some national deviations butit is argued they can be explained by highlyspecific historically based institutional factorsrather than by systematic relationship to othervariable characteristics of national societiesThere seems to be only one exception to thisldquounsystematic fluctuationrdquo Comparison of theCASMIN countries found that a small but sig-nificant portion of the international variationin mobility is related to economic inequalitywith higher inequality leading to less fluidity(Erikson and Goldthorpe 1992a chapter 12also see Tyree Semyonov and Hodge 1979for an earlier analysis)

This finding supports the resource approachlinking the two distributive phenomenaAdditional evidence consistent with theresources approach was presented by Jonssonand Mills (1993) who compared Sweden andEngland and found higher fluidity in the more

equal Swedish society by Bjorklund and Jantti(1997) who compared earnings mobility inthe United States and Sweden and found mobil-ity to be higher in Sweden and by analyses ofintergenerational earnings mobility in severalOrganization for Economic Cooperation andDevelopment (OECD) countries which sug-gest higher mobility in more equal nations(Solon 2002)2

However a more recent comparative analy-sis of mobility trends between 1970 and 1990in 11 industrialized countries does not findevidence of a relationship between inequalityand fluidity (Breen and Luijkx 2004a396)This supports an earlier indication of no asso-ciation between the two distributional phe-nomena (Grusky and Hauser 1984)

In sum evidence about a potential mobili-ty-inequality association based on cross-coun-try comparisons is inconclusive not onlybecause of divergent f indings but alsobecause of the small number of countriesincluded and the potentially high collinearitybetween inequality and other explanatory fac-tors

Empirical analysis of mobility trends with-in countries has not provided a conclusiveanswer either The CASMIN finding of ldquocon-stant social fluidityrdquo over time has been recent-ly tested in the aforementioned comparativeanalysis of mobility in 11 countries Thisanalysis including 10 European nations andIsrael (Breen 2004) finds growing fluidity insome countries but null or slight temporalchange in others While Britain Israel andless conclusively Germany display ldquoconstantfluidityrdquo some indication of growing open-ness is detected in France Hungary IrelandItaly the Netherlands Norway Poland andSweden However in France Hungary Polandand Sweden all change occurred between the1970s and 1980s and stability has prevailedsince then (Breen and Luijkx 2004b54)Changes in Ireland and Italy are quite minor(Layte and Whelan 2004 Pisati andSchizzerotto 2004) and only the Netherlandsdisplays a sustained increase in mobility overthe entire period considered (Ganzeboom andLuijkx 2004) Mobility has also been found to

442266mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

2 To my knowledge there is no empirical evidencesupporting the incentive approach

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

increase over time in the United Statesbetween the early 1970s and the mid 1980s(Hout 1988) However these increases in flu-idity are not univocally preceded by a reduc-tion of inequality thus casting doubts on apotential association between inequality andmobility

A careful analysis of mobility trends inRussia shows a significant decline in mobili-ty associated with an increase in inequalityafter the market transformation of the early1990s (Gerber and Hout 2004) which pro-vides additional support for the resourceapproach The advantage of the Russian studyis that ldquothe market transition in Russia ||altered so many fundamental economic insti-tutions so rapidly that we can confidentlyascribe changes in social mobility || to thissource rather than to cultural change or indus-trializationrdquo (Gerber and Hout 2004678) It isnot clear however whether it is growth ininequality recession some other change asso-ciated with the radical liberalization of theeconomy or a combination of these factorsthat triggered a decline in fluidity in contem-porary Russia

In this context Chile presents an ideal casefor an examination of the association betweeninequality and mobility Given the extremeeconomic inequality in the country if theunequal distribution of resources or incentiveshas an impact on mobility opportunities theChilean level of fluidity should be signifi-cantly different from that of industrializednations Additionally given the increase ininequality associated with the market reformof the 1970s and 1980s analysis of mobilitybefore and after the reform would provide sup-plementary evidence concerning the potentialmobilityndashinequality relationship Naturally asingle case study will not supply a definitiveanswer to these questions However by com-bining an examination of the Chilean mobili-ty regime in the context of historicaltransformations in the country with an inter-national comparative analysis and with anassessment of mobility trends over time thisarticle provides important insights into theexistence of a link between mobility andinequality and the mechanisms driving it

TTHHEE CCHHIILLEEAANN CCOONNTTEEXXTT IINNCCOOMMPPAARRAATTIIVVEE PPEERRSSPPEECCTTIIVVEE

During the second half of the 20th centuryChile transformed from an agrarian semifeu-dal society into an urban service-based oneBetween 1950 and 2000 the rural populationdeclined from almost 40 to 17 percent (Braunet al 2000 INE 2002) This defines Chile asa mostly urban country with an 83 percentrate of urbanization larger than the 78 percentrate in the United States In tandem with urban-ization Chile experienced a reallocation ofemployment from the agricultural to the terti-ary sector of the economy The share of agri-culture in total employment declined from 38percent in 1950 to 17 percent in 2000 Whereasthe share of manufacturing remained constantat about 18 percent the share of the servicesector rose from 42 to 65 percent (Braun et al2000 INE 2002)

Urbanization and tertiarization are process-es that virtually all countries experienced dur-ing the 20th century Within this secular trendthe Chilean political economy is marked byspecific institutional developments that shapeits stratification structure From the 1940s tothe 1970s the Chilean economic landscape asthat of its Latin American neighbors wasdefined by import-substitutive industrialization(ISI) Emerging as a reaction to the collapse ofinternational trade caused by the GreatDepression (Ellworth 1945) and based on theldquodeterioration of the terms of traderdquo theory(Prebisch 1950) ISI was based on two typesof policies The first was oriented to closing theChilean economy to international markets andthe second was oriented to promoting nation-al industrialization The Chilean state becamethe leading productive agent supporting indus-try through credit investments and technicalassistance and taking a direct productive rolethrough the creation of public enterprises(Stallings 1978) After a sanguine beginningwith industrial production growing at almost7 percent per year between 1940 and 1950(Mamalakis 1976 Munoz 1968) ISI started tofail economic growth stagnated and socialturmoil resulting from massive urban migra-tion and vast social inequalities increased

In 1964 a progressive administration tookpower and adopted as its mandate the correc-tion of extreme inequality in the country by a

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442277

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ldquoRevolution in Freedomrdquo (Gazmuri 2000)This progressive government launched redis-tributive policies including an agrarian reformand an educational reform The redistributiveagenda boosted by a socialist administrationthat came to power in 1970 was abruptly halt-ed by a military coup in 1973 The militarytook power and retained it until 1990 Duringthese 17 years the military regime conduct-ed a deep market-oriented transformationNow turned into the ldquoWashington consensusrdquoparadigm (Williamson 1990) this reformincluded macroeconomic stabilization dereg-ulation of prices and markets and the priva-tization of enterprises and social services Ittransformed Chile from a closed economywith heavy state intervention into one of themost open market-based economies in theworld with the productive and welfare role ofthe state reduced to a minimum (Edwards andCox-Edwards 1991 Martinez and Diaz 1999Meller 1996 Velasco 1994)

The depth of the market reform coupledwith a world recession led in the late 1970sand early 1980s to the deepest economic cri-sis since the Great Depression A third of thelabor force was unemployed and povertyafflicted nearly half of Chilean households(Meller 1991) The post-crisis recovery start-ing in the late 1980s was substantial and sus-tained and coincided with theredemocratization of the country The grossdomestic product (GDP) per capita grew morethan 6 percent annually for 15 years a novelrate for Chile only comparable in recent timesto the ldquoEast Asian Miraclerdquo (World Bank1993) The sharp economic growth has trans-formed Chile from one of the poorest coun-tries in Latin America (Hofman 2000) into aldquomiddle-income economyrdquo (World Bank2003a) In the year 2000 the income per capi-ta was approximately US$5000 much lowerthan the United States average of US$31910but the highest in Latin America (World Bank2003a) As a consequence of the sharp eco-nomic growth Chileans have reached levels ofconsumption unthinkable two decades agoand the poverty rate fell from 45 to 21 percentbetween 1985 and 2000 (MIDEPLAN 2000Raczynski 2000)

TTHHEE CCHHIILLEEAANN LLEEVVEELL AANNDD PPAATTTTEERRNN OOFF

IINNEEQQUUAALLIITTYY

The dark side of this ldquosuccess storyrdquo is eco-nomic inequality Structurally rooted in a feu-dal agrarian structure the institutional legacyof the colonial period and in the slow expan-sion of education (Engerman and Sokoloff1997 World Bank 2003b) inequality hasremained persistently high during the 20thcentury The Gini coefficient reached 58 in the1990s which compares with a much lowerGini of 34 among the industrialized coun-tries and is large even in the highly unequalLatin American context with its average Giniof 49 (Deininger and Squire 1996 Marceland Solimano 1994) As Figure 1 indicatesChilean inequality is almost twice that in mostindustrialized countries and 15 times that inthe United States the most unequal nation ofthe industrialized world

Not only the level but also the pattern ofinequality in Chile significantly departs fromthat of the industrialized world With thewealthiest Chilean decile receiving 423 per-cent of the total national income (MIDEPLAN2001) the Chilean pattern of inequality ischaracterized by high ldquoconcentration at thetoprdquo Although inequality is by definition relat-ed to concentration the Chilean case isextreme as compared with the industrializedworld and even with other Latin Americancountries A comparison between the incomeof each decile and the income of the preced-ing (poorer) decile illustrates the point Theratio between the wealthiest and the secondwealthiest decile is twice as large in Chile asin the United States and England and one ofthe largest in Latin America depicting highelite concentration In contrast the ratiobetween the second poorest and the poorestdeciles in Chile is half that of the United Statesand England indicating that inequality at thebottom of the income distribution is muchlower in Chile than in these industrializednations (Szekely and Hilgert 1999)

In fact as Figure 2 indicates Chile is thefourth most unequal country in the mostunequal region of the world However if thewealthiest decile is excluded Chilean inequal-ity is dramatically reduced and Chile becomesthe most equal Latin American country evenmore equal than the United States (Inter-American Development Bank 1999)

442288mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442299

2322-ASR 703 filename70304-torche

Figure 1 Income Inequality in Chile and Industrialized Countries Circa 2000

Note Jap = Japan Swe = Sweden Ita = Italy Ger = Germany Neth = Netherlands Fra = France Ire = Republicof Ireland Eng = England Source World Bank 2001

Figure 2 Gini Coefficient for Total Population and Excluding Wealthiest Decile Chile other Latin AmericanCountries and Unites States in 1998

Note Par = Paraguay Bra = Brazil Ecu = Ecuador Arg = Argentina Bol = Bolivia Ven = Venezuela Uru =Uruguay Source Inter-American Development Bank 1999

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Therefore Chile is unequal largely because theelite concentrates an extremely high proportionof the national income Across the noneliteclasses the distribution of resources is muchmore uniform

Inequality is not a new development inChilean society but rather has deep historicalroots Figure 3 presents the longest availableseries on inequality in Chile depicting earningsdistribution since 1957 in Santiago3 and show-ing persistently high inequality over the lasthalf century However there is significant vari-ation over time a short decline of inequality dur-ing the progressive administrations of themid-1960s and early 1970s an increase since themilitary regime took power in 1973 a peak in1984 and a small decline after the democratictransition to levels still higher than those thatpreceded the military regime

LLAATTEE IINNDDUUSSTTRRIIAALLIIZZAATTIIOONN MMAARRKKEETTRREEFFOORRMM AANNDD TTHHEE CCHHIILLEEAANN CCLLAASSSSSSTTRRUUCCTTUURREE

The ldquocommon social fluidityrdquo hypothesis assertsthat all international variation in mobility pat-terns is attributable to highly specific histori-cally formed national characteristics In theChilean case these factors may be related to themarket reform of the 1970s and 1980s and therapid economic growth since the late 1980sNational mobility studies suggest that coun-tries experiencing rapid industrialization may becharacterized by weaker sector barriers to mobil-ity specif ically barriers between the self-employed and employees and betweenagricultural and nonagricultural classes(Goldthorpe Yaish and Kraus 1997 and Yaish2004 for Israel Ishida Goldthorpe and Erikson1991 for Japan Park 2004 for Korea and Costa-Ribeiro 2003 for Brazil)

In addition two components of the Chileanmarket transformation may have contributed tothe weakness of mobility barriers across sectorsof the economy The first component is theagrarian reform and subsequent counterreformundertaken by the military regime The secondone is the opening of the economy to interna-tional trade and the retrenchment of the statewhich led to a decline in formal employment

443300mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Figure 3 Earnings Inequality in Greatest Santiago 1957ndash2000Source Larranaga 1999 and calculations by the author

3 Santiago the Chilean capital comprises aboutone-third of the countryrsquos population Assessmentby Chilean experts suggests that Santiago trendspresent an unbiased picture of trends at the nationallevel Although earnings inequality is not as com-prehensive a measure as total income inequality it isan adequate much more accurately measured proxy(Galbraith 2002)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

and to the growth and segmentation of the self-employed sector

The main objective of the Chilean agrarianreform (1962ndash1973) was to reduce the extremeland concentration in the country The reformwas very successful In 1955 the top 7 percentof landowners held 65 percent of the land andthe bottom 37 percent held only 1 percent By1973 the last year of the reform 43 percent ofthe land had been expropriated (Kay 2002 Scott1996) An agrarian counterreform was launchedby the military regime as a way of returning theland to its former owners It did not howeverrestore the traditional hacienda order (Gomezand Echenique 1988 Rivera 1988) In fact themilitary government returned only one-third ofthe plots to their old owners sold another thirdand adjudicated the remaining third to the smallproprietors benefited by the agrarian reformUnable to compete with the now cheap foodimports many small proprietors opted to selltheir plots As a consequence as much as two-thirds of the Chilean agricultural land came upfor sale creating an active land market Themain beneficiaries of the ldquomarketization of thecountrysiderdquo were a new group of export-ori-ented entrepreneurs and international investorswho bought large tracts of land (Gwynne 1996Kay 2002) Thus the reform and counterreformtriggered a massive ldquochange of handsrdquo in theChilean countryside likely altering the patternsof land inheritance

The second major effect of the market reformon the class structure was the reduction of for-mal employment caused by the decline of theindustrial working class and by the shrinkage ofthe state Incapable of competing with the nowcheap imports the working class plummetedfrom 34 percent of total employment in theearly 1970s to 20 percent in 1980 while pub-lic employment dropped from 14 to 84 per-cent in the same period (Leon and Martinez2000 Schkolnick 2000 Velasquez 1990) Thedecline in industrial and public employmentled to an increase in self-employment fromabout 15 percent in 1970 to about 28 percent in1980 (Thomas 1996) followed by a declinethrough the economic recovery before stabi-lization at about 22 percent of the total employ-ment in 2000 (MIDEPLAN 2001) In contrastto industrialized countries in which the self-employment rate usually is within the one-digitrange and involves capital ownership in Chile

it encompasses more than one-fifth of the laborforce Furthermore self-employment hadbecome a survival strategy for a large numberof former industrial and public employeesappropriately labeled ldquoforced entrepreneurial-ismrdquo (Infante and Klein 1995 Portes Castellsand Benton 1989) As a consequence theChilean self-employed sector is voluminousand segmented It includes a small segment ofentrepreneurs oriented to capital accumulationand able to hire employees and a large seg-ment of self-employed workers mostly engagedin ldquosurvival activitiesrdquo and unable to hire thelabor of others Differences in economic well-being between these two groups are massivewith the former group earning on averagethree times more than the latter (MIDEPLAN2001)

To be sure a large self-employed sector is nota result of the market transformation Self-employment has been historically high in LatinAmerica because according to some scholarsthe region has experienced ldquodependent inte-grationrdquo into the world capitalistic system(Hopkins and Wallerstein 1982 Luxembourg1951) However the internal heterogeneity ofthe self-employed sector increased during themarket reform in a pattern that foretold whathappened in the rest of Latin America during the1990s (Klein and Tokman 2000 Portes andHoffman 2003) and what may be current devel-opments in the industrialized world (Arum andMuumlller 2004 Noorderhaven et al 2003) Thesecharacteristics of the Chilean class structureare considered in the comparative analysis ofChilean mobility

DDAATTAA AANNDD AANNAALLYYTTIICCAALL AAPPPPRROOAACCHH

This study uses the 2001 Chilean MobilitySurvey (CMS) The CMS is a nationally repre-sentative multistage stratified sample of maleheads of household ages 24 to 69 The samplingstrategy includes the following stages First 87primary sampling units (PSUs) (counties) areselected Then blocks within the PSUs areselected and finally households within blocksare chosen Counties are stratified according tosize (fewer than 20000 20000 to 100000100000 to 200000 and more than 200000inhabitants) and by geographic zone (NorthCenter South) All PSUs in the large size stra-tum are included in the sample to increase effi-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443311

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ciency The fieldwork conducted between Apriland June 2001 consists of face-to-face inter-views in the respondentrsquos household carried outby trained personnel The survey excludesnonndashhead-of-household males which represent17 percent of the male population of the rele-vant age (MIDEPLAN 2001) Among thoseexcluded 865 percent are the sons of the headof the household 124 percent are another rel-ative of the head and 11 percent are other non-related males Their occupational distributionwith control used for age is almost identical tothat of the heads of household The small pro-portion represented by this group and their sim-ilar occupational distribution suggest that theirinclusion would not significantly alter the find-ings presented in this article Excluding thehouseholds not eligible for the survey theresponse rate is 63 percent Although nonre-sponse rates usually are not reported in Chileansurveys exchange with Chilean experts indi-cates that the nonresponse rate is about 20 to 25percent for face-to-face household surveys Thehigher nonresponse rate of the CMS is likelyattributable to the difficulty contacting maleheads of household4 The total sample size is3544 I exclude individuals outside the agerange of 25 to 64 years which is conventional-ly used in comparative mobility research andcases with unusable data After this exclusionthe usable sample size is 3002

CCLLAASSSS SSCCHHEEMMAA

This study uses a class perspective to describethe Chilean social structure It utilizes the seven-category version of the classification devel-oped by the CASMIN project Thisclassification is widely used in internationalcomparative research and describes the basic

stratification of advanced industrial societiesbased on ldquoemployment relationshipsrdquo (Eriksonand Goldthorpe 1992a35ndash47 Goldthorpe andHeath 1992) The seven classes included in theschema are I+II (service class) III (routine non-manual workers) IVab (self-employed work-ers) V+VI (manual supervisors and skilledmanual workers) VIIa (unskilled manual work-ers) IVc (farmers) and VIIb (farm workers)

Use of the 7-class categorization instead ofan alternative detailed 12-class version pro-duced by the CASMIN project (Erikson andGoldthorpe 1992b) induces some loss of infor-mation about the originndashdestination associa-tion A global test of all five aggregationsobtained from the ratio of L2 tests for the inde-pendence model in the collapsed and full tablesindicates that the seven-class schema masks 29percent of the association shown by the full setof classes ([L2 full table ndash L2 collapsed table]L2 full table = [108644 ndash 77191]108644 =29) This proportion is high compared withthat of the CASMIN countries and is surpassedonly by Sweden with 31 percent (Hout andHauser 1992 Table 2) A collapsed sevenfoldclass schema was however preferred because ofthe moderate sample size of the Chilean surveyand because it grants international comparabil-ity of the findings5

Operationalization of this class classifica-tion is based on detailed information concern-ing job title (recoded into the standard ISCO-88classification [ILO 1990]) industry occupa-tional status and supervisory status of workersBecause the CMS is specifically designed toassess class mobility it includes all the infor-mation necessary to produce the CASMIN classschema The CASMIN project has not produceda standard algorithm to generate the class clas-sification but other researchers have producedsuch an algorithm which I use in this analysis(Ganzeboom and Treiman 2003) Table 1 pres-ents the basic intergenerational mobility tableto be used in this analysis including the countsand marginal percentage distribution for originand destination classes

443322mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

4 The nonresponse rate can yield bias if those whowere unreachable or refused to participate differ fromthose included in the sample To estimate the mag-nitude of this bias I compared the distribution of keyvariables with the CASEN 2000 survey CASEN isa large survey (n = 252595) conducted by the Chileangovernment and has a refusal rate lower than 10percent The comparison (available upon request)suggests that the CMS slightly underrepresents agri-cultural workers and the upper class Overall how-ever there is no indication of major nonresponsebias

5 Analyses were replicated using the detailed 12-class table and no difference in substantive find-ings was found

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

MMEETTHHOODDSS

I fit two alternative models to the Chilean mobil-ity table The first is the ldquocore model of social flu-idityrdquo which claims to capture basic internationalsimilarity in mobility Although the core modelhas become a benchmark for mobility researchand currently is the only model that provides acomparative framework for mobility analysisanalysts have drawn attention to a number of itslimitations The model has been criticized forinadequately representing hierarchical mobilityeffects for being tailored to a few industrializedcountries for the ad hoc nature of some effects itincludes (specifically the affinity effects) andfor lacking a criterion to determine how muchdivergence is necessary for it to be rejected(Ganzeboom Luijkx and Treiman 1989 Houtand Hauser 1992 Sorensen 1992 Yamaguchi1987) I therefore also test an alternative ldquohybridmodelrdquo that combines the rowndashcolumn (II) asso-ciation model (Goodman 1979) with parametersaccounting for class immobility The internationaland temporal comparison of Chilean mobilitypatterns uses the uniform-difference (UNIDIFF)model also known as the multiplicative layereffect model (Erikson and Goldthorpe 1992aXie 1992) Model comparison and selection isbased on the Bayesian Information Criterion(BIC) statistic (Raftery 1995)

AANNAALLYYSSIISS OOFF CCHHIILLEEAANN MMOOBBIILLIITTYY IINNCCOOMMPPAARRAATTIIVVEE PPEERRSSPPEECCTTIIVVEE

TTHHEE CCOORREE MMOODDEELL OOFF SSOOCCIIAALL FFLLUUIIDDIITTYY

I start by fitting the core model to the Chileanmobility table The core model represents the

basic international commonality in social flu-idity using a topological formulation In contrastto standard topological models constructed froma single allocation of cells (Hauser 1978) thecore model uses eight matrices each of whichis designed to capture a particular effect thatenhances or reduces mobility between specificclasses (Erikson and Goldthorpe 1987a 1992achapter 4) In this way the model permits thesources of commonality across nations to beidentified clearly and likewise the departuresfrom it The effects are of four types hierarchyinheritance sector and affinity6

HIERARCHY EFFECTS Hierarchy effects reflectthe impact of status distances between classeson fluidity between them To estimate theseeffects the mobility table is divided into threestrata reflecting differences in resources andrewards across classes The strata are the fol-lowing upper stratum (class I+II) middle stra-tum (classes III IVab IVc and V-VI) and lowerstratum (classes VIIa and VIIb) Because thereare three strata two hierarchy effects (HI1 andHI2) can be identified each representing thecrossing of one additional hierarchical barrierThe association between each pair of classes isexpressed as an inverse function of the numberof hierarchical strata crossed

INHERITANCE EFFECTS Inheritance effects aredesigned to capture the propensity for class

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443333

2322-ASR 703 filename70304-torche

Table 1 Seven-Class Intergenerational Mobility Table for Chilean Men

Class of Destination

I-II III IVab IVc V-VI VIIa VIIb N

I-II Service Class 229 49 81 2 37 30 2 430 14III Routine Nonmanual 61 15 34 0 24 27 1 162 5IVab Self-employed 123 37 173 28 102 90 20 573 19IVc Farmers 42 16 68 54 58 54 24 316 11V-VI Skilled Manual 91 33 116 13 155 135 36 579 19VIIa Semiskilled and Unskilled Manual 49 43 115 13 111 123 44 498 17VIIb Farm Workers 28 16 78 17 89 101 115 444 15N 623 209 665 127 576 560 242 3002 21 7 22 4 19 19 8

Note Cell entries represent counts

6 Design matrices in Erikson and Goldthorpe(1992a124ndash129)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

immobility They include three matrices IN1identifies immobility across all classes alike IN2identifies the higher immobility of the serviceclass (I+II) the self-employed group (IVab) andfarmers (IVc) and IN3 accounts for the highestimmobility of farmers (IVc)

SECTOR EFFECTS Sector effects are intended toidentify the difficulty of moving between theagricultural and nonagricultural sectors of theeconomy (SE1)

AFFINITY EFFECTS Affinity effects are intend-ed to capture specific discontinuities (negativeaffinities) or linkages (positive affinities) betweenclasses which either reinforce or offset the over-all effects of hierarchy and sector There are twoaffinity effects The first AF1 identifies the dif-ficulty of moving between the service class (I+II)and the class of farm workers (VIIb) which addsto their hierarchical distance In contrast AF2identifies instances in which mobility is morefrequent than accounted for by hierarchy and sec-tor effects and includes affinities within the non-manual sector (I+II and III) and within the manualsector (V+VI and VIIa) a symmetrical affinitybased on capital possession (IVab and IVc and Iand IVab) and an asymmetrical link between agri-cultural classes of origin (IVc and VIIb) and theunskilled manual class of destination (VIIa)7

TTHHEE MMOOBBIILLIITTYY PPAATTTTEERRNN IINN CCHHIILLEECCOONNVVEERRGGEENNCCEE OORR DDIIVVEERRGGEENNCCEE

Table 2 presents the fit of several models appliedto the Chilean data and Table 3 displays theparameter estimates for each of these modelsModel 1 in Table 2 indicates that the fit of theoriginal core model to the Chilean table is unsat-isfactory under standard statistical criteria butit explains a significant 775 percent of associ-ation under independence To assess the strengthof the different factors driving mobility oppor-tunities I compare the Chilean coefficients withthose obtained for the CASMIN countriesCoefficients from the original core model fit-ted to the Chilean table are reported in row 1 ofTable 3 CASMIN coefficients are reported inrow 2 and differences in coefficientsrsquo magni-tude and the statistical significance of the dif-ferences are reported in row 38

All Chilean coefficients are significant andhave the same sign as for the CASMIN coun-tries Comparison of the magnitude of the coef-ficients however shows significant departuresfrom the core for five effects which revealinteresting differences between Chile and theCASMIN countries The hierarchy coefficientsindicate that whereas the short-range hierar-chical barriers are somewhat weaker in Chile

443344mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

8 CASMIN coefficients were obtained from fittingthe core model to the combined table of France andEngland standardizing both tables to have 10000cases as recommended by Erikson and Goldthorpe(1992a121) The coefficients obtained are virtuallyidentical to those presented by Erikson andGoldthorpe (1992a Table 44)

Table 2 Fit statistics selected models to the Chilean Mobility Table

AssociationL2 BIC df p Explained

Core Modelmdash1 Core Model Original 1375 285 28 000 775mdash2 Core Model EHE 1081 ndash08 28 000 860mdash3 Chilean version of Core Model ([Model 2] + Chilean Parameters) 724 ndash326 27 000 906Association Models 000mdash4 Heterogeneous Quasi-RC(II) Model 734 ndash161 23 000 905mdash5 Homogeneous Quasi-RC(II) Model 803 ndash287 28 000 896mdash6 Quasi-Linear by Linear Association (SES row and column ranking)a 1563 279 33 000 798

Note EHE = empirical hierarchical effects SES = eocioeconomic statusa Linear-by-linear association model using socioeconomic status scores as presented in Appendix Figure A1 torank origin and destination classes

7 For a detailed account of the empirical derivationof the model see Erikson and Goldthorpe (1987a1987b and 1992a chapter 4)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443355

2322-ASR 703 filename70304-torche

Tab

le 3

M

odel

Par

amet

er E

stim

ates

for

the

Chi

lean

Mob

ilit

y Ta

ble

Eff

ect

1 C

ore

mod

el o

rigi

nala

mdash 2 C

ore

mod

el C

AS

MIN

b

mdash 3 D

iffe

renc

e C

ASM

INmdash

Chi

lec

mdash 4 C

ore

mod

el E

HE

d

mdash 5 C

hile

an v

ersi

on o

f co

ree

mdash Not

eD

ata

show

n ar

e m

odel

par

amet

er e

stim

ates

wit

h st

anda

rd e

rror

s in

par

enth

eses

EH

E =

em

piri

cal h

iera

rchi

cal e

ffec

ts

aM

odel

1 f

rom

Tab

le 2

b

Ori

gina

l Cor

e M

odel

fit

ted

to th

e E

ngli

sh a

nd F

renc

h m

obil

ity

tabl

es S

ee te

xt f

or d

etai

ls

cD

iffe

renc

e in

par

amet

er m

agni

tude

bet

wee

n or

igin

al c

ore

mod

el f

itte

d to

the

Chi

lean

tabl

e (r

ow 1

) an

d to

the

CA

SM

IN ta

ble

(row

2)

dM

odel

2 f

rom

Tab

le 2

e

Mod

el 3

fro

m T

able

2

dagger p

lt 1

p

lt 0

5

p

lt 0

1

p

lt 0

01 (

two-

tail

ed)

HI1

ndash12

(

05)

ndash22

(02

)ndash

10 dagger

(06

)ndash

03(

07)

ndash09

dagger(

06)

HI2

ndash65

(11

)ndash

45

(

04)

20

dagger(

12)

ndash74

(09

)ndash

47

(

10)

IN1

41

(

08)

43

(

03)

Insi

g

37

(

07)

34

(

07)

IN2

39

(

10)

84

(

04)

45

(

11)

31

(13

)mdash

IN3

68

(

24)

101

(11

)In

sig

61

(24

)4

8(

23)

SE

1

ndash38

(07

)ndash1

06

(

05)

ndash68

(08

)ndash

42

(

07)

ndash20

(

08)

AF

1

ndash76

(21

)ndash

83

(

13)

Insi

g

ndash84

(20

)ndash

63

(21

)

AF

2

33

(

05)

45

(

02)

ndash12

(

06)

20

(

06)

26

(

06)

AF

1ndashC

mdash mdash mdash mdash

ndash14

9

(46

)

IN2ndash

C

mdash mdash mdash mdash 71

(

14)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(HI1) long-range hierarchical mobility is moredifficult in this country (HI2) This suggestssubstantial inequality between the extremes ofthe class hierarchy combined with smaller dif-ferentiation in the middle of the distribution Asto the inheritance effects the service class theself-employed group and the farmers (classesI+II IVab and IVc respectively) have a sig-nificantly lower propensity to immobility inChile than in the CASMIN countries suggest-ing that class positions usually associated withindependent work are less able to reproducetheir class status intergenerationally than depict-ed by the core Most impressive the barrierbetween the agricultural and nonagriculturalsectors of the economy (SE1) is much weakerin Chile as is the affinity effect linking non-manual classes manual classes and the class-es that own capital among themselves (AF2)

Interestingly all the significant differences incoefficients with the single exception of the bar-rier to long-range hierarchical mobility sug-gest that Chile is more fluid than the core modeldepicts Higher fluidity seems to be driven bythe weakness of horizontal barriers separatingthe agricultural manual and self-employed sec-tors of the economy This is consistent withfindings of weaker sector effects in countriesthat have experienced rapid industrializationsuch as Korea (Park 2004) Brazil (Costa-Ribeiro 2003) Israel (Yaish 2004 Goldthorpeet al 1997) and Japan (Ishida et al 1991)

Additionally in the Chilean case the weak-ness of sector barriers may have been intensi-fied by the deep market transformation of the1970s and 1980s On the one hand the ldquochangeof handsrdquo in the agricultural sector induced bythe agrarian reform and counterreform likelyincreased fluidity between the agricultural andurban classes On the other hand the role of self-employment as an ephemeral refuge against(industrial and public sector) unemploymentmay have reduced barriers between the self-employed group and the rest of the social struc-ture Added to the larger difficulty of long-rangehierarchical mobility these features suggestweak sector cleavages in the Chilean fluiditypattern and high salience of the hierarchicaldimension of mobility

For the countries in which the fit of the coremodel is not adequate Erikson and Goldthorpe(1992a145) suggest adding parameters thataccount for country-specific deviations from

the model I therefore introduce adjustments toreflect salient characteristics of the Chilean flu-idity pattern producing a ldquoChilean versionrdquo ofthe core model

First I modify the hierarchical effects Fittingof the core model suggests that hierarchicalbarriers are a crucial determinant of mobility inChile This conclusion is however obscuredby the fact that the hierarchical strata distin-guished by the core model are not empiricallyobtained and may not accurately represent theChilean ranking of classes in terms of socioe-conomic status (SES) To model hierarchicaleffects in Chile adequately I rank classes usingthe unweighted average of their schooling andearning levels as a proxy for SES I then collapsethe seven classes into three hierarchical stratausing cluster analysis These three empiricallyobtained strata are the following Upper stratum(service class I+II) middle stratum (routinenonmanual and self-employed group III andIVab) and lower stratum (manual and agricul-tural classes V-VI VIIa IVc and VIIb) Detailson the ranking of classes cluster procedures andthe comparison of the CASMIN and empiricalChilean class ranking can be found in theAppendix Comparison of the CASMIN and theempirical class ranking suggests higher differ-entiation at the top than at the bottom of thesocial structure in the Chilean case

Model 2 in Table 2 uses the empiricallyobtained hierarchical strata (EHE) Theimprovement in fit is massive (the difference inBIC is 293 without using any degrees of free-dom) The model now accounts for 86 percentof the association under independence

The large improvement in fit when empiricalhierarchical strata are used challenges theassumption of an internationally homogeneoushierarchical class ranking At the empiricallevel there is no proof that the hierarchicalranking of classes assumed by the core modelfits all CASMIN countries9 Moreover someevidence points to significant cross-country dif-

443366mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

9 Erikson and Goldthorpe(1992a Table 22) pres-ent the average ranking of classes based on standardstatus measures but they do not show the empiricalranking of classes for each country If the occupa-tional composition of classes varies across countriesthere may be significant international differences inthe hierarchical ranking of classes

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ferences in the ranking of classes (Hout andHauser 1992 Sorensen 1992) At the theoreti-cal level the creators of the core model high-light the relevance of highly specific historicallyformed national attributes to explain interna-tional variation in mobility These national par-ticularities may very well result in differenthierarchical positions of classes across coun-tries10 Furthermore even if the homogeneousranking of classes imposed by the CASMINresearchers applied to most CASMIN coun-tries significant departures may be found inthe late industrialized countries currently beingadded to the comparative mobility project TheChilean findings suggest that instead of impos-ing international homogeneity to the hierarchi-cal dimension of mobility the ranking of classesshould be treated as nationally specific param-eters to be estimated as mobility researchersroutinely treat row and column marginals ofthe mobility table The issue is importantbecause the hierarchical dimension is the mostrelevant dimension of mobility (Hout andHauser 1992 Wong 1992) and inadequate mod-eling can bias assessment of not only hierar-chical effects of mobility but also other effects

Two other changes were introduced toaccount for Chilean-specific deviations fromthe core model First the IN2 effect wasrearranged to exclude the self-employed class(IVab) and to include the class of farm workers(VIIb) This rearrangement accounts for thefact that the Chilean self-employed classappeared less likely than depicted by the coremodel to reproduce its class position acrossgenerations and the inheritance level of Chileanfarm workers is higher than in the CASMINcountries These changes are expressed in theChilean-specific inheritance parameter IN2-CSecond an affinity parameter (AF1-C) wasadded to capture the difficulty of long-rangedownward elite mobility This negative asym-metrical affinity links flows from the service

class to the agricultural classes and from theroutine nonmanual class to the class of farmworkers Note that the AF1-C adds to the hier-archical distance between the two extremes ofthe class structure and to the negative affinitycaptured by AF1 to depict the extreme difficultyof long-range downward mobility from theChilean elite The fit of this lsquoChilean versionrsquoof the core model is reported in Table 2 model3 The improvement in BIC from model 2 is sub-stantial Furthermore even if the model fallsshort in terms of standard statistical criteria itaccounts for a large 91 of the associationunder independence thereby supporting theclaim that it adequately represents the Chileanmobility pattern The parameter estimates for theChilean version of the core are reported in row5 Table 3

Because of the core modelrsquos undesirable prop-erties I also use an alternative approach toexamine Chilean mobility patterns Thisapproach uses the now standard associationrowndashcolumn (II) (RC[II]) model (Goodman1979 see also Hout 1983 and Wong 1992) Inthis model both origin and destination classesare scaled so that the association can beexpressed as a linear-by-linear interaction by asingle parameter The multiplicative form ofthe model can be expressed as follows

Fij = iO j

D exp( microi νj)

where i indexes class of origin j indexes classof destination Fij is the expected frequency inthe (ij) cell is the grand mean i

O pertainsto the class of origin marginal effect j

D pertainsto the class of destination marginal effect isa global association parameter microi is the scalevalue for the ith class of origin and νj is the scalescore for the jth class of destination subject tothe following normalization constraints Σmicroi =Σνj = 0 (normalization of the location) and Σmicroi2

= Σνj2 = 1 (normalization of the scale)The class scale scores reflect a latent con-

tinuous variable made manifest by the class cat-egories Empirically obtained from the datathese ldquodistancerdquo scores produce an optimalranking of classes for the purpose of mobilityanalysis The RC(II) model is appropriate formodeling the association of any two ordinalvariables but mobility tables are distinctivebecause of the correspondence between class oforigin and class of destination (Gerber and Hout2004691)

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443377

2322-ASR 703 filename70304-torche

10 For instance during the socialist period inHungary the status of the proletariat may have beenindistinguishable from the status of the routine non-manual class (Szelenyi 199862) In Scotland givento early industrialization the hierarchical barrierbetween the skilled and unskilled working classes thatcharacterizes other CASMIN countries may not exist(Erikson and Goldthorpe 1992a163)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I introduce two adjustments to account forthis correspondence transforming the formu-lation into a ldquohybrid modelrdquo First topologicalparameters will account for the large counts insome diagonal cells reflecting immobility(inheritance) effects These parameters distin-guish two inheritance levels ldquohigh inheritancerdquofor the service class (I+II) farmers (IVc) andfarm workers (VIIb) and ldquolow inheritancerdquo forall other classes with the exception of the rou-tine nonmanual workers

Model 4 in Table 2 reports the fit statistic forthis quasi-RC(II) model The BIC statisticrejects it and prefers the more parsimoniousChilean version of the core model ThereforeI introduce a second adjustment to account forthe correspondence between rows and columnsin mobility tables Origin and destination scoresare constrained to be the same assuming thatthe scaling of classes has not significantlychanged over time Model 5 in Table 2 reportsfit statistics for this homogeneous quasi-RC(II)model On the basis of the BIC statistic the fitof the model is as good as that of the Chileanversion of the core model (Wong [1994] showsthat BIC differences of less than 5 points shouldbe considered indeterminate) Thus under stan-dard statistical criteria the Chilean core modeland the homogeneous quasi-RC(II) model areindistinguishable Given that the Chilean versionof the core model was tailored to fit the Chileantable the fit of model 5 is impressive and it indi-cates that a unidimensional scale can adequatelycapture mobility distances between classes

The logical next question is what this scalerepresents and specifically whether it repre-sents the hierarchical ranking of classes in termsof SES If a close correspondence is foundbetween the mobility distances among classesand their distances in terms of SES this find-ing would suggest that the intragenerationaldistribution of resources and rewards expressedin the SES ranking of classes drives the processof intergenerational mobility In other words thisfinding would support the resource approachrsquosclaim that differences in resources associatedwith diverse class positions determine mobili-ty opportunities across generations

To examine this possibility I compare theclass scores obtained from the homogeneousquasi-RC(II) model with the empiricallyobtained SES ranking of classes based on edu-cation and earnings (Appendix) Figure 4 pres-

ents the comparison between the standardizedvalues of both sets of scores The correspon-dence is striking The mobility distancesbetween classes closely reproduce their dis-tances in terms of SES This close isomorphismsuggests that at least in Chile the intragener-ational inequality in the distribution of socialresources and rewards across classes largelydrives the intergenerational mobility process Totest the isomorphism further I use the SESclass scores as row and column rankings to esti-mate a quasindashlinear-by-linear association model(model 6 Table 2) The fit of model 6 is sig-nificantly worse than that of model 5 almost aspoor as that of the original core model This indi-cates that although the inter- and intragenera-tional dimensions of inequality are very similarin Chile they have a somewhat different struc-ture

In summary the central characteristic of theChilean mobility regime is the predominance oflong-range hierarchical barriers and the weak-ness of sector cleavages separating classes in themiddle and at the lower end of the hierarchicalstructure Findings from the ldquoChilean versionrdquoof the core model and the quasi-RC(II) hybridmodels are consistent depicting a social struc-ture characterized by signif icant barriersbetween the top echelon and the rest of the classstructure The good fit of the homogeneousquasi-RC(II) model provides preliminary sup-port for the resource approach linking inequal-ity and mobility Mobility dynamics seem to bedriven largely by the hierarchical distancesbetween classes and overall Chile seems not tobe less fluid than the CASMIN countries

CCOOMMPPAARRIISSOONN OOFF TTHHEE OORRIIGGIINNAALL AANNDD CCHHIILLEEAANN

VVEERRSSIIOONNSS OOFF TTHHEE CCOORREE MMOODDEELL

Although Chilean mobility dynamics seem to bedriven by inequality across classes in a veryunequal society the fitting of the core modelprovides no indication that Chile is significantlyless fluid than depicted by the core model Asa preliminary analysis of the Chilean level of flu-idity I compare the predicted propensities formobility and immobility between specific pairsof classes in Chile with those in the CASMINcountries Propensities are obtained from theoriginal and Chilean versions of the core modelrespectively (models 2 and 5 of Table 3)

443388mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I begin with the immobility of those at the topof the class structure namely the service class(I+II) The service classrsquo propensity for immo-bility in the original core model is captured bytwo parameters IN1 and IN2 This propensityis more than three times what it would be in theabsence of these effects (e43+84 = 356) InChile the propensity for immobility of class I+IIis captured by the same two parameters and itis lower than what the core model expresses286 times larger than it would be in the absenceof these effects (e34+71) What about the immo-bility of the two agricultural classes In theCASMIN countries immobility of the IVc andVIIb classes is respectively 785 and 154 timeswhat it would be in the absence of these param-eters (e43+84+101ndash22 and e43 respectively) InChile the comparable values are 422 and 286(e34+71+48ndash09 and e43+71 respectively) indi-cating the relatively low inheritance associatedwith land ownership in the Latin Americannation Thus if attention is focused on theimmobility in the upper and lower ends of thesocial hierarchy the comparison does not indi-cate less fluidity in the highly unequal Chileansociety

What about the mobility rates between thetwo most distant classes in the hierarchical rank-ing namely the service class (I+II) and farmworkers (VIIb) In the case of the CASMINcountries the mobility chances between theseclasses are captured by four parameters (HI1HI2 SE1 and AF1) and is 077 times smallerthan it would be in the absence of these effects(endash22 ndash45ndash106ndash83) In Chile the distance fromclass I to class VIIb is expressed by these samefour parameters in addition to the asymmetri-cal disaffinity AF1-C This yields mobilitychances 056 times what would be obtained inthe absence of these effects (endash09 ndash47ndash20ndash63ndash149)Thus the barrier to long-range downwardmobility from the elite is more significant inChile than in the CASMIN countries The sameoccurs with long-range downward mobility fromclass III

If we measure the reciprocal flow (ieupward mobility from class VIIb to class I+II)we obtain the same value for the CASMINcountries because all parameters are symmet-rical in the original core model For Chile how-ever this calculation removes the asymmetricalbarrier to elite downward mobility which rais-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443399

2322-ASR 703 filename70304-torche

Figure 4 Class Scores Obtained from the Homogeneous Quasi-RC(II) Model and from Socioeconomic Status(SES) Ranking

Note (1) = Class standardized socioeconomic status scores based on schooling and earnings (2) = Class stan-dardized scores based on the Homogeneous Quasi-RC(II) model (see text for details)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 4: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

is why ldquoindividuals are anonymous for inequal-ity but not for mobilityrdquo (Behrman 199972) Infact mobility analysis assumes path dependencein the reproduction of the social structure acrossgenerations and studies the extent and mor-phology of such intergenerational dependence

The conceptual difference may have impor-tant practical implications As argued byFriedman (1962) and echoed more recently bysome researchers and policymakers a givenextent of income inequality under conditions ofgreat mobility and change may be less a causefor concern than the same degree of inequalityin a rigid system wherein the position of ldquopar-ticular families in the income distribution doesnot vary widely over timerdquo (p 171)

The key question is whether these two dis-tributional phenomena are independent Twotheoretical approaches address this questionBoth focus on the micro-level mechanisms thatpurportedly link these macrophenomena theindividual decision-making processes Theldquoincentive approachrdquo claims that the motiva-tion to pursue mobility is proportional to theamount of cross-sectional inequality If inequal-ity approaches zero so does the payoff formobility Conversely great inequality increas-es both the inducement to pursue mobility forthose who are initially disadvantaged and theincentive to resist mobility for those who are ini-tially advantaged In other words inequalityraises the stakes of mobility (Hout 2004970 seealso Tahlin 2004)

The ldquoresource approachrdquo contends thatmobility depends critically on resources ratherthan on incentives The higher the inequality thegreater the distance in terms of human finan-cial cultural and social resources across dif-ferent social origins In a highly unequal societythe stakes of mobility will be high across theboard but the crucial resources controlled byindividuals will be so unevenly distributed thatcompetition certainly will benefit those moreadvantaged (see Goldthorpe 2000254 Stephens197954 and Tawney 1965 for a seminal elab-oration of this argument) The outcome of highcross-sectional inequality will be the rigid repro-duction of the class structure over time ielow social fluidity

Because resources and incentives work indifferent directions the effect of inequality onmobility can be thought of as an additive cal-culation If the impact of resources outweighs

that of incentives we should expect a negativeassociation between inequality and mobilityConversely the association should be positiveif incentives are more relevant than resources inthe competition for advantaged positions Inwhat follows I argue that consideration ofresources and incentives affecting mobility at themicro-level should be preceded by a macro-level understanding of both mobility andinequality This understanding should considerthe pattern of these two distributional phenom-ena not only their overall level and shouldfocus on the location and depth of the majorcleavages in the social structure

TTHHEE EEMMPPIIRRIICCAALL EEVVIIDDEENNCCEE

Empirical evidence concerning themobilityndashinequality association is scarceAlthough high-quality data on inequality existfor a large number of countries (eg Deiningerand Squire 1996) comparable data on mobili-ty are available only for a few mostly industri-alized nations The most important source ofcomparable mobility data is the ComparativeAnalysis of Social Mobility in IndustrializedCountries (CASMIN) project which includes 15industrialized nations1 The findings of theCASMIN project are highly relevant for theassociation between mobility and inequalityThe CASMIN project concluded that whereasabsolute mobility varies across countries andover time because of differences in nationaloccupational structures relative mobilitymdashtheassociation between class of origin and desti-nation net of differences in marginal occupa-tional distributionsmdashis extremely homogeneousacross countries and over time These findingswere expressed in the ldquocommonrdquo and ldquocon-stantrdquo social fluidity hypotheses respectivelyThe basic temporal and international similari-ty was systematized into a ldquocore modelrdquo ofsocial fluidity (Erikson and Goldthorpe 1987a1987b 1992a see Featherman Jones andHauser 1975 and Hauser et al 1975a 1975b for

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442255

2322-ASR 703 filename70304-torche

1 These include 12 European countries (EnglandFrance the former Federal Republic of GermanyHungary Ireland Northern Ireland Poland ScotlandSweden the former Czechoslovakia Italy and theNetherlands) and three industrial non-European coun-tries (the United States Australia and Japan)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

the original hypotheses) Common social flu-idity is claimed to hold (at least) in industrialsocieties with a ldquomarket economy and nuclearfamilyrdquo (Featherman et al 1975340) and itwould be explained by ldquothe substantial uni-formity in the economic resources and desir-ability of occupationsrdquo (Grusky and Hauser1984)

The significant international similarity inmobility patterns has been confirmed largelyin pairwise or three-country comparisons in afew industrialized countries EriksonGoldthorpe and Portocarero (1979) Eriksonet al (1982) and Hauser (1984a 1984b) forEngland France and Sweden McRoberts andSelbec (1981) for the United States andCanada and Kerckhoff et al (1985) forEngland and the United States As more indus-trialized countries were added to the compar-ative template it was shown that Sweden andthe Netherlands are more fluid (Erikson et al1982 Ganzeboom and De Graaf 1984) andthat Germany Japan and Ireland are morerigid (Hout and Jackson 1986 Ishida 1993Muumlller 1986) However the most striking find-ing was the significant similarity across coun-tries despite the different historicalbackgrounds and institutional arrangements

On the basis of the CASMIN project a rel-ative consensus has emerged favoring thehypothesis of substantial international simi-larity in mobility patterns Similarity does notmean complete homogeneity however Therecertainly exist some national deviations butit is argued they can be explained by highlyspecific historically based institutional factorsrather than by systematic relationship to othervariable characteristics of national societiesThere seems to be only one exception to thisldquounsystematic fluctuationrdquo Comparison of theCASMIN countries found that a small but sig-nificant portion of the international variationin mobility is related to economic inequalitywith higher inequality leading to less fluidity(Erikson and Goldthorpe 1992a chapter 12also see Tyree Semyonov and Hodge 1979for an earlier analysis)

This finding supports the resource approachlinking the two distributive phenomenaAdditional evidence consistent with theresources approach was presented by Jonssonand Mills (1993) who compared Sweden andEngland and found higher fluidity in the more

equal Swedish society by Bjorklund and Jantti(1997) who compared earnings mobility inthe United States and Sweden and found mobil-ity to be higher in Sweden and by analyses ofintergenerational earnings mobility in severalOrganization for Economic Cooperation andDevelopment (OECD) countries which sug-gest higher mobility in more equal nations(Solon 2002)2

However a more recent comparative analy-sis of mobility trends between 1970 and 1990in 11 industrialized countries does not findevidence of a relationship between inequalityand fluidity (Breen and Luijkx 2004a396)This supports an earlier indication of no asso-ciation between the two distributional phe-nomena (Grusky and Hauser 1984)

In sum evidence about a potential mobili-ty-inequality association based on cross-coun-try comparisons is inconclusive not onlybecause of divergent f indings but alsobecause of the small number of countriesincluded and the potentially high collinearitybetween inequality and other explanatory fac-tors

Empirical analysis of mobility trends with-in countries has not provided a conclusiveanswer either The CASMIN finding of ldquocon-stant social fluidityrdquo over time has been recent-ly tested in the aforementioned comparativeanalysis of mobility in 11 countries Thisanalysis including 10 European nations andIsrael (Breen 2004) finds growing fluidity insome countries but null or slight temporalchange in others While Britain Israel andless conclusively Germany display ldquoconstantfluidityrdquo some indication of growing open-ness is detected in France Hungary IrelandItaly the Netherlands Norway Poland andSweden However in France Hungary Polandand Sweden all change occurred between the1970s and 1980s and stability has prevailedsince then (Breen and Luijkx 2004b54)Changes in Ireland and Italy are quite minor(Layte and Whelan 2004 Pisati andSchizzerotto 2004) and only the Netherlandsdisplays a sustained increase in mobility overthe entire period considered (Ganzeboom andLuijkx 2004) Mobility has also been found to

442266mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

2 To my knowledge there is no empirical evidencesupporting the incentive approach

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

increase over time in the United Statesbetween the early 1970s and the mid 1980s(Hout 1988) However these increases in flu-idity are not univocally preceded by a reduc-tion of inequality thus casting doubts on apotential association between inequality andmobility

A careful analysis of mobility trends inRussia shows a significant decline in mobili-ty associated with an increase in inequalityafter the market transformation of the early1990s (Gerber and Hout 2004) which pro-vides additional support for the resourceapproach The advantage of the Russian studyis that ldquothe market transition in Russia ||altered so many fundamental economic insti-tutions so rapidly that we can confidentlyascribe changes in social mobility || to thissource rather than to cultural change or indus-trializationrdquo (Gerber and Hout 2004678) It isnot clear however whether it is growth ininequality recession some other change asso-ciated with the radical liberalization of theeconomy or a combination of these factorsthat triggered a decline in fluidity in contem-porary Russia

In this context Chile presents an ideal casefor an examination of the association betweeninequality and mobility Given the extremeeconomic inequality in the country if theunequal distribution of resources or incentiveshas an impact on mobility opportunities theChilean level of fluidity should be signifi-cantly different from that of industrializednations Additionally given the increase ininequality associated with the market reformof the 1970s and 1980s analysis of mobilitybefore and after the reform would provide sup-plementary evidence concerning the potentialmobilityndashinequality relationship Naturally asingle case study will not supply a definitiveanswer to these questions However by com-bining an examination of the Chilean mobili-ty regime in the context of historicaltransformations in the country with an inter-national comparative analysis and with anassessment of mobility trends over time thisarticle provides important insights into theexistence of a link between mobility andinequality and the mechanisms driving it

TTHHEE CCHHIILLEEAANN CCOONNTTEEXXTT IINNCCOOMMPPAARRAATTIIVVEE PPEERRSSPPEECCTTIIVVEE

During the second half of the 20th centuryChile transformed from an agrarian semifeu-dal society into an urban service-based oneBetween 1950 and 2000 the rural populationdeclined from almost 40 to 17 percent (Braunet al 2000 INE 2002) This defines Chile asa mostly urban country with an 83 percentrate of urbanization larger than the 78 percentrate in the United States In tandem with urban-ization Chile experienced a reallocation ofemployment from the agricultural to the terti-ary sector of the economy The share of agri-culture in total employment declined from 38percent in 1950 to 17 percent in 2000 Whereasthe share of manufacturing remained constantat about 18 percent the share of the servicesector rose from 42 to 65 percent (Braun et al2000 INE 2002)

Urbanization and tertiarization are process-es that virtually all countries experienced dur-ing the 20th century Within this secular trendthe Chilean political economy is marked byspecific institutional developments that shapeits stratification structure From the 1940s tothe 1970s the Chilean economic landscape asthat of its Latin American neighbors wasdefined by import-substitutive industrialization(ISI) Emerging as a reaction to the collapse ofinternational trade caused by the GreatDepression (Ellworth 1945) and based on theldquodeterioration of the terms of traderdquo theory(Prebisch 1950) ISI was based on two typesof policies The first was oriented to closing theChilean economy to international markets andthe second was oriented to promoting nation-al industrialization The Chilean state becamethe leading productive agent supporting indus-try through credit investments and technicalassistance and taking a direct productive rolethrough the creation of public enterprises(Stallings 1978) After a sanguine beginningwith industrial production growing at almost7 percent per year between 1940 and 1950(Mamalakis 1976 Munoz 1968) ISI started tofail economic growth stagnated and socialturmoil resulting from massive urban migra-tion and vast social inequalities increased

In 1964 a progressive administration tookpower and adopted as its mandate the correc-tion of extreme inequality in the country by a

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442277

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ldquoRevolution in Freedomrdquo (Gazmuri 2000)This progressive government launched redis-tributive policies including an agrarian reformand an educational reform The redistributiveagenda boosted by a socialist administrationthat came to power in 1970 was abruptly halt-ed by a military coup in 1973 The militarytook power and retained it until 1990 Duringthese 17 years the military regime conduct-ed a deep market-oriented transformationNow turned into the ldquoWashington consensusrdquoparadigm (Williamson 1990) this reformincluded macroeconomic stabilization dereg-ulation of prices and markets and the priva-tization of enterprises and social services Ittransformed Chile from a closed economywith heavy state intervention into one of themost open market-based economies in theworld with the productive and welfare role ofthe state reduced to a minimum (Edwards andCox-Edwards 1991 Martinez and Diaz 1999Meller 1996 Velasco 1994)

The depth of the market reform coupledwith a world recession led in the late 1970sand early 1980s to the deepest economic cri-sis since the Great Depression A third of thelabor force was unemployed and povertyafflicted nearly half of Chilean households(Meller 1991) The post-crisis recovery start-ing in the late 1980s was substantial and sus-tained and coincided with theredemocratization of the country The grossdomestic product (GDP) per capita grew morethan 6 percent annually for 15 years a novelrate for Chile only comparable in recent timesto the ldquoEast Asian Miraclerdquo (World Bank1993) The sharp economic growth has trans-formed Chile from one of the poorest coun-tries in Latin America (Hofman 2000) into aldquomiddle-income economyrdquo (World Bank2003a) In the year 2000 the income per capi-ta was approximately US$5000 much lowerthan the United States average of US$31910but the highest in Latin America (World Bank2003a) As a consequence of the sharp eco-nomic growth Chileans have reached levels ofconsumption unthinkable two decades agoand the poverty rate fell from 45 to 21 percentbetween 1985 and 2000 (MIDEPLAN 2000Raczynski 2000)

TTHHEE CCHHIILLEEAANN LLEEVVEELL AANNDD PPAATTTTEERRNN OOFF

IINNEEQQUUAALLIITTYY

The dark side of this ldquosuccess storyrdquo is eco-nomic inequality Structurally rooted in a feu-dal agrarian structure the institutional legacyof the colonial period and in the slow expan-sion of education (Engerman and Sokoloff1997 World Bank 2003b) inequality hasremained persistently high during the 20thcentury The Gini coefficient reached 58 in the1990s which compares with a much lowerGini of 34 among the industrialized coun-tries and is large even in the highly unequalLatin American context with its average Giniof 49 (Deininger and Squire 1996 Marceland Solimano 1994) As Figure 1 indicatesChilean inequality is almost twice that in mostindustrialized countries and 15 times that inthe United States the most unequal nation ofthe industrialized world

Not only the level but also the pattern ofinequality in Chile significantly departs fromthat of the industrialized world With thewealthiest Chilean decile receiving 423 per-cent of the total national income (MIDEPLAN2001) the Chilean pattern of inequality ischaracterized by high ldquoconcentration at thetoprdquo Although inequality is by definition relat-ed to concentration the Chilean case isextreme as compared with the industrializedworld and even with other Latin Americancountries A comparison between the incomeof each decile and the income of the preced-ing (poorer) decile illustrates the point Theratio between the wealthiest and the secondwealthiest decile is twice as large in Chile asin the United States and England and one ofthe largest in Latin America depicting highelite concentration In contrast the ratiobetween the second poorest and the poorestdeciles in Chile is half that of the United Statesand England indicating that inequality at thebottom of the income distribution is muchlower in Chile than in these industrializednations (Szekely and Hilgert 1999)

In fact as Figure 2 indicates Chile is thefourth most unequal country in the mostunequal region of the world However if thewealthiest decile is excluded Chilean inequal-ity is dramatically reduced and Chile becomesthe most equal Latin American country evenmore equal than the United States (Inter-American Development Bank 1999)

442288mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442299

2322-ASR 703 filename70304-torche

Figure 1 Income Inequality in Chile and Industrialized Countries Circa 2000

Note Jap = Japan Swe = Sweden Ita = Italy Ger = Germany Neth = Netherlands Fra = France Ire = Republicof Ireland Eng = England Source World Bank 2001

Figure 2 Gini Coefficient for Total Population and Excluding Wealthiest Decile Chile other Latin AmericanCountries and Unites States in 1998

Note Par = Paraguay Bra = Brazil Ecu = Ecuador Arg = Argentina Bol = Bolivia Ven = Venezuela Uru =Uruguay Source Inter-American Development Bank 1999

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Therefore Chile is unequal largely because theelite concentrates an extremely high proportionof the national income Across the noneliteclasses the distribution of resources is muchmore uniform

Inequality is not a new development inChilean society but rather has deep historicalroots Figure 3 presents the longest availableseries on inequality in Chile depicting earningsdistribution since 1957 in Santiago3 and show-ing persistently high inequality over the lasthalf century However there is significant vari-ation over time a short decline of inequality dur-ing the progressive administrations of themid-1960s and early 1970s an increase since themilitary regime took power in 1973 a peak in1984 and a small decline after the democratictransition to levels still higher than those thatpreceded the military regime

LLAATTEE IINNDDUUSSTTRRIIAALLIIZZAATTIIOONN MMAARRKKEETTRREEFFOORRMM AANNDD TTHHEE CCHHIILLEEAANN CCLLAASSSSSSTTRRUUCCTTUURREE

The ldquocommon social fluidityrdquo hypothesis assertsthat all international variation in mobility pat-terns is attributable to highly specific histori-cally formed national characteristics In theChilean case these factors may be related to themarket reform of the 1970s and 1980s and therapid economic growth since the late 1980sNational mobility studies suggest that coun-tries experiencing rapid industrialization may becharacterized by weaker sector barriers to mobil-ity specif ically barriers between the self-employed and employees and betweenagricultural and nonagricultural classes(Goldthorpe Yaish and Kraus 1997 and Yaish2004 for Israel Ishida Goldthorpe and Erikson1991 for Japan Park 2004 for Korea and Costa-Ribeiro 2003 for Brazil)

In addition two components of the Chileanmarket transformation may have contributed tothe weakness of mobility barriers across sectorsof the economy The first component is theagrarian reform and subsequent counterreformundertaken by the military regime The secondone is the opening of the economy to interna-tional trade and the retrenchment of the statewhich led to a decline in formal employment

443300mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Figure 3 Earnings Inequality in Greatest Santiago 1957ndash2000Source Larranaga 1999 and calculations by the author

3 Santiago the Chilean capital comprises aboutone-third of the countryrsquos population Assessmentby Chilean experts suggests that Santiago trendspresent an unbiased picture of trends at the nationallevel Although earnings inequality is not as com-prehensive a measure as total income inequality it isan adequate much more accurately measured proxy(Galbraith 2002)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

and to the growth and segmentation of the self-employed sector

The main objective of the Chilean agrarianreform (1962ndash1973) was to reduce the extremeland concentration in the country The reformwas very successful In 1955 the top 7 percentof landowners held 65 percent of the land andthe bottom 37 percent held only 1 percent By1973 the last year of the reform 43 percent ofthe land had been expropriated (Kay 2002 Scott1996) An agrarian counterreform was launchedby the military regime as a way of returning theland to its former owners It did not howeverrestore the traditional hacienda order (Gomezand Echenique 1988 Rivera 1988) In fact themilitary government returned only one-third ofthe plots to their old owners sold another thirdand adjudicated the remaining third to the smallproprietors benefited by the agrarian reformUnable to compete with the now cheap foodimports many small proprietors opted to selltheir plots As a consequence as much as two-thirds of the Chilean agricultural land came upfor sale creating an active land market Themain beneficiaries of the ldquomarketization of thecountrysiderdquo were a new group of export-ori-ented entrepreneurs and international investorswho bought large tracts of land (Gwynne 1996Kay 2002) Thus the reform and counterreformtriggered a massive ldquochange of handsrdquo in theChilean countryside likely altering the patternsof land inheritance

The second major effect of the market reformon the class structure was the reduction of for-mal employment caused by the decline of theindustrial working class and by the shrinkage ofthe state Incapable of competing with the nowcheap imports the working class plummetedfrom 34 percent of total employment in theearly 1970s to 20 percent in 1980 while pub-lic employment dropped from 14 to 84 per-cent in the same period (Leon and Martinez2000 Schkolnick 2000 Velasquez 1990) Thedecline in industrial and public employmentled to an increase in self-employment fromabout 15 percent in 1970 to about 28 percent in1980 (Thomas 1996) followed by a declinethrough the economic recovery before stabi-lization at about 22 percent of the total employ-ment in 2000 (MIDEPLAN 2001) In contrastto industrialized countries in which the self-employment rate usually is within the one-digitrange and involves capital ownership in Chile

it encompasses more than one-fifth of the laborforce Furthermore self-employment hadbecome a survival strategy for a large numberof former industrial and public employeesappropriately labeled ldquoforced entrepreneurial-ismrdquo (Infante and Klein 1995 Portes Castellsand Benton 1989) As a consequence theChilean self-employed sector is voluminousand segmented It includes a small segment ofentrepreneurs oriented to capital accumulationand able to hire employees and a large seg-ment of self-employed workers mostly engagedin ldquosurvival activitiesrdquo and unable to hire thelabor of others Differences in economic well-being between these two groups are massivewith the former group earning on averagethree times more than the latter (MIDEPLAN2001)

To be sure a large self-employed sector is nota result of the market transformation Self-employment has been historically high in LatinAmerica because according to some scholarsthe region has experienced ldquodependent inte-grationrdquo into the world capitalistic system(Hopkins and Wallerstein 1982 Luxembourg1951) However the internal heterogeneity ofthe self-employed sector increased during themarket reform in a pattern that foretold whathappened in the rest of Latin America during the1990s (Klein and Tokman 2000 Portes andHoffman 2003) and what may be current devel-opments in the industrialized world (Arum andMuumlller 2004 Noorderhaven et al 2003) Thesecharacteristics of the Chilean class structureare considered in the comparative analysis ofChilean mobility

DDAATTAA AANNDD AANNAALLYYTTIICCAALL AAPPPPRROOAACCHH

This study uses the 2001 Chilean MobilitySurvey (CMS) The CMS is a nationally repre-sentative multistage stratified sample of maleheads of household ages 24 to 69 The samplingstrategy includes the following stages First 87primary sampling units (PSUs) (counties) areselected Then blocks within the PSUs areselected and finally households within blocksare chosen Counties are stratified according tosize (fewer than 20000 20000 to 100000100000 to 200000 and more than 200000inhabitants) and by geographic zone (NorthCenter South) All PSUs in the large size stra-tum are included in the sample to increase effi-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443311

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ciency The fieldwork conducted between Apriland June 2001 consists of face-to-face inter-views in the respondentrsquos household carried outby trained personnel The survey excludesnonndashhead-of-household males which represent17 percent of the male population of the rele-vant age (MIDEPLAN 2001) Among thoseexcluded 865 percent are the sons of the headof the household 124 percent are another rel-ative of the head and 11 percent are other non-related males Their occupational distributionwith control used for age is almost identical tothat of the heads of household The small pro-portion represented by this group and their sim-ilar occupational distribution suggest that theirinclusion would not significantly alter the find-ings presented in this article Excluding thehouseholds not eligible for the survey theresponse rate is 63 percent Although nonre-sponse rates usually are not reported in Chileansurveys exchange with Chilean experts indi-cates that the nonresponse rate is about 20 to 25percent for face-to-face household surveys Thehigher nonresponse rate of the CMS is likelyattributable to the difficulty contacting maleheads of household4 The total sample size is3544 I exclude individuals outside the agerange of 25 to 64 years which is conventional-ly used in comparative mobility research andcases with unusable data After this exclusionthe usable sample size is 3002

CCLLAASSSS SSCCHHEEMMAA

This study uses a class perspective to describethe Chilean social structure It utilizes the seven-category version of the classification devel-oped by the CASMIN project Thisclassification is widely used in internationalcomparative research and describes the basic

stratification of advanced industrial societiesbased on ldquoemployment relationshipsrdquo (Eriksonand Goldthorpe 1992a35ndash47 Goldthorpe andHeath 1992) The seven classes included in theschema are I+II (service class) III (routine non-manual workers) IVab (self-employed work-ers) V+VI (manual supervisors and skilledmanual workers) VIIa (unskilled manual work-ers) IVc (farmers) and VIIb (farm workers)

Use of the 7-class categorization instead ofan alternative detailed 12-class version pro-duced by the CASMIN project (Erikson andGoldthorpe 1992b) induces some loss of infor-mation about the originndashdestination associa-tion A global test of all five aggregationsobtained from the ratio of L2 tests for the inde-pendence model in the collapsed and full tablesindicates that the seven-class schema masks 29percent of the association shown by the full setof classes ([L2 full table ndash L2 collapsed table]L2 full table = [108644 ndash 77191]108644 =29) This proportion is high compared withthat of the CASMIN countries and is surpassedonly by Sweden with 31 percent (Hout andHauser 1992 Table 2) A collapsed sevenfoldclass schema was however preferred because ofthe moderate sample size of the Chilean surveyand because it grants international comparabil-ity of the findings5

Operationalization of this class classifica-tion is based on detailed information concern-ing job title (recoded into the standard ISCO-88classification [ILO 1990]) industry occupa-tional status and supervisory status of workersBecause the CMS is specifically designed toassess class mobility it includes all the infor-mation necessary to produce the CASMIN classschema The CASMIN project has not produceda standard algorithm to generate the class clas-sification but other researchers have producedsuch an algorithm which I use in this analysis(Ganzeboom and Treiman 2003) Table 1 pres-ents the basic intergenerational mobility tableto be used in this analysis including the countsand marginal percentage distribution for originand destination classes

443322mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

4 The nonresponse rate can yield bias if those whowere unreachable or refused to participate differ fromthose included in the sample To estimate the mag-nitude of this bias I compared the distribution of keyvariables with the CASEN 2000 survey CASEN isa large survey (n = 252595) conducted by the Chileangovernment and has a refusal rate lower than 10percent The comparison (available upon request)suggests that the CMS slightly underrepresents agri-cultural workers and the upper class Overall how-ever there is no indication of major nonresponsebias

5 Analyses were replicated using the detailed 12-class table and no difference in substantive find-ings was found

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

MMEETTHHOODDSS

I fit two alternative models to the Chilean mobil-ity table The first is the ldquocore model of social flu-idityrdquo which claims to capture basic internationalsimilarity in mobility Although the core modelhas become a benchmark for mobility researchand currently is the only model that provides acomparative framework for mobility analysisanalysts have drawn attention to a number of itslimitations The model has been criticized forinadequately representing hierarchical mobilityeffects for being tailored to a few industrializedcountries for the ad hoc nature of some effects itincludes (specifically the affinity effects) andfor lacking a criterion to determine how muchdivergence is necessary for it to be rejected(Ganzeboom Luijkx and Treiman 1989 Houtand Hauser 1992 Sorensen 1992 Yamaguchi1987) I therefore also test an alternative ldquohybridmodelrdquo that combines the rowndashcolumn (II) asso-ciation model (Goodman 1979) with parametersaccounting for class immobility The internationaland temporal comparison of Chilean mobilitypatterns uses the uniform-difference (UNIDIFF)model also known as the multiplicative layereffect model (Erikson and Goldthorpe 1992aXie 1992) Model comparison and selection isbased on the Bayesian Information Criterion(BIC) statistic (Raftery 1995)

AANNAALLYYSSIISS OOFF CCHHIILLEEAANN MMOOBBIILLIITTYY IINNCCOOMMPPAARRAATTIIVVEE PPEERRSSPPEECCTTIIVVEE

TTHHEE CCOORREE MMOODDEELL OOFF SSOOCCIIAALL FFLLUUIIDDIITTYY

I start by fitting the core model to the Chileanmobility table The core model represents the

basic international commonality in social flu-idity using a topological formulation In contrastto standard topological models constructed froma single allocation of cells (Hauser 1978) thecore model uses eight matrices each of whichis designed to capture a particular effect thatenhances or reduces mobility between specificclasses (Erikson and Goldthorpe 1987a 1992achapter 4) In this way the model permits thesources of commonality across nations to beidentified clearly and likewise the departuresfrom it The effects are of four types hierarchyinheritance sector and affinity6

HIERARCHY EFFECTS Hierarchy effects reflectthe impact of status distances between classeson fluidity between them To estimate theseeffects the mobility table is divided into threestrata reflecting differences in resources andrewards across classes The strata are the fol-lowing upper stratum (class I+II) middle stra-tum (classes III IVab IVc and V-VI) and lowerstratum (classes VIIa and VIIb) Because thereare three strata two hierarchy effects (HI1 andHI2) can be identified each representing thecrossing of one additional hierarchical barrierThe association between each pair of classes isexpressed as an inverse function of the numberof hierarchical strata crossed

INHERITANCE EFFECTS Inheritance effects aredesigned to capture the propensity for class

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443333

2322-ASR 703 filename70304-torche

Table 1 Seven-Class Intergenerational Mobility Table for Chilean Men

Class of Destination

I-II III IVab IVc V-VI VIIa VIIb N

I-II Service Class 229 49 81 2 37 30 2 430 14III Routine Nonmanual 61 15 34 0 24 27 1 162 5IVab Self-employed 123 37 173 28 102 90 20 573 19IVc Farmers 42 16 68 54 58 54 24 316 11V-VI Skilled Manual 91 33 116 13 155 135 36 579 19VIIa Semiskilled and Unskilled Manual 49 43 115 13 111 123 44 498 17VIIb Farm Workers 28 16 78 17 89 101 115 444 15N 623 209 665 127 576 560 242 3002 21 7 22 4 19 19 8

Note Cell entries represent counts

6 Design matrices in Erikson and Goldthorpe(1992a124ndash129)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

immobility They include three matrices IN1identifies immobility across all classes alike IN2identifies the higher immobility of the serviceclass (I+II) the self-employed group (IVab) andfarmers (IVc) and IN3 accounts for the highestimmobility of farmers (IVc)

SECTOR EFFECTS Sector effects are intended toidentify the difficulty of moving between theagricultural and nonagricultural sectors of theeconomy (SE1)

AFFINITY EFFECTS Affinity effects are intend-ed to capture specific discontinuities (negativeaffinities) or linkages (positive affinities) betweenclasses which either reinforce or offset the over-all effects of hierarchy and sector There are twoaffinity effects The first AF1 identifies the dif-ficulty of moving between the service class (I+II)and the class of farm workers (VIIb) which addsto their hierarchical distance In contrast AF2identifies instances in which mobility is morefrequent than accounted for by hierarchy and sec-tor effects and includes affinities within the non-manual sector (I+II and III) and within the manualsector (V+VI and VIIa) a symmetrical affinitybased on capital possession (IVab and IVc and Iand IVab) and an asymmetrical link between agri-cultural classes of origin (IVc and VIIb) and theunskilled manual class of destination (VIIa)7

TTHHEE MMOOBBIILLIITTYY PPAATTTTEERRNN IINN CCHHIILLEECCOONNVVEERRGGEENNCCEE OORR DDIIVVEERRGGEENNCCEE

Table 2 presents the fit of several models appliedto the Chilean data and Table 3 displays theparameter estimates for each of these modelsModel 1 in Table 2 indicates that the fit of theoriginal core model to the Chilean table is unsat-isfactory under standard statistical criteria butit explains a significant 775 percent of associ-ation under independence To assess the strengthof the different factors driving mobility oppor-tunities I compare the Chilean coefficients withthose obtained for the CASMIN countriesCoefficients from the original core model fit-ted to the Chilean table are reported in row 1 ofTable 3 CASMIN coefficients are reported inrow 2 and differences in coefficientsrsquo magni-tude and the statistical significance of the dif-ferences are reported in row 38

All Chilean coefficients are significant andhave the same sign as for the CASMIN coun-tries Comparison of the magnitude of the coef-ficients however shows significant departuresfrom the core for five effects which revealinteresting differences between Chile and theCASMIN countries The hierarchy coefficientsindicate that whereas the short-range hierar-chical barriers are somewhat weaker in Chile

443344mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

8 CASMIN coefficients were obtained from fittingthe core model to the combined table of France andEngland standardizing both tables to have 10000cases as recommended by Erikson and Goldthorpe(1992a121) The coefficients obtained are virtuallyidentical to those presented by Erikson andGoldthorpe (1992a Table 44)

Table 2 Fit statistics selected models to the Chilean Mobility Table

AssociationL2 BIC df p Explained

Core Modelmdash1 Core Model Original 1375 285 28 000 775mdash2 Core Model EHE 1081 ndash08 28 000 860mdash3 Chilean version of Core Model ([Model 2] + Chilean Parameters) 724 ndash326 27 000 906Association Models 000mdash4 Heterogeneous Quasi-RC(II) Model 734 ndash161 23 000 905mdash5 Homogeneous Quasi-RC(II) Model 803 ndash287 28 000 896mdash6 Quasi-Linear by Linear Association (SES row and column ranking)a 1563 279 33 000 798

Note EHE = empirical hierarchical effects SES = eocioeconomic statusa Linear-by-linear association model using socioeconomic status scores as presented in Appendix Figure A1 torank origin and destination classes

7 For a detailed account of the empirical derivationof the model see Erikson and Goldthorpe (1987a1987b and 1992a chapter 4)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443355

2322-ASR 703 filename70304-torche

Tab

le 3

M

odel

Par

amet

er E

stim

ates

for

the

Chi

lean

Mob

ilit

y Ta

ble

Eff

ect

1 C

ore

mod

el o

rigi

nala

mdash 2 C

ore

mod

el C

AS

MIN

b

mdash 3 D

iffe

renc

e C

ASM

INmdash

Chi

lec

mdash 4 C

ore

mod

el E

HE

d

mdash 5 C

hile

an v

ersi

on o

f co

ree

mdash Not

eD

ata

show

n ar

e m

odel

par

amet

er e

stim

ates

wit

h st

anda

rd e

rror

s in

par

enth

eses

EH

E =

em

piri

cal h

iera

rchi

cal e

ffec

ts

aM

odel

1 f

rom

Tab

le 2

b

Ori

gina

l Cor

e M

odel

fit

ted

to th

e E

ngli

sh a

nd F

renc

h m

obil

ity

tabl

es S

ee te

xt f

or d

etai

ls

cD

iffe

renc

e in

par

amet

er m

agni

tude

bet

wee

n or

igin

al c

ore

mod

el f

itte

d to

the

Chi

lean

tabl

e (r

ow 1

) an

d to

the

CA

SM

IN ta

ble

(row

2)

dM

odel

2 f

rom

Tab

le 2

e

Mod

el 3

fro

m T

able

2

dagger p

lt 1

p

lt 0

5

p

lt 0

1

p

lt 0

01 (

two-

tail

ed)

HI1

ndash12

(

05)

ndash22

(02

)ndash

10 dagger

(06

)ndash

03(

07)

ndash09

dagger(

06)

HI2

ndash65

(11

)ndash

45

(

04)

20

dagger(

12)

ndash74

(09

)ndash

47

(

10)

IN1

41

(

08)

43

(

03)

Insi

g

37

(

07)

34

(

07)

IN2

39

(

10)

84

(

04)

45

(

11)

31

(13

)mdash

IN3

68

(

24)

101

(11

)In

sig

61

(24

)4

8(

23)

SE

1

ndash38

(07

)ndash1

06

(

05)

ndash68

(08

)ndash

42

(

07)

ndash20

(

08)

AF

1

ndash76

(21

)ndash

83

(

13)

Insi

g

ndash84

(20

)ndash

63

(21

)

AF

2

33

(

05)

45

(

02)

ndash12

(

06)

20

(

06)

26

(

06)

AF

1ndashC

mdash mdash mdash mdash

ndash14

9

(46

)

IN2ndash

C

mdash mdash mdash mdash 71

(

14)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(HI1) long-range hierarchical mobility is moredifficult in this country (HI2) This suggestssubstantial inequality between the extremes ofthe class hierarchy combined with smaller dif-ferentiation in the middle of the distribution Asto the inheritance effects the service class theself-employed group and the farmers (classesI+II IVab and IVc respectively) have a sig-nificantly lower propensity to immobility inChile than in the CASMIN countries suggest-ing that class positions usually associated withindependent work are less able to reproducetheir class status intergenerationally than depict-ed by the core Most impressive the barrierbetween the agricultural and nonagriculturalsectors of the economy (SE1) is much weakerin Chile as is the affinity effect linking non-manual classes manual classes and the class-es that own capital among themselves (AF2)

Interestingly all the significant differences incoefficients with the single exception of the bar-rier to long-range hierarchical mobility sug-gest that Chile is more fluid than the core modeldepicts Higher fluidity seems to be driven bythe weakness of horizontal barriers separatingthe agricultural manual and self-employed sec-tors of the economy This is consistent withfindings of weaker sector effects in countriesthat have experienced rapid industrializationsuch as Korea (Park 2004) Brazil (Costa-Ribeiro 2003) Israel (Yaish 2004 Goldthorpeet al 1997) and Japan (Ishida et al 1991)

Additionally in the Chilean case the weak-ness of sector barriers may have been intensi-fied by the deep market transformation of the1970s and 1980s On the one hand the ldquochangeof handsrdquo in the agricultural sector induced bythe agrarian reform and counterreform likelyincreased fluidity between the agricultural andurban classes On the other hand the role of self-employment as an ephemeral refuge against(industrial and public sector) unemploymentmay have reduced barriers between the self-employed group and the rest of the social struc-ture Added to the larger difficulty of long-rangehierarchical mobility these features suggestweak sector cleavages in the Chilean fluiditypattern and high salience of the hierarchicaldimension of mobility

For the countries in which the fit of the coremodel is not adequate Erikson and Goldthorpe(1992a145) suggest adding parameters thataccount for country-specific deviations from

the model I therefore introduce adjustments toreflect salient characteristics of the Chilean flu-idity pattern producing a ldquoChilean versionrdquo ofthe core model

First I modify the hierarchical effects Fittingof the core model suggests that hierarchicalbarriers are a crucial determinant of mobility inChile This conclusion is however obscuredby the fact that the hierarchical strata distin-guished by the core model are not empiricallyobtained and may not accurately represent theChilean ranking of classes in terms of socioe-conomic status (SES) To model hierarchicaleffects in Chile adequately I rank classes usingthe unweighted average of their schooling andearning levels as a proxy for SES I then collapsethe seven classes into three hierarchical stratausing cluster analysis These three empiricallyobtained strata are the following Upper stratum(service class I+II) middle stratum (routinenonmanual and self-employed group III andIVab) and lower stratum (manual and agricul-tural classes V-VI VIIa IVc and VIIb) Detailson the ranking of classes cluster procedures andthe comparison of the CASMIN and empiricalChilean class ranking can be found in theAppendix Comparison of the CASMIN and theempirical class ranking suggests higher differ-entiation at the top than at the bottom of thesocial structure in the Chilean case

Model 2 in Table 2 uses the empiricallyobtained hierarchical strata (EHE) Theimprovement in fit is massive (the difference inBIC is 293 without using any degrees of free-dom) The model now accounts for 86 percentof the association under independence

The large improvement in fit when empiricalhierarchical strata are used challenges theassumption of an internationally homogeneoushierarchical class ranking At the empiricallevel there is no proof that the hierarchicalranking of classes assumed by the core modelfits all CASMIN countries9 Moreover someevidence points to significant cross-country dif-

443366mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

9 Erikson and Goldthorpe(1992a Table 22) pres-ent the average ranking of classes based on standardstatus measures but they do not show the empiricalranking of classes for each country If the occupa-tional composition of classes varies across countriesthere may be significant international differences inthe hierarchical ranking of classes

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ferences in the ranking of classes (Hout andHauser 1992 Sorensen 1992) At the theoreti-cal level the creators of the core model high-light the relevance of highly specific historicallyformed national attributes to explain interna-tional variation in mobility These national par-ticularities may very well result in differenthierarchical positions of classes across coun-tries10 Furthermore even if the homogeneousranking of classes imposed by the CASMINresearchers applied to most CASMIN coun-tries significant departures may be found inthe late industrialized countries currently beingadded to the comparative mobility project TheChilean findings suggest that instead of impos-ing international homogeneity to the hierarchi-cal dimension of mobility the ranking of classesshould be treated as nationally specific param-eters to be estimated as mobility researchersroutinely treat row and column marginals ofthe mobility table The issue is importantbecause the hierarchical dimension is the mostrelevant dimension of mobility (Hout andHauser 1992 Wong 1992) and inadequate mod-eling can bias assessment of not only hierar-chical effects of mobility but also other effects

Two other changes were introduced toaccount for Chilean-specific deviations fromthe core model First the IN2 effect wasrearranged to exclude the self-employed class(IVab) and to include the class of farm workers(VIIb) This rearrangement accounts for thefact that the Chilean self-employed classappeared less likely than depicted by the coremodel to reproduce its class position acrossgenerations and the inheritance level of Chileanfarm workers is higher than in the CASMINcountries These changes are expressed in theChilean-specific inheritance parameter IN2-CSecond an affinity parameter (AF1-C) wasadded to capture the difficulty of long-rangedownward elite mobility This negative asym-metrical affinity links flows from the service

class to the agricultural classes and from theroutine nonmanual class to the class of farmworkers Note that the AF1-C adds to the hier-archical distance between the two extremes ofthe class structure and to the negative affinitycaptured by AF1 to depict the extreme difficultyof long-range downward mobility from theChilean elite The fit of this lsquoChilean versionrsquoof the core model is reported in Table 2 model3 The improvement in BIC from model 2 is sub-stantial Furthermore even if the model fallsshort in terms of standard statistical criteria itaccounts for a large 91 of the associationunder independence thereby supporting theclaim that it adequately represents the Chileanmobility pattern The parameter estimates for theChilean version of the core are reported in row5 Table 3

Because of the core modelrsquos undesirable prop-erties I also use an alternative approach toexamine Chilean mobility patterns Thisapproach uses the now standard associationrowndashcolumn (II) (RC[II]) model (Goodman1979 see also Hout 1983 and Wong 1992) Inthis model both origin and destination classesare scaled so that the association can beexpressed as a linear-by-linear interaction by asingle parameter The multiplicative form ofthe model can be expressed as follows

Fij = iO j

D exp( microi νj)

where i indexes class of origin j indexes classof destination Fij is the expected frequency inthe (ij) cell is the grand mean i

O pertainsto the class of origin marginal effect j

D pertainsto the class of destination marginal effect isa global association parameter microi is the scalevalue for the ith class of origin and νj is the scalescore for the jth class of destination subject tothe following normalization constraints Σmicroi =Σνj = 0 (normalization of the location) and Σmicroi2

= Σνj2 = 1 (normalization of the scale)The class scale scores reflect a latent con-

tinuous variable made manifest by the class cat-egories Empirically obtained from the datathese ldquodistancerdquo scores produce an optimalranking of classes for the purpose of mobilityanalysis The RC(II) model is appropriate formodeling the association of any two ordinalvariables but mobility tables are distinctivebecause of the correspondence between class oforigin and class of destination (Gerber and Hout2004691)

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443377

2322-ASR 703 filename70304-torche

10 For instance during the socialist period inHungary the status of the proletariat may have beenindistinguishable from the status of the routine non-manual class (Szelenyi 199862) In Scotland givento early industrialization the hierarchical barrierbetween the skilled and unskilled working classes thatcharacterizes other CASMIN countries may not exist(Erikson and Goldthorpe 1992a163)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I introduce two adjustments to account forthis correspondence transforming the formu-lation into a ldquohybrid modelrdquo First topologicalparameters will account for the large counts insome diagonal cells reflecting immobility(inheritance) effects These parameters distin-guish two inheritance levels ldquohigh inheritancerdquofor the service class (I+II) farmers (IVc) andfarm workers (VIIb) and ldquolow inheritancerdquo forall other classes with the exception of the rou-tine nonmanual workers

Model 4 in Table 2 reports the fit statistic forthis quasi-RC(II) model The BIC statisticrejects it and prefers the more parsimoniousChilean version of the core model ThereforeI introduce a second adjustment to account forthe correspondence between rows and columnsin mobility tables Origin and destination scoresare constrained to be the same assuming thatthe scaling of classes has not significantlychanged over time Model 5 in Table 2 reportsfit statistics for this homogeneous quasi-RC(II)model On the basis of the BIC statistic the fitof the model is as good as that of the Chileanversion of the core model (Wong [1994] showsthat BIC differences of less than 5 points shouldbe considered indeterminate) Thus under stan-dard statistical criteria the Chilean core modeland the homogeneous quasi-RC(II) model areindistinguishable Given that the Chilean versionof the core model was tailored to fit the Chileantable the fit of model 5 is impressive and it indi-cates that a unidimensional scale can adequatelycapture mobility distances between classes

The logical next question is what this scalerepresents and specifically whether it repre-sents the hierarchical ranking of classes in termsof SES If a close correspondence is foundbetween the mobility distances among classesand their distances in terms of SES this find-ing would suggest that the intragenerationaldistribution of resources and rewards expressedin the SES ranking of classes drives the processof intergenerational mobility In other words thisfinding would support the resource approachrsquosclaim that differences in resources associatedwith diverse class positions determine mobili-ty opportunities across generations

To examine this possibility I compare theclass scores obtained from the homogeneousquasi-RC(II) model with the empiricallyobtained SES ranking of classes based on edu-cation and earnings (Appendix) Figure 4 pres-

ents the comparison between the standardizedvalues of both sets of scores The correspon-dence is striking The mobility distancesbetween classes closely reproduce their dis-tances in terms of SES This close isomorphismsuggests that at least in Chile the intragener-ational inequality in the distribution of socialresources and rewards across classes largelydrives the intergenerational mobility process Totest the isomorphism further I use the SESclass scores as row and column rankings to esti-mate a quasindashlinear-by-linear association model(model 6 Table 2) The fit of model 6 is sig-nificantly worse than that of model 5 almost aspoor as that of the original core model This indi-cates that although the inter- and intragenera-tional dimensions of inequality are very similarin Chile they have a somewhat different struc-ture

In summary the central characteristic of theChilean mobility regime is the predominance oflong-range hierarchical barriers and the weak-ness of sector cleavages separating classes in themiddle and at the lower end of the hierarchicalstructure Findings from the ldquoChilean versionrdquoof the core model and the quasi-RC(II) hybridmodels are consistent depicting a social struc-ture characterized by signif icant barriersbetween the top echelon and the rest of the classstructure The good fit of the homogeneousquasi-RC(II) model provides preliminary sup-port for the resource approach linking inequal-ity and mobility Mobility dynamics seem to bedriven largely by the hierarchical distancesbetween classes and overall Chile seems not tobe less fluid than the CASMIN countries

CCOOMMPPAARRIISSOONN OOFF TTHHEE OORRIIGGIINNAALL AANNDD CCHHIILLEEAANN

VVEERRSSIIOONNSS OOFF TTHHEE CCOORREE MMOODDEELL

Although Chilean mobility dynamics seem to bedriven by inequality across classes in a veryunequal society the fitting of the core modelprovides no indication that Chile is significantlyless fluid than depicted by the core model Asa preliminary analysis of the Chilean level of flu-idity I compare the predicted propensities formobility and immobility between specific pairsof classes in Chile with those in the CASMINcountries Propensities are obtained from theoriginal and Chilean versions of the core modelrespectively (models 2 and 5 of Table 3)

443388mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I begin with the immobility of those at the topof the class structure namely the service class(I+II) The service classrsquo propensity for immo-bility in the original core model is captured bytwo parameters IN1 and IN2 This propensityis more than three times what it would be in theabsence of these effects (e43+84 = 356) InChile the propensity for immobility of class I+IIis captured by the same two parameters and itis lower than what the core model expresses286 times larger than it would be in the absenceof these effects (e34+71) What about the immo-bility of the two agricultural classes In theCASMIN countries immobility of the IVc andVIIb classes is respectively 785 and 154 timeswhat it would be in the absence of these param-eters (e43+84+101ndash22 and e43 respectively) InChile the comparable values are 422 and 286(e34+71+48ndash09 and e43+71 respectively) indi-cating the relatively low inheritance associatedwith land ownership in the Latin Americannation Thus if attention is focused on theimmobility in the upper and lower ends of thesocial hierarchy the comparison does not indi-cate less fluidity in the highly unequal Chileansociety

What about the mobility rates between thetwo most distant classes in the hierarchical rank-ing namely the service class (I+II) and farmworkers (VIIb) In the case of the CASMINcountries the mobility chances between theseclasses are captured by four parameters (HI1HI2 SE1 and AF1) and is 077 times smallerthan it would be in the absence of these effects(endash22 ndash45ndash106ndash83) In Chile the distance fromclass I to class VIIb is expressed by these samefour parameters in addition to the asymmetri-cal disaffinity AF1-C This yields mobilitychances 056 times what would be obtained inthe absence of these effects (endash09 ndash47ndash20ndash63ndash149)Thus the barrier to long-range downwardmobility from the elite is more significant inChile than in the CASMIN countries The sameoccurs with long-range downward mobility fromclass III

If we measure the reciprocal flow (ieupward mobility from class VIIb to class I+II)we obtain the same value for the CASMINcountries because all parameters are symmet-rical in the original core model For Chile how-ever this calculation removes the asymmetricalbarrier to elite downward mobility which rais-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443399

2322-ASR 703 filename70304-torche

Figure 4 Class Scores Obtained from the Homogeneous Quasi-RC(II) Model and from Socioeconomic Status(SES) Ranking

Note (1) = Class standardized socioeconomic status scores based on schooling and earnings (2) = Class stan-dardized scores based on the Homogeneous Quasi-RC(II) model (see text for details)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 5: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

the original hypotheses) Common social flu-idity is claimed to hold (at least) in industrialsocieties with a ldquomarket economy and nuclearfamilyrdquo (Featherman et al 1975340) and itwould be explained by ldquothe substantial uni-formity in the economic resources and desir-ability of occupationsrdquo (Grusky and Hauser1984)

The significant international similarity inmobility patterns has been confirmed largelyin pairwise or three-country comparisons in afew industrialized countries EriksonGoldthorpe and Portocarero (1979) Eriksonet al (1982) and Hauser (1984a 1984b) forEngland France and Sweden McRoberts andSelbec (1981) for the United States andCanada and Kerckhoff et al (1985) forEngland and the United States As more indus-trialized countries were added to the compar-ative template it was shown that Sweden andthe Netherlands are more fluid (Erikson et al1982 Ganzeboom and De Graaf 1984) andthat Germany Japan and Ireland are morerigid (Hout and Jackson 1986 Ishida 1993Muumlller 1986) However the most striking find-ing was the significant similarity across coun-tries despite the different historicalbackgrounds and institutional arrangements

On the basis of the CASMIN project a rel-ative consensus has emerged favoring thehypothesis of substantial international simi-larity in mobility patterns Similarity does notmean complete homogeneity however Therecertainly exist some national deviations butit is argued they can be explained by highlyspecific historically based institutional factorsrather than by systematic relationship to othervariable characteristics of national societiesThere seems to be only one exception to thisldquounsystematic fluctuationrdquo Comparison of theCASMIN countries found that a small but sig-nificant portion of the international variationin mobility is related to economic inequalitywith higher inequality leading to less fluidity(Erikson and Goldthorpe 1992a chapter 12also see Tyree Semyonov and Hodge 1979for an earlier analysis)

This finding supports the resource approachlinking the two distributive phenomenaAdditional evidence consistent with theresources approach was presented by Jonssonand Mills (1993) who compared Sweden andEngland and found higher fluidity in the more

equal Swedish society by Bjorklund and Jantti(1997) who compared earnings mobility inthe United States and Sweden and found mobil-ity to be higher in Sweden and by analyses ofintergenerational earnings mobility in severalOrganization for Economic Cooperation andDevelopment (OECD) countries which sug-gest higher mobility in more equal nations(Solon 2002)2

However a more recent comparative analy-sis of mobility trends between 1970 and 1990in 11 industrialized countries does not findevidence of a relationship between inequalityand fluidity (Breen and Luijkx 2004a396)This supports an earlier indication of no asso-ciation between the two distributional phe-nomena (Grusky and Hauser 1984)

In sum evidence about a potential mobili-ty-inequality association based on cross-coun-try comparisons is inconclusive not onlybecause of divergent f indings but alsobecause of the small number of countriesincluded and the potentially high collinearitybetween inequality and other explanatory fac-tors

Empirical analysis of mobility trends with-in countries has not provided a conclusiveanswer either The CASMIN finding of ldquocon-stant social fluidityrdquo over time has been recent-ly tested in the aforementioned comparativeanalysis of mobility in 11 countries Thisanalysis including 10 European nations andIsrael (Breen 2004) finds growing fluidity insome countries but null or slight temporalchange in others While Britain Israel andless conclusively Germany display ldquoconstantfluidityrdquo some indication of growing open-ness is detected in France Hungary IrelandItaly the Netherlands Norway Poland andSweden However in France Hungary Polandand Sweden all change occurred between the1970s and 1980s and stability has prevailedsince then (Breen and Luijkx 2004b54)Changes in Ireland and Italy are quite minor(Layte and Whelan 2004 Pisati andSchizzerotto 2004) and only the Netherlandsdisplays a sustained increase in mobility overthe entire period considered (Ganzeboom andLuijkx 2004) Mobility has also been found to

442266mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

2 To my knowledge there is no empirical evidencesupporting the incentive approach

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

increase over time in the United Statesbetween the early 1970s and the mid 1980s(Hout 1988) However these increases in flu-idity are not univocally preceded by a reduc-tion of inequality thus casting doubts on apotential association between inequality andmobility

A careful analysis of mobility trends inRussia shows a significant decline in mobili-ty associated with an increase in inequalityafter the market transformation of the early1990s (Gerber and Hout 2004) which pro-vides additional support for the resourceapproach The advantage of the Russian studyis that ldquothe market transition in Russia ||altered so many fundamental economic insti-tutions so rapidly that we can confidentlyascribe changes in social mobility || to thissource rather than to cultural change or indus-trializationrdquo (Gerber and Hout 2004678) It isnot clear however whether it is growth ininequality recession some other change asso-ciated with the radical liberalization of theeconomy or a combination of these factorsthat triggered a decline in fluidity in contem-porary Russia

In this context Chile presents an ideal casefor an examination of the association betweeninequality and mobility Given the extremeeconomic inequality in the country if theunequal distribution of resources or incentiveshas an impact on mobility opportunities theChilean level of fluidity should be signifi-cantly different from that of industrializednations Additionally given the increase ininequality associated with the market reformof the 1970s and 1980s analysis of mobilitybefore and after the reform would provide sup-plementary evidence concerning the potentialmobilityndashinequality relationship Naturally asingle case study will not supply a definitiveanswer to these questions However by com-bining an examination of the Chilean mobili-ty regime in the context of historicaltransformations in the country with an inter-national comparative analysis and with anassessment of mobility trends over time thisarticle provides important insights into theexistence of a link between mobility andinequality and the mechanisms driving it

TTHHEE CCHHIILLEEAANN CCOONNTTEEXXTT IINNCCOOMMPPAARRAATTIIVVEE PPEERRSSPPEECCTTIIVVEE

During the second half of the 20th centuryChile transformed from an agrarian semifeu-dal society into an urban service-based oneBetween 1950 and 2000 the rural populationdeclined from almost 40 to 17 percent (Braunet al 2000 INE 2002) This defines Chile asa mostly urban country with an 83 percentrate of urbanization larger than the 78 percentrate in the United States In tandem with urban-ization Chile experienced a reallocation ofemployment from the agricultural to the terti-ary sector of the economy The share of agri-culture in total employment declined from 38percent in 1950 to 17 percent in 2000 Whereasthe share of manufacturing remained constantat about 18 percent the share of the servicesector rose from 42 to 65 percent (Braun et al2000 INE 2002)

Urbanization and tertiarization are process-es that virtually all countries experienced dur-ing the 20th century Within this secular trendthe Chilean political economy is marked byspecific institutional developments that shapeits stratification structure From the 1940s tothe 1970s the Chilean economic landscape asthat of its Latin American neighbors wasdefined by import-substitutive industrialization(ISI) Emerging as a reaction to the collapse ofinternational trade caused by the GreatDepression (Ellworth 1945) and based on theldquodeterioration of the terms of traderdquo theory(Prebisch 1950) ISI was based on two typesof policies The first was oriented to closing theChilean economy to international markets andthe second was oriented to promoting nation-al industrialization The Chilean state becamethe leading productive agent supporting indus-try through credit investments and technicalassistance and taking a direct productive rolethrough the creation of public enterprises(Stallings 1978) After a sanguine beginningwith industrial production growing at almost7 percent per year between 1940 and 1950(Mamalakis 1976 Munoz 1968) ISI started tofail economic growth stagnated and socialturmoil resulting from massive urban migra-tion and vast social inequalities increased

In 1964 a progressive administration tookpower and adopted as its mandate the correc-tion of extreme inequality in the country by a

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442277

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ldquoRevolution in Freedomrdquo (Gazmuri 2000)This progressive government launched redis-tributive policies including an agrarian reformand an educational reform The redistributiveagenda boosted by a socialist administrationthat came to power in 1970 was abruptly halt-ed by a military coup in 1973 The militarytook power and retained it until 1990 Duringthese 17 years the military regime conduct-ed a deep market-oriented transformationNow turned into the ldquoWashington consensusrdquoparadigm (Williamson 1990) this reformincluded macroeconomic stabilization dereg-ulation of prices and markets and the priva-tization of enterprises and social services Ittransformed Chile from a closed economywith heavy state intervention into one of themost open market-based economies in theworld with the productive and welfare role ofthe state reduced to a minimum (Edwards andCox-Edwards 1991 Martinez and Diaz 1999Meller 1996 Velasco 1994)

The depth of the market reform coupledwith a world recession led in the late 1970sand early 1980s to the deepest economic cri-sis since the Great Depression A third of thelabor force was unemployed and povertyafflicted nearly half of Chilean households(Meller 1991) The post-crisis recovery start-ing in the late 1980s was substantial and sus-tained and coincided with theredemocratization of the country The grossdomestic product (GDP) per capita grew morethan 6 percent annually for 15 years a novelrate for Chile only comparable in recent timesto the ldquoEast Asian Miraclerdquo (World Bank1993) The sharp economic growth has trans-formed Chile from one of the poorest coun-tries in Latin America (Hofman 2000) into aldquomiddle-income economyrdquo (World Bank2003a) In the year 2000 the income per capi-ta was approximately US$5000 much lowerthan the United States average of US$31910but the highest in Latin America (World Bank2003a) As a consequence of the sharp eco-nomic growth Chileans have reached levels ofconsumption unthinkable two decades agoand the poverty rate fell from 45 to 21 percentbetween 1985 and 2000 (MIDEPLAN 2000Raczynski 2000)

TTHHEE CCHHIILLEEAANN LLEEVVEELL AANNDD PPAATTTTEERRNN OOFF

IINNEEQQUUAALLIITTYY

The dark side of this ldquosuccess storyrdquo is eco-nomic inequality Structurally rooted in a feu-dal agrarian structure the institutional legacyof the colonial period and in the slow expan-sion of education (Engerman and Sokoloff1997 World Bank 2003b) inequality hasremained persistently high during the 20thcentury The Gini coefficient reached 58 in the1990s which compares with a much lowerGini of 34 among the industrialized coun-tries and is large even in the highly unequalLatin American context with its average Giniof 49 (Deininger and Squire 1996 Marceland Solimano 1994) As Figure 1 indicatesChilean inequality is almost twice that in mostindustrialized countries and 15 times that inthe United States the most unequal nation ofthe industrialized world

Not only the level but also the pattern ofinequality in Chile significantly departs fromthat of the industrialized world With thewealthiest Chilean decile receiving 423 per-cent of the total national income (MIDEPLAN2001) the Chilean pattern of inequality ischaracterized by high ldquoconcentration at thetoprdquo Although inequality is by definition relat-ed to concentration the Chilean case isextreme as compared with the industrializedworld and even with other Latin Americancountries A comparison between the incomeof each decile and the income of the preced-ing (poorer) decile illustrates the point Theratio between the wealthiest and the secondwealthiest decile is twice as large in Chile asin the United States and England and one ofthe largest in Latin America depicting highelite concentration In contrast the ratiobetween the second poorest and the poorestdeciles in Chile is half that of the United Statesand England indicating that inequality at thebottom of the income distribution is muchlower in Chile than in these industrializednations (Szekely and Hilgert 1999)

In fact as Figure 2 indicates Chile is thefourth most unequal country in the mostunequal region of the world However if thewealthiest decile is excluded Chilean inequal-ity is dramatically reduced and Chile becomesthe most equal Latin American country evenmore equal than the United States (Inter-American Development Bank 1999)

442288mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442299

2322-ASR 703 filename70304-torche

Figure 1 Income Inequality in Chile and Industrialized Countries Circa 2000

Note Jap = Japan Swe = Sweden Ita = Italy Ger = Germany Neth = Netherlands Fra = France Ire = Republicof Ireland Eng = England Source World Bank 2001

Figure 2 Gini Coefficient for Total Population and Excluding Wealthiest Decile Chile other Latin AmericanCountries and Unites States in 1998

Note Par = Paraguay Bra = Brazil Ecu = Ecuador Arg = Argentina Bol = Bolivia Ven = Venezuela Uru =Uruguay Source Inter-American Development Bank 1999

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Therefore Chile is unequal largely because theelite concentrates an extremely high proportionof the national income Across the noneliteclasses the distribution of resources is muchmore uniform

Inequality is not a new development inChilean society but rather has deep historicalroots Figure 3 presents the longest availableseries on inequality in Chile depicting earningsdistribution since 1957 in Santiago3 and show-ing persistently high inequality over the lasthalf century However there is significant vari-ation over time a short decline of inequality dur-ing the progressive administrations of themid-1960s and early 1970s an increase since themilitary regime took power in 1973 a peak in1984 and a small decline after the democratictransition to levels still higher than those thatpreceded the military regime

LLAATTEE IINNDDUUSSTTRRIIAALLIIZZAATTIIOONN MMAARRKKEETTRREEFFOORRMM AANNDD TTHHEE CCHHIILLEEAANN CCLLAASSSSSSTTRRUUCCTTUURREE

The ldquocommon social fluidityrdquo hypothesis assertsthat all international variation in mobility pat-terns is attributable to highly specific histori-cally formed national characteristics In theChilean case these factors may be related to themarket reform of the 1970s and 1980s and therapid economic growth since the late 1980sNational mobility studies suggest that coun-tries experiencing rapid industrialization may becharacterized by weaker sector barriers to mobil-ity specif ically barriers between the self-employed and employees and betweenagricultural and nonagricultural classes(Goldthorpe Yaish and Kraus 1997 and Yaish2004 for Israel Ishida Goldthorpe and Erikson1991 for Japan Park 2004 for Korea and Costa-Ribeiro 2003 for Brazil)

In addition two components of the Chileanmarket transformation may have contributed tothe weakness of mobility barriers across sectorsof the economy The first component is theagrarian reform and subsequent counterreformundertaken by the military regime The secondone is the opening of the economy to interna-tional trade and the retrenchment of the statewhich led to a decline in formal employment

443300mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Figure 3 Earnings Inequality in Greatest Santiago 1957ndash2000Source Larranaga 1999 and calculations by the author

3 Santiago the Chilean capital comprises aboutone-third of the countryrsquos population Assessmentby Chilean experts suggests that Santiago trendspresent an unbiased picture of trends at the nationallevel Although earnings inequality is not as com-prehensive a measure as total income inequality it isan adequate much more accurately measured proxy(Galbraith 2002)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

and to the growth and segmentation of the self-employed sector

The main objective of the Chilean agrarianreform (1962ndash1973) was to reduce the extremeland concentration in the country The reformwas very successful In 1955 the top 7 percentof landowners held 65 percent of the land andthe bottom 37 percent held only 1 percent By1973 the last year of the reform 43 percent ofthe land had been expropriated (Kay 2002 Scott1996) An agrarian counterreform was launchedby the military regime as a way of returning theland to its former owners It did not howeverrestore the traditional hacienda order (Gomezand Echenique 1988 Rivera 1988) In fact themilitary government returned only one-third ofthe plots to their old owners sold another thirdand adjudicated the remaining third to the smallproprietors benefited by the agrarian reformUnable to compete with the now cheap foodimports many small proprietors opted to selltheir plots As a consequence as much as two-thirds of the Chilean agricultural land came upfor sale creating an active land market Themain beneficiaries of the ldquomarketization of thecountrysiderdquo were a new group of export-ori-ented entrepreneurs and international investorswho bought large tracts of land (Gwynne 1996Kay 2002) Thus the reform and counterreformtriggered a massive ldquochange of handsrdquo in theChilean countryside likely altering the patternsof land inheritance

The second major effect of the market reformon the class structure was the reduction of for-mal employment caused by the decline of theindustrial working class and by the shrinkage ofthe state Incapable of competing with the nowcheap imports the working class plummetedfrom 34 percent of total employment in theearly 1970s to 20 percent in 1980 while pub-lic employment dropped from 14 to 84 per-cent in the same period (Leon and Martinez2000 Schkolnick 2000 Velasquez 1990) Thedecline in industrial and public employmentled to an increase in self-employment fromabout 15 percent in 1970 to about 28 percent in1980 (Thomas 1996) followed by a declinethrough the economic recovery before stabi-lization at about 22 percent of the total employ-ment in 2000 (MIDEPLAN 2001) In contrastto industrialized countries in which the self-employment rate usually is within the one-digitrange and involves capital ownership in Chile

it encompasses more than one-fifth of the laborforce Furthermore self-employment hadbecome a survival strategy for a large numberof former industrial and public employeesappropriately labeled ldquoforced entrepreneurial-ismrdquo (Infante and Klein 1995 Portes Castellsand Benton 1989) As a consequence theChilean self-employed sector is voluminousand segmented It includes a small segment ofentrepreneurs oriented to capital accumulationand able to hire employees and a large seg-ment of self-employed workers mostly engagedin ldquosurvival activitiesrdquo and unable to hire thelabor of others Differences in economic well-being between these two groups are massivewith the former group earning on averagethree times more than the latter (MIDEPLAN2001)

To be sure a large self-employed sector is nota result of the market transformation Self-employment has been historically high in LatinAmerica because according to some scholarsthe region has experienced ldquodependent inte-grationrdquo into the world capitalistic system(Hopkins and Wallerstein 1982 Luxembourg1951) However the internal heterogeneity ofthe self-employed sector increased during themarket reform in a pattern that foretold whathappened in the rest of Latin America during the1990s (Klein and Tokman 2000 Portes andHoffman 2003) and what may be current devel-opments in the industrialized world (Arum andMuumlller 2004 Noorderhaven et al 2003) Thesecharacteristics of the Chilean class structureare considered in the comparative analysis ofChilean mobility

DDAATTAA AANNDD AANNAALLYYTTIICCAALL AAPPPPRROOAACCHH

This study uses the 2001 Chilean MobilitySurvey (CMS) The CMS is a nationally repre-sentative multistage stratified sample of maleheads of household ages 24 to 69 The samplingstrategy includes the following stages First 87primary sampling units (PSUs) (counties) areselected Then blocks within the PSUs areselected and finally households within blocksare chosen Counties are stratified according tosize (fewer than 20000 20000 to 100000100000 to 200000 and more than 200000inhabitants) and by geographic zone (NorthCenter South) All PSUs in the large size stra-tum are included in the sample to increase effi-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443311

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ciency The fieldwork conducted between Apriland June 2001 consists of face-to-face inter-views in the respondentrsquos household carried outby trained personnel The survey excludesnonndashhead-of-household males which represent17 percent of the male population of the rele-vant age (MIDEPLAN 2001) Among thoseexcluded 865 percent are the sons of the headof the household 124 percent are another rel-ative of the head and 11 percent are other non-related males Their occupational distributionwith control used for age is almost identical tothat of the heads of household The small pro-portion represented by this group and their sim-ilar occupational distribution suggest that theirinclusion would not significantly alter the find-ings presented in this article Excluding thehouseholds not eligible for the survey theresponse rate is 63 percent Although nonre-sponse rates usually are not reported in Chileansurveys exchange with Chilean experts indi-cates that the nonresponse rate is about 20 to 25percent for face-to-face household surveys Thehigher nonresponse rate of the CMS is likelyattributable to the difficulty contacting maleheads of household4 The total sample size is3544 I exclude individuals outside the agerange of 25 to 64 years which is conventional-ly used in comparative mobility research andcases with unusable data After this exclusionthe usable sample size is 3002

CCLLAASSSS SSCCHHEEMMAA

This study uses a class perspective to describethe Chilean social structure It utilizes the seven-category version of the classification devel-oped by the CASMIN project Thisclassification is widely used in internationalcomparative research and describes the basic

stratification of advanced industrial societiesbased on ldquoemployment relationshipsrdquo (Eriksonand Goldthorpe 1992a35ndash47 Goldthorpe andHeath 1992) The seven classes included in theschema are I+II (service class) III (routine non-manual workers) IVab (self-employed work-ers) V+VI (manual supervisors and skilledmanual workers) VIIa (unskilled manual work-ers) IVc (farmers) and VIIb (farm workers)

Use of the 7-class categorization instead ofan alternative detailed 12-class version pro-duced by the CASMIN project (Erikson andGoldthorpe 1992b) induces some loss of infor-mation about the originndashdestination associa-tion A global test of all five aggregationsobtained from the ratio of L2 tests for the inde-pendence model in the collapsed and full tablesindicates that the seven-class schema masks 29percent of the association shown by the full setof classes ([L2 full table ndash L2 collapsed table]L2 full table = [108644 ndash 77191]108644 =29) This proportion is high compared withthat of the CASMIN countries and is surpassedonly by Sweden with 31 percent (Hout andHauser 1992 Table 2) A collapsed sevenfoldclass schema was however preferred because ofthe moderate sample size of the Chilean surveyand because it grants international comparabil-ity of the findings5

Operationalization of this class classifica-tion is based on detailed information concern-ing job title (recoded into the standard ISCO-88classification [ILO 1990]) industry occupa-tional status and supervisory status of workersBecause the CMS is specifically designed toassess class mobility it includes all the infor-mation necessary to produce the CASMIN classschema The CASMIN project has not produceda standard algorithm to generate the class clas-sification but other researchers have producedsuch an algorithm which I use in this analysis(Ganzeboom and Treiman 2003) Table 1 pres-ents the basic intergenerational mobility tableto be used in this analysis including the countsand marginal percentage distribution for originand destination classes

443322mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

4 The nonresponse rate can yield bias if those whowere unreachable or refused to participate differ fromthose included in the sample To estimate the mag-nitude of this bias I compared the distribution of keyvariables with the CASEN 2000 survey CASEN isa large survey (n = 252595) conducted by the Chileangovernment and has a refusal rate lower than 10percent The comparison (available upon request)suggests that the CMS slightly underrepresents agri-cultural workers and the upper class Overall how-ever there is no indication of major nonresponsebias

5 Analyses were replicated using the detailed 12-class table and no difference in substantive find-ings was found

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

MMEETTHHOODDSS

I fit two alternative models to the Chilean mobil-ity table The first is the ldquocore model of social flu-idityrdquo which claims to capture basic internationalsimilarity in mobility Although the core modelhas become a benchmark for mobility researchand currently is the only model that provides acomparative framework for mobility analysisanalysts have drawn attention to a number of itslimitations The model has been criticized forinadequately representing hierarchical mobilityeffects for being tailored to a few industrializedcountries for the ad hoc nature of some effects itincludes (specifically the affinity effects) andfor lacking a criterion to determine how muchdivergence is necessary for it to be rejected(Ganzeboom Luijkx and Treiman 1989 Houtand Hauser 1992 Sorensen 1992 Yamaguchi1987) I therefore also test an alternative ldquohybridmodelrdquo that combines the rowndashcolumn (II) asso-ciation model (Goodman 1979) with parametersaccounting for class immobility The internationaland temporal comparison of Chilean mobilitypatterns uses the uniform-difference (UNIDIFF)model also known as the multiplicative layereffect model (Erikson and Goldthorpe 1992aXie 1992) Model comparison and selection isbased on the Bayesian Information Criterion(BIC) statistic (Raftery 1995)

AANNAALLYYSSIISS OOFF CCHHIILLEEAANN MMOOBBIILLIITTYY IINNCCOOMMPPAARRAATTIIVVEE PPEERRSSPPEECCTTIIVVEE

TTHHEE CCOORREE MMOODDEELL OOFF SSOOCCIIAALL FFLLUUIIDDIITTYY

I start by fitting the core model to the Chileanmobility table The core model represents the

basic international commonality in social flu-idity using a topological formulation In contrastto standard topological models constructed froma single allocation of cells (Hauser 1978) thecore model uses eight matrices each of whichis designed to capture a particular effect thatenhances or reduces mobility between specificclasses (Erikson and Goldthorpe 1987a 1992achapter 4) In this way the model permits thesources of commonality across nations to beidentified clearly and likewise the departuresfrom it The effects are of four types hierarchyinheritance sector and affinity6

HIERARCHY EFFECTS Hierarchy effects reflectthe impact of status distances between classeson fluidity between them To estimate theseeffects the mobility table is divided into threestrata reflecting differences in resources andrewards across classes The strata are the fol-lowing upper stratum (class I+II) middle stra-tum (classes III IVab IVc and V-VI) and lowerstratum (classes VIIa and VIIb) Because thereare three strata two hierarchy effects (HI1 andHI2) can be identified each representing thecrossing of one additional hierarchical barrierThe association between each pair of classes isexpressed as an inverse function of the numberof hierarchical strata crossed

INHERITANCE EFFECTS Inheritance effects aredesigned to capture the propensity for class

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443333

2322-ASR 703 filename70304-torche

Table 1 Seven-Class Intergenerational Mobility Table for Chilean Men

Class of Destination

I-II III IVab IVc V-VI VIIa VIIb N

I-II Service Class 229 49 81 2 37 30 2 430 14III Routine Nonmanual 61 15 34 0 24 27 1 162 5IVab Self-employed 123 37 173 28 102 90 20 573 19IVc Farmers 42 16 68 54 58 54 24 316 11V-VI Skilled Manual 91 33 116 13 155 135 36 579 19VIIa Semiskilled and Unskilled Manual 49 43 115 13 111 123 44 498 17VIIb Farm Workers 28 16 78 17 89 101 115 444 15N 623 209 665 127 576 560 242 3002 21 7 22 4 19 19 8

Note Cell entries represent counts

6 Design matrices in Erikson and Goldthorpe(1992a124ndash129)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

immobility They include three matrices IN1identifies immobility across all classes alike IN2identifies the higher immobility of the serviceclass (I+II) the self-employed group (IVab) andfarmers (IVc) and IN3 accounts for the highestimmobility of farmers (IVc)

SECTOR EFFECTS Sector effects are intended toidentify the difficulty of moving between theagricultural and nonagricultural sectors of theeconomy (SE1)

AFFINITY EFFECTS Affinity effects are intend-ed to capture specific discontinuities (negativeaffinities) or linkages (positive affinities) betweenclasses which either reinforce or offset the over-all effects of hierarchy and sector There are twoaffinity effects The first AF1 identifies the dif-ficulty of moving between the service class (I+II)and the class of farm workers (VIIb) which addsto their hierarchical distance In contrast AF2identifies instances in which mobility is morefrequent than accounted for by hierarchy and sec-tor effects and includes affinities within the non-manual sector (I+II and III) and within the manualsector (V+VI and VIIa) a symmetrical affinitybased on capital possession (IVab and IVc and Iand IVab) and an asymmetrical link between agri-cultural classes of origin (IVc and VIIb) and theunskilled manual class of destination (VIIa)7

TTHHEE MMOOBBIILLIITTYY PPAATTTTEERRNN IINN CCHHIILLEECCOONNVVEERRGGEENNCCEE OORR DDIIVVEERRGGEENNCCEE

Table 2 presents the fit of several models appliedto the Chilean data and Table 3 displays theparameter estimates for each of these modelsModel 1 in Table 2 indicates that the fit of theoriginal core model to the Chilean table is unsat-isfactory under standard statistical criteria butit explains a significant 775 percent of associ-ation under independence To assess the strengthof the different factors driving mobility oppor-tunities I compare the Chilean coefficients withthose obtained for the CASMIN countriesCoefficients from the original core model fit-ted to the Chilean table are reported in row 1 ofTable 3 CASMIN coefficients are reported inrow 2 and differences in coefficientsrsquo magni-tude and the statistical significance of the dif-ferences are reported in row 38

All Chilean coefficients are significant andhave the same sign as for the CASMIN coun-tries Comparison of the magnitude of the coef-ficients however shows significant departuresfrom the core for five effects which revealinteresting differences between Chile and theCASMIN countries The hierarchy coefficientsindicate that whereas the short-range hierar-chical barriers are somewhat weaker in Chile

443344mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

8 CASMIN coefficients were obtained from fittingthe core model to the combined table of France andEngland standardizing both tables to have 10000cases as recommended by Erikson and Goldthorpe(1992a121) The coefficients obtained are virtuallyidentical to those presented by Erikson andGoldthorpe (1992a Table 44)

Table 2 Fit statistics selected models to the Chilean Mobility Table

AssociationL2 BIC df p Explained

Core Modelmdash1 Core Model Original 1375 285 28 000 775mdash2 Core Model EHE 1081 ndash08 28 000 860mdash3 Chilean version of Core Model ([Model 2] + Chilean Parameters) 724 ndash326 27 000 906Association Models 000mdash4 Heterogeneous Quasi-RC(II) Model 734 ndash161 23 000 905mdash5 Homogeneous Quasi-RC(II) Model 803 ndash287 28 000 896mdash6 Quasi-Linear by Linear Association (SES row and column ranking)a 1563 279 33 000 798

Note EHE = empirical hierarchical effects SES = eocioeconomic statusa Linear-by-linear association model using socioeconomic status scores as presented in Appendix Figure A1 torank origin and destination classes

7 For a detailed account of the empirical derivationof the model see Erikson and Goldthorpe (1987a1987b and 1992a chapter 4)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443355

2322-ASR 703 filename70304-torche

Tab

le 3

M

odel

Par

amet

er E

stim

ates

for

the

Chi

lean

Mob

ilit

y Ta

ble

Eff

ect

1 C

ore

mod

el o

rigi

nala

mdash 2 C

ore

mod

el C

AS

MIN

b

mdash 3 D

iffe

renc

e C

ASM

INmdash

Chi

lec

mdash 4 C

ore

mod

el E

HE

d

mdash 5 C

hile

an v

ersi

on o

f co

ree

mdash Not

eD

ata

show

n ar

e m

odel

par

amet

er e

stim

ates

wit

h st

anda

rd e

rror

s in

par

enth

eses

EH

E =

em

piri

cal h

iera

rchi

cal e

ffec

ts

aM

odel

1 f

rom

Tab

le 2

b

Ori

gina

l Cor

e M

odel

fit

ted

to th

e E

ngli

sh a

nd F

renc

h m

obil

ity

tabl

es S

ee te

xt f

or d

etai

ls

cD

iffe

renc

e in

par

amet

er m

agni

tude

bet

wee

n or

igin

al c

ore

mod

el f

itte

d to

the

Chi

lean

tabl

e (r

ow 1

) an

d to

the

CA

SM

IN ta

ble

(row

2)

dM

odel

2 f

rom

Tab

le 2

e

Mod

el 3

fro

m T

able

2

dagger p

lt 1

p

lt 0

5

p

lt 0

1

p

lt 0

01 (

two-

tail

ed)

HI1

ndash12

(

05)

ndash22

(02

)ndash

10 dagger

(06

)ndash

03(

07)

ndash09

dagger(

06)

HI2

ndash65

(11

)ndash

45

(

04)

20

dagger(

12)

ndash74

(09

)ndash

47

(

10)

IN1

41

(

08)

43

(

03)

Insi

g

37

(

07)

34

(

07)

IN2

39

(

10)

84

(

04)

45

(

11)

31

(13

)mdash

IN3

68

(

24)

101

(11

)In

sig

61

(24

)4

8(

23)

SE

1

ndash38

(07

)ndash1

06

(

05)

ndash68

(08

)ndash

42

(

07)

ndash20

(

08)

AF

1

ndash76

(21

)ndash

83

(

13)

Insi

g

ndash84

(20

)ndash

63

(21

)

AF

2

33

(

05)

45

(

02)

ndash12

(

06)

20

(

06)

26

(

06)

AF

1ndashC

mdash mdash mdash mdash

ndash14

9

(46

)

IN2ndash

C

mdash mdash mdash mdash 71

(

14)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(HI1) long-range hierarchical mobility is moredifficult in this country (HI2) This suggestssubstantial inequality between the extremes ofthe class hierarchy combined with smaller dif-ferentiation in the middle of the distribution Asto the inheritance effects the service class theself-employed group and the farmers (classesI+II IVab and IVc respectively) have a sig-nificantly lower propensity to immobility inChile than in the CASMIN countries suggest-ing that class positions usually associated withindependent work are less able to reproducetheir class status intergenerationally than depict-ed by the core Most impressive the barrierbetween the agricultural and nonagriculturalsectors of the economy (SE1) is much weakerin Chile as is the affinity effect linking non-manual classes manual classes and the class-es that own capital among themselves (AF2)

Interestingly all the significant differences incoefficients with the single exception of the bar-rier to long-range hierarchical mobility sug-gest that Chile is more fluid than the core modeldepicts Higher fluidity seems to be driven bythe weakness of horizontal barriers separatingthe agricultural manual and self-employed sec-tors of the economy This is consistent withfindings of weaker sector effects in countriesthat have experienced rapid industrializationsuch as Korea (Park 2004) Brazil (Costa-Ribeiro 2003) Israel (Yaish 2004 Goldthorpeet al 1997) and Japan (Ishida et al 1991)

Additionally in the Chilean case the weak-ness of sector barriers may have been intensi-fied by the deep market transformation of the1970s and 1980s On the one hand the ldquochangeof handsrdquo in the agricultural sector induced bythe agrarian reform and counterreform likelyincreased fluidity between the agricultural andurban classes On the other hand the role of self-employment as an ephemeral refuge against(industrial and public sector) unemploymentmay have reduced barriers between the self-employed group and the rest of the social struc-ture Added to the larger difficulty of long-rangehierarchical mobility these features suggestweak sector cleavages in the Chilean fluiditypattern and high salience of the hierarchicaldimension of mobility

For the countries in which the fit of the coremodel is not adequate Erikson and Goldthorpe(1992a145) suggest adding parameters thataccount for country-specific deviations from

the model I therefore introduce adjustments toreflect salient characteristics of the Chilean flu-idity pattern producing a ldquoChilean versionrdquo ofthe core model

First I modify the hierarchical effects Fittingof the core model suggests that hierarchicalbarriers are a crucial determinant of mobility inChile This conclusion is however obscuredby the fact that the hierarchical strata distin-guished by the core model are not empiricallyobtained and may not accurately represent theChilean ranking of classes in terms of socioe-conomic status (SES) To model hierarchicaleffects in Chile adequately I rank classes usingthe unweighted average of their schooling andearning levels as a proxy for SES I then collapsethe seven classes into three hierarchical stratausing cluster analysis These three empiricallyobtained strata are the following Upper stratum(service class I+II) middle stratum (routinenonmanual and self-employed group III andIVab) and lower stratum (manual and agricul-tural classes V-VI VIIa IVc and VIIb) Detailson the ranking of classes cluster procedures andthe comparison of the CASMIN and empiricalChilean class ranking can be found in theAppendix Comparison of the CASMIN and theempirical class ranking suggests higher differ-entiation at the top than at the bottom of thesocial structure in the Chilean case

Model 2 in Table 2 uses the empiricallyobtained hierarchical strata (EHE) Theimprovement in fit is massive (the difference inBIC is 293 without using any degrees of free-dom) The model now accounts for 86 percentof the association under independence

The large improvement in fit when empiricalhierarchical strata are used challenges theassumption of an internationally homogeneoushierarchical class ranking At the empiricallevel there is no proof that the hierarchicalranking of classes assumed by the core modelfits all CASMIN countries9 Moreover someevidence points to significant cross-country dif-

443366mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

9 Erikson and Goldthorpe(1992a Table 22) pres-ent the average ranking of classes based on standardstatus measures but they do not show the empiricalranking of classes for each country If the occupa-tional composition of classes varies across countriesthere may be significant international differences inthe hierarchical ranking of classes

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ferences in the ranking of classes (Hout andHauser 1992 Sorensen 1992) At the theoreti-cal level the creators of the core model high-light the relevance of highly specific historicallyformed national attributes to explain interna-tional variation in mobility These national par-ticularities may very well result in differenthierarchical positions of classes across coun-tries10 Furthermore even if the homogeneousranking of classes imposed by the CASMINresearchers applied to most CASMIN coun-tries significant departures may be found inthe late industrialized countries currently beingadded to the comparative mobility project TheChilean findings suggest that instead of impos-ing international homogeneity to the hierarchi-cal dimension of mobility the ranking of classesshould be treated as nationally specific param-eters to be estimated as mobility researchersroutinely treat row and column marginals ofthe mobility table The issue is importantbecause the hierarchical dimension is the mostrelevant dimension of mobility (Hout andHauser 1992 Wong 1992) and inadequate mod-eling can bias assessment of not only hierar-chical effects of mobility but also other effects

Two other changes were introduced toaccount for Chilean-specific deviations fromthe core model First the IN2 effect wasrearranged to exclude the self-employed class(IVab) and to include the class of farm workers(VIIb) This rearrangement accounts for thefact that the Chilean self-employed classappeared less likely than depicted by the coremodel to reproduce its class position acrossgenerations and the inheritance level of Chileanfarm workers is higher than in the CASMINcountries These changes are expressed in theChilean-specific inheritance parameter IN2-CSecond an affinity parameter (AF1-C) wasadded to capture the difficulty of long-rangedownward elite mobility This negative asym-metrical affinity links flows from the service

class to the agricultural classes and from theroutine nonmanual class to the class of farmworkers Note that the AF1-C adds to the hier-archical distance between the two extremes ofthe class structure and to the negative affinitycaptured by AF1 to depict the extreme difficultyof long-range downward mobility from theChilean elite The fit of this lsquoChilean versionrsquoof the core model is reported in Table 2 model3 The improvement in BIC from model 2 is sub-stantial Furthermore even if the model fallsshort in terms of standard statistical criteria itaccounts for a large 91 of the associationunder independence thereby supporting theclaim that it adequately represents the Chileanmobility pattern The parameter estimates for theChilean version of the core are reported in row5 Table 3

Because of the core modelrsquos undesirable prop-erties I also use an alternative approach toexamine Chilean mobility patterns Thisapproach uses the now standard associationrowndashcolumn (II) (RC[II]) model (Goodman1979 see also Hout 1983 and Wong 1992) Inthis model both origin and destination classesare scaled so that the association can beexpressed as a linear-by-linear interaction by asingle parameter The multiplicative form ofthe model can be expressed as follows

Fij = iO j

D exp( microi νj)

where i indexes class of origin j indexes classof destination Fij is the expected frequency inthe (ij) cell is the grand mean i

O pertainsto the class of origin marginal effect j

D pertainsto the class of destination marginal effect isa global association parameter microi is the scalevalue for the ith class of origin and νj is the scalescore for the jth class of destination subject tothe following normalization constraints Σmicroi =Σνj = 0 (normalization of the location) and Σmicroi2

= Σνj2 = 1 (normalization of the scale)The class scale scores reflect a latent con-

tinuous variable made manifest by the class cat-egories Empirically obtained from the datathese ldquodistancerdquo scores produce an optimalranking of classes for the purpose of mobilityanalysis The RC(II) model is appropriate formodeling the association of any two ordinalvariables but mobility tables are distinctivebecause of the correspondence between class oforigin and class of destination (Gerber and Hout2004691)

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443377

2322-ASR 703 filename70304-torche

10 For instance during the socialist period inHungary the status of the proletariat may have beenindistinguishable from the status of the routine non-manual class (Szelenyi 199862) In Scotland givento early industrialization the hierarchical barrierbetween the skilled and unskilled working classes thatcharacterizes other CASMIN countries may not exist(Erikson and Goldthorpe 1992a163)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I introduce two adjustments to account forthis correspondence transforming the formu-lation into a ldquohybrid modelrdquo First topologicalparameters will account for the large counts insome diagonal cells reflecting immobility(inheritance) effects These parameters distin-guish two inheritance levels ldquohigh inheritancerdquofor the service class (I+II) farmers (IVc) andfarm workers (VIIb) and ldquolow inheritancerdquo forall other classes with the exception of the rou-tine nonmanual workers

Model 4 in Table 2 reports the fit statistic forthis quasi-RC(II) model The BIC statisticrejects it and prefers the more parsimoniousChilean version of the core model ThereforeI introduce a second adjustment to account forthe correspondence between rows and columnsin mobility tables Origin and destination scoresare constrained to be the same assuming thatthe scaling of classes has not significantlychanged over time Model 5 in Table 2 reportsfit statistics for this homogeneous quasi-RC(II)model On the basis of the BIC statistic the fitof the model is as good as that of the Chileanversion of the core model (Wong [1994] showsthat BIC differences of less than 5 points shouldbe considered indeterminate) Thus under stan-dard statistical criteria the Chilean core modeland the homogeneous quasi-RC(II) model areindistinguishable Given that the Chilean versionof the core model was tailored to fit the Chileantable the fit of model 5 is impressive and it indi-cates that a unidimensional scale can adequatelycapture mobility distances between classes

The logical next question is what this scalerepresents and specifically whether it repre-sents the hierarchical ranking of classes in termsof SES If a close correspondence is foundbetween the mobility distances among classesand their distances in terms of SES this find-ing would suggest that the intragenerationaldistribution of resources and rewards expressedin the SES ranking of classes drives the processof intergenerational mobility In other words thisfinding would support the resource approachrsquosclaim that differences in resources associatedwith diverse class positions determine mobili-ty opportunities across generations

To examine this possibility I compare theclass scores obtained from the homogeneousquasi-RC(II) model with the empiricallyobtained SES ranking of classes based on edu-cation and earnings (Appendix) Figure 4 pres-

ents the comparison between the standardizedvalues of both sets of scores The correspon-dence is striking The mobility distancesbetween classes closely reproduce their dis-tances in terms of SES This close isomorphismsuggests that at least in Chile the intragener-ational inequality in the distribution of socialresources and rewards across classes largelydrives the intergenerational mobility process Totest the isomorphism further I use the SESclass scores as row and column rankings to esti-mate a quasindashlinear-by-linear association model(model 6 Table 2) The fit of model 6 is sig-nificantly worse than that of model 5 almost aspoor as that of the original core model This indi-cates that although the inter- and intragenera-tional dimensions of inequality are very similarin Chile they have a somewhat different struc-ture

In summary the central characteristic of theChilean mobility regime is the predominance oflong-range hierarchical barriers and the weak-ness of sector cleavages separating classes in themiddle and at the lower end of the hierarchicalstructure Findings from the ldquoChilean versionrdquoof the core model and the quasi-RC(II) hybridmodels are consistent depicting a social struc-ture characterized by signif icant barriersbetween the top echelon and the rest of the classstructure The good fit of the homogeneousquasi-RC(II) model provides preliminary sup-port for the resource approach linking inequal-ity and mobility Mobility dynamics seem to bedriven largely by the hierarchical distancesbetween classes and overall Chile seems not tobe less fluid than the CASMIN countries

CCOOMMPPAARRIISSOONN OOFF TTHHEE OORRIIGGIINNAALL AANNDD CCHHIILLEEAANN

VVEERRSSIIOONNSS OOFF TTHHEE CCOORREE MMOODDEELL

Although Chilean mobility dynamics seem to bedriven by inequality across classes in a veryunequal society the fitting of the core modelprovides no indication that Chile is significantlyless fluid than depicted by the core model Asa preliminary analysis of the Chilean level of flu-idity I compare the predicted propensities formobility and immobility between specific pairsof classes in Chile with those in the CASMINcountries Propensities are obtained from theoriginal and Chilean versions of the core modelrespectively (models 2 and 5 of Table 3)

443388mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I begin with the immobility of those at the topof the class structure namely the service class(I+II) The service classrsquo propensity for immo-bility in the original core model is captured bytwo parameters IN1 and IN2 This propensityis more than three times what it would be in theabsence of these effects (e43+84 = 356) InChile the propensity for immobility of class I+IIis captured by the same two parameters and itis lower than what the core model expresses286 times larger than it would be in the absenceof these effects (e34+71) What about the immo-bility of the two agricultural classes In theCASMIN countries immobility of the IVc andVIIb classes is respectively 785 and 154 timeswhat it would be in the absence of these param-eters (e43+84+101ndash22 and e43 respectively) InChile the comparable values are 422 and 286(e34+71+48ndash09 and e43+71 respectively) indi-cating the relatively low inheritance associatedwith land ownership in the Latin Americannation Thus if attention is focused on theimmobility in the upper and lower ends of thesocial hierarchy the comparison does not indi-cate less fluidity in the highly unequal Chileansociety

What about the mobility rates between thetwo most distant classes in the hierarchical rank-ing namely the service class (I+II) and farmworkers (VIIb) In the case of the CASMINcountries the mobility chances between theseclasses are captured by four parameters (HI1HI2 SE1 and AF1) and is 077 times smallerthan it would be in the absence of these effects(endash22 ndash45ndash106ndash83) In Chile the distance fromclass I to class VIIb is expressed by these samefour parameters in addition to the asymmetri-cal disaffinity AF1-C This yields mobilitychances 056 times what would be obtained inthe absence of these effects (endash09 ndash47ndash20ndash63ndash149)Thus the barrier to long-range downwardmobility from the elite is more significant inChile than in the CASMIN countries The sameoccurs with long-range downward mobility fromclass III

If we measure the reciprocal flow (ieupward mobility from class VIIb to class I+II)we obtain the same value for the CASMINcountries because all parameters are symmet-rical in the original core model For Chile how-ever this calculation removes the asymmetricalbarrier to elite downward mobility which rais-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443399

2322-ASR 703 filename70304-torche

Figure 4 Class Scores Obtained from the Homogeneous Quasi-RC(II) Model and from Socioeconomic Status(SES) Ranking

Note (1) = Class standardized socioeconomic status scores based on schooling and earnings (2) = Class stan-dardized scores based on the Homogeneous Quasi-RC(II) model (see text for details)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 6: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

increase over time in the United Statesbetween the early 1970s and the mid 1980s(Hout 1988) However these increases in flu-idity are not univocally preceded by a reduc-tion of inequality thus casting doubts on apotential association between inequality andmobility

A careful analysis of mobility trends inRussia shows a significant decline in mobili-ty associated with an increase in inequalityafter the market transformation of the early1990s (Gerber and Hout 2004) which pro-vides additional support for the resourceapproach The advantage of the Russian studyis that ldquothe market transition in Russia ||altered so many fundamental economic insti-tutions so rapidly that we can confidentlyascribe changes in social mobility || to thissource rather than to cultural change or indus-trializationrdquo (Gerber and Hout 2004678) It isnot clear however whether it is growth ininequality recession some other change asso-ciated with the radical liberalization of theeconomy or a combination of these factorsthat triggered a decline in fluidity in contem-porary Russia

In this context Chile presents an ideal casefor an examination of the association betweeninequality and mobility Given the extremeeconomic inequality in the country if theunequal distribution of resources or incentiveshas an impact on mobility opportunities theChilean level of fluidity should be signifi-cantly different from that of industrializednations Additionally given the increase ininequality associated with the market reformof the 1970s and 1980s analysis of mobilitybefore and after the reform would provide sup-plementary evidence concerning the potentialmobilityndashinequality relationship Naturally asingle case study will not supply a definitiveanswer to these questions However by com-bining an examination of the Chilean mobili-ty regime in the context of historicaltransformations in the country with an inter-national comparative analysis and with anassessment of mobility trends over time thisarticle provides important insights into theexistence of a link between mobility andinequality and the mechanisms driving it

TTHHEE CCHHIILLEEAANN CCOONNTTEEXXTT IINNCCOOMMPPAARRAATTIIVVEE PPEERRSSPPEECCTTIIVVEE

During the second half of the 20th centuryChile transformed from an agrarian semifeu-dal society into an urban service-based oneBetween 1950 and 2000 the rural populationdeclined from almost 40 to 17 percent (Braunet al 2000 INE 2002) This defines Chile asa mostly urban country with an 83 percentrate of urbanization larger than the 78 percentrate in the United States In tandem with urban-ization Chile experienced a reallocation ofemployment from the agricultural to the terti-ary sector of the economy The share of agri-culture in total employment declined from 38percent in 1950 to 17 percent in 2000 Whereasthe share of manufacturing remained constantat about 18 percent the share of the servicesector rose from 42 to 65 percent (Braun et al2000 INE 2002)

Urbanization and tertiarization are process-es that virtually all countries experienced dur-ing the 20th century Within this secular trendthe Chilean political economy is marked byspecific institutional developments that shapeits stratification structure From the 1940s tothe 1970s the Chilean economic landscape asthat of its Latin American neighbors wasdefined by import-substitutive industrialization(ISI) Emerging as a reaction to the collapse ofinternational trade caused by the GreatDepression (Ellworth 1945) and based on theldquodeterioration of the terms of traderdquo theory(Prebisch 1950) ISI was based on two typesof policies The first was oriented to closing theChilean economy to international markets andthe second was oriented to promoting nation-al industrialization The Chilean state becamethe leading productive agent supporting indus-try through credit investments and technicalassistance and taking a direct productive rolethrough the creation of public enterprises(Stallings 1978) After a sanguine beginningwith industrial production growing at almost7 percent per year between 1940 and 1950(Mamalakis 1976 Munoz 1968) ISI started tofail economic growth stagnated and socialturmoil resulting from massive urban migra-tion and vast social inequalities increased

In 1964 a progressive administration tookpower and adopted as its mandate the correc-tion of extreme inequality in the country by a

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442277

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ldquoRevolution in Freedomrdquo (Gazmuri 2000)This progressive government launched redis-tributive policies including an agrarian reformand an educational reform The redistributiveagenda boosted by a socialist administrationthat came to power in 1970 was abruptly halt-ed by a military coup in 1973 The militarytook power and retained it until 1990 Duringthese 17 years the military regime conduct-ed a deep market-oriented transformationNow turned into the ldquoWashington consensusrdquoparadigm (Williamson 1990) this reformincluded macroeconomic stabilization dereg-ulation of prices and markets and the priva-tization of enterprises and social services Ittransformed Chile from a closed economywith heavy state intervention into one of themost open market-based economies in theworld with the productive and welfare role ofthe state reduced to a minimum (Edwards andCox-Edwards 1991 Martinez and Diaz 1999Meller 1996 Velasco 1994)

The depth of the market reform coupledwith a world recession led in the late 1970sand early 1980s to the deepest economic cri-sis since the Great Depression A third of thelabor force was unemployed and povertyafflicted nearly half of Chilean households(Meller 1991) The post-crisis recovery start-ing in the late 1980s was substantial and sus-tained and coincided with theredemocratization of the country The grossdomestic product (GDP) per capita grew morethan 6 percent annually for 15 years a novelrate for Chile only comparable in recent timesto the ldquoEast Asian Miraclerdquo (World Bank1993) The sharp economic growth has trans-formed Chile from one of the poorest coun-tries in Latin America (Hofman 2000) into aldquomiddle-income economyrdquo (World Bank2003a) In the year 2000 the income per capi-ta was approximately US$5000 much lowerthan the United States average of US$31910but the highest in Latin America (World Bank2003a) As a consequence of the sharp eco-nomic growth Chileans have reached levels ofconsumption unthinkable two decades agoand the poverty rate fell from 45 to 21 percentbetween 1985 and 2000 (MIDEPLAN 2000Raczynski 2000)

TTHHEE CCHHIILLEEAANN LLEEVVEELL AANNDD PPAATTTTEERRNN OOFF

IINNEEQQUUAALLIITTYY

The dark side of this ldquosuccess storyrdquo is eco-nomic inequality Structurally rooted in a feu-dal agrarian structure the institutional legacyof the colonial period and in the slow expan-sion of education (Engerman and Sokoloff1997 World Bank 2003b) inequality hasremained persistently high during the 20thcentury The Gini coefficient reached 58 in the1990s which compares with a much lowerGini of 34 among the industrialized coun-tries and is large even in the highly unequalLatin American context with its average Giniof 49 (Deininger and Squire 1996 Marceland Solimano 1994) As Figure 1 indicatesChilean inequality is almost twice that in mostindustrialized countries and 15 times that inthe United States the most unequal nation ofthe industrialized world

Not only the level but also the pattern ofinequality in Chile significantly departs fromthat of the industrialized world With thewealthiest Chilean decile receiving 423 per-cent of the total national income (MIDEPLAN2001) the Chilean pattern of inequality ischaracterized by high ldquoconcentration at thetoprdquo Although inequality is by definition relat-ed to concentration the Chilean case isextreme as compared with the industrializedworld and even with other Latin Americancountries A comparison between the incomeof each decile and the income of the preced-ing (poorer) decile illustrates the point Theratio between the wealthiest and the secondwealthiest decile is twice as large in Chile asin the United States and England and one ofthe largest in Latin America depicting highelite concentration In contrast the ratiobetween the second poorest and the poorestdeciles in Chile is half that of the United Statesand England indicating that inequality at thebottom of the income distribution is muchlower in Chile than in these industrializednations (Szekely and Hilgert 1999)

In fact as Figure 2 indicates Chile is thefourth most unequal country in the mostunequal region of the world However if thewealthiest decile is excluded Chilean inequal-ity is dramatically reduced and Chile becomesthe most equal Latin American country evenmore equal than the United States (Inter-American Development Bank 1999)

442288mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442299

2322-ASR 703 filename70304-torche

Figure 1 Income Inequality in Chile and Industrialized Countries Circa 2000

Note Jap = Japan Swe = Sweden Ita = Italy Ger = Germany Neth = Netherlands Fra = France Ire = Republicof Ireland Eng = England Source World Bank 2001

Figure 2 Gini Coefficient for Total Population and Excluding Wealthiest Decile Chile other Latin AmericanCountries and Unites States in 1998

Note Par = Paraguay Bra = Brazil Ecu = Ecuador Arg = Argentina Bol = Bolivia Ven = Venezuela Uru =Uruguay Source Inter-American Development Bank 1999

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Therefore Chile is unequal largely because theelite concentrates an extremely high proportionof the national income Across the noneliteclasses the distribution of resources is muchmore uniform

Inequality is not a new development inChilean society but rather has deep historicalroots Figure 3 presents the longest availableseries on inequality in Chile depicting earningsdistribution since 1957 in Santiago3 and show-ing persistently high inequality over the lasthalf century However there is significant vari-ation over time a short decline of inequality dur-ing the progressive administrations of themid-1960s and early 1970s an increase since themilitary regime took power in 1973 a peak in1984 and a small decline after the democratictransition to levels still higher than those thatpreceded the military regime

LLAATTEE IINNDDUUSSTTRRIIAALLIIZZAATTIIOONN MMAARRKKEETTRREEFFOORRMM AANNDD TTHHEE CCHHIILLEEAANN CCLLAASSSSSSTTRRUUCCTTUURREE

The ldquocommon social fluidityrdquo hypothesis assertsthat all international variation in mobility pat-terns is attributable to highly specific histori-cally formed national characteristics In theChilean case these factors may be related to themarket reform of the 1970s and 1980s and therapid economic growth since the late 1980sNational mobility studies suggest that coun-tries experiencing rapid industrialization may becharacterized by weaker sector barriers to mobil-ity specif ically barriers between the self-employed and employees and betweenagricultural and nonagricultural classes(Goldthorpe Yaish and Kraus 1997 and Yaish2004 for Israel Ishida Goldthorpe and Erikson1991 for Japan Park 2004 for Korea and Costa-Ribeiro 2003 for Brazil)

In addition two components of the Chileanmarket transformation may have contributed tothe weakness of mobility barriers across sectorsof the economy The first component is theagrarian reform and subsequent counterreformundertaken by the military regime The secondone is the opening of the economy to interna-tional trade and the retrenchment of the statewhich led to a decline in formal employment

443300mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Figure 3 Earnings Inequality in Greatest Santiago 1957ndash2000Source Larranaga 1999 and calculations by the author

3 Santiago the Chilean capital comprises aboutone-third of the countryrsquos population Assessmentby Chilean experts suggests that Santiago trendspresent an unbiased picture of trends at the nationallevel Although earnings inequality is not as com-prehensive a measure as total income inequality it isan adequate much more accurately measured proxy(Galbraith 2002)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

and to the growth and segmentation of the self-employed sector

The main objective of the Chilean agrarianreform (1962ndash1973) was to reduce the extremeland concentration in the country The reformwas very successful In 1955 the top 7 percentof landowners held 65 percent of the land andthe bottom 37 percent held only 1 percent By1973 the last year of the reform 43 percent ofthe land had been expropriated (Kay 2002 Scott1996) An agrarian counterreform was launchedby the military regime as a way of returning theland to its former owners It did not howeverrestore the traditional hacienda order (Gomezand Echenique 1988 Rivera 1988) In fact themilitary government returned only one-third ofthe plots to their old owners sold another thirdand adjudicated the remaining third to the smallproprietors benefited by the agrarian reformUnable to compete with the now cheap foodimports many small proprietors opted to selltheir plots As a consequence as much as two-thirds of the Chilean agricultural land came upfor sale creating an active land market Themain beneficiaries of the ldquomarketization of thecountrysiderdquo were a new group of export-ori-ented entrepreneurs and international investorswho bought large tracts of land (Gwynne 1996Kay 2002) Thus the reform and counterreformtriggered a massive ldquochange of handsrdquo in theChilean countryside likely altering the patternsof land inheritance

The second major effect of the market reformon the class structure was the reduction of for-mal employment caused by the decline of theindustrial working class and by the shrinkage ofthe state Incapable of competing with the nowcheap imports the working class plummetedfrom 34 percent of total employment in theearly 1970s to 20 percent in 1980 while pub-lic employment dropped from 14 to 84 per-cent in the same period (Leon and Martinez2000 Schkolnick 2000 Velasquez 1990) Thedecline in industrial and public employmentled to an increase in self-employment fromabout 15 percent in 1970 to about 28 percent in1980 (Thomas 1996) followed by a declinethrough the economic recovery before stabi-lization at about 22 percent of the total employ-ment in 2000 (MIDEPLAN 2001) In contrastto industrialized countries in which the self-employment rate usually is within the one-digitrange and involves capital ownership in Chile

it encompasses more than one-fifth of the laborforce Furthermore self-employment hadbecome a survival strategy for a large numberof former industrial and public employeesappropriately labeled ldquoforced entrepreneurial-ismrdquo (Infante and Klein 1995 Portes Castellsand Benton 1989) As a consequence theChilean self-employed sector is voluminousand segmented It includes a small segment ofentrepreneurs oriented to capital accumulationand able to hire employees and a large seg-ment of self-employed workers mostly engagedin ldquosurvival activitiesrdquo and unable to hire thelabor of others Differences in economic well-being between these two groups are massivewith the former group earning on averagethree times more than the latter (MIDEPLAN2001)

To be sure a large self-employed sector is nota result of the market transformation Self-employment has been historically high in LatinAmerica because according to some scholarsthe region has experienced ldquodependent inte-grationrdquo into the world capitalistic system(Hopkins and Wallerstein 1982 Luxembourg1951) However the internal heterogeneity ofthe self-employed sector increased during themarket reform in a pattern that foretold whathappened in the rest of Latin America during the1990s (Klein and Tokman 2000 Portes andHoffman 2003) and what may be current devel-opments in the industrialized world (Arum andMuumlller 2004 Noorderhaven et al 2003) Thesecharacteristics of the Chilean class structureare considered in the comparative analysis ofChilean mobility

DDAATTAA AANNDD AANNAALLYYTTIICCAALL AAPPPPRROOAACCHH

This study uses the 2001 Chilean MobilitySurvey (CMS) The CMS is a nationally repre-sentative multistage stratified sample of maleheads of household ages 24 to 69 The samplingstrategy includes the following stages First 87primary sampling units (PSUs) (counties) areselected Then blocks within the PSUs areselected and finally households within blocksare chosen Counties are stratified according tosize (fewer than 20000 20000 to 100000100000 to 200000 and more than 200000inhabitants) and by geographic zone (NorthCenter South) All PSUs in the large size stra-tum are included in the sample to increase effi-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443311

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ciency The fieldwork conducted between Apriland June 2001 consists of face-to-face inter-views in the respondentrsquos household carried outby trained personnel The survey excludesnonndashhead-of-household males which represent17 percent of the male population of the rele-vant age (MIDEPLAN 2001) Among thoseexcluded 865 percent are the sons of the headof the household 124 percent are another rel-ative of the head and 11 percent are other non-related males Their occupational distributionwith control used for age is almost identical tothat of the heads of household The small pro-portion represented by this group and their sim-ilar occupational distribution suggest that theirinclusion would not significantly alter the find-ings presented in this article Excluding thehouseholds not eligible for the survey theresponse rate is 63 percent Although nonre-sponse rates usually are not reported in Chileansurveys exchange with Chilean experts indi-cates that the nonresponse rate is about 20 to 25percent for face-to-face household surveys Thehigher nonresponse rate of the CMS is likelyattributable to the difficulty contacting maleheads of household4 The total sample size is3544 I exclude individuals outside the agerange of 25 to 64 years which is conventional-ly used in comparative mobility research andcases with unusable data After this exclusionthe usable sample size is 3002

CCLLAASSSS SSCCHHEEMMAA

This study uses a class perspective to describethe Chilean social structure It utilizes the seven-category version of the classification devel-oped by the CASMIN project Thisclassification is widely used in internationalcomparative research and describes the basic

stratification of advanced industrial societiesbased on ldquoemployment relationshipsrdquo (Eriksonand Goldthorpe 1992a35ndash47 Goldthorpe andHeath 1992) The seven classes included in theschema are I+II (service class) III (routine non-manual workers) IVab (self-employed work-ers) V+VI (manual supervisors and skilledmanual workers) VIIa (unskilled manual work-ers) IVc (farmers) and VIIb (farm workers)

Use of the 7-class categorization instead ofan alternative detailed 12-class version pro-duced by the CASMIN project (Erikson andGoldthorpe 1992b) induces some loss of infor-mation about the originndashdestination associa-tion A global test of all five aggregationsobtained from the ratio of L2 tests for the inde-pendence model in the collapsed and full tablesindicates that the seven-class schema masks 29percent of the association shown by the full setof classes ([L2 full table ndash L2 collapsed table]L2 full table = [108644 ndash 77191]108644 =29) This proportion is high compared withthat of the CASMIN countries and is surpassedonly by Sweden with 31 percent (Hout andHauser 1992 Table 2) A collapsed sevenfoldclass schema was however preferred because ofthe moderate sample size of the Chilean surveyand because it grants international comparabil-ity of the findings5

Operationalization of this class classifica-tion is based on detailed information concern-ing job title (recoded into the standard ISCO-88classification [ILO 1990]) industry occupa-tional status and supervisory status of workersBecause the CMS is specifically designed toassess class mobility it includes all the infor-mation necessary to produce the CASMIN classschema The CASMIN project has not produceda standard algorithm to generate the class clas-sification but other researchers have producedsuch an algorithm which I use in this analysis(Ganzeboom and Treiman 2003) Table 1 pres-ents the basic intergenerational mobility tableto be used in this analysis including the countsand marginal percentage distribution for originand destination classes

443322mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

4 The nonresponse rate can yield bias if those whowere unreachable or refused to participate differ fromthose included in the sample To estimate the mag-nitude of this bias I compared the distribution of keyvariables with the CASEN 2000 survey CASEN isa large survey (n = 252595) conducted by the Chileangovernment and has a refusal rate lower than 10percent The comparison (available upon request)suggests that the CMS slightly underrepresents agri-cultural workers and the upper class Overall how-ever there is no indication of major nonresponsebias

5 Analyses were replicated using the detailed 12-class table and no difference in substantive find-ings was found

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

MMEETTHHOODDSS

I fit two alternative models to the Chilean mobil-ity table The first is the ldquocore model of social flu-idityrdquo which claims to capture basic internationalsimilarity in mobility Although the core modelhas become a benchmark for mobility researchand currently is the only model that provides acomparative framework for mobility analysisanalysts have drawn attention to a number of itslimitations The model has been criticized forinadequately representing hierarchical mobilityeffects for being tailored to a few industrializedcountries for the ad hoc nature of some effects itincludes (specifically the affinity effects) andfor lacking a criterion to determine how muchdivergence is necessary for it to be rejected(Ganzeboom Luijkx and Treiman 1989 Houtand Hauser 1992 Sorensen 1992 Yamaguchi1987) I therefore also test an alternative ldquohybridmodelrdquo that combines the rowndashcolumn (II) asso-ciation model (Goodman 1979) with parametersaccounting for class immobility The internationaland temporal comparison of Chilean mobilitypatterns uses the uniform-difference (UNIDIFF)model also known as the multiplicative layereffect model (Erikson and Goldthorpe 1992aXie 1992) Model comparison and selection isbased on the Bayesian Information Criterion(BIC) statistic (Raftery 1995)

AANNAALLYYSSIISS OOFF CCHHIILLEEAANN MMOOBBIILLIITTYY IINNCCOOMMPPAARRAATTIIVVEE PPEERRSSPPEECCTTIIVVEE

TTHHEE CCOORREE MMOODDEELL OOFF SSOOCCIIAALL FFLLUUIIDDIITTYY

I start by fitting the core model to the Chileanmobility table The core model represents the

basic international commonality in social flu-idity using a topological formulation In contrastto standard topological models constructed froma single allocation of cells (Hauser 1978) thecore model uses eight matrices each of whichis designed to capture a particular effect thatenhances or reduces mobility between specificclasses (Erikson and Goldthorpe 1987a 1992achapter 4) In this way the model permits thesources of commonality across nations to beidentified clearly and likewise the departuresfrom it The effects are of four types hierarchyinheritance sector and affinity6

HIERARCHY EFFECTS Hierarchy effects reflectthe impact of status distances between classeson fluidity between them To estimate theseeffects the mobility table is divided into threestrata reflecting differences in resources andrewards across classes The strata are the fol-lowing upper stratum (class I+II) middle stra-tum (classes III IVab IVc and V-VI) and lowerstratum (classes VIIa and VIIb) Because thereare three strata two hierarchy effects (HI1 andHI2) can be identified each representing thecrossing of one additional hierarchical barrierThe association between each pair of classes isexpressed as an inverse function of the numberof hierarchical strata crossed

INHERITANCE EFFECTS Inheritance effects aredesigned to capture the propensity for class

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443333

2322-ASR 703 filename70304-torche

Table 1 Seven-Class Intergenerational Mobility Table for Chilean Men

Class of Destination

I-II III IVab IVc V-VI VIIa VIIb N

I-II Service Class 229 49 81 2 37 30 2 430 14III Routine Nonmanual 61 15 34 0 24 27 1 162 5IVab Self-employed 123 37 173 28 102 90 20 573 19IVc Farmers 42 16 68 54 58 54 24 316 11V-VI Skilled Manual 91 33 116 13 155 135 36 579 19VIIa Semiskilled and Unskilled Manual 49 43 115 13 111 123 44 498 17VIIb Farm Workers 28 16 78 17 89 101 115 444 15N 623 209 665 127 576 560 242 3002 21 7 22 4 19 19 8

Note Cell entries represent counts

6 Design matrices in Erikson and Goldthorpe(1992a124ndash129)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

immobility They include three matrices IN1identifies immobility across all classes alike IN2identifies the higher immobility of the serviceclass (I+II) the self-employed group (IVab) andfarmers (IVc) and IN3 accounts for the highestimmobility of farmers (IVc)

SECTOR EFFECTS Sector effects are intended toidentify the difficulty of moving between theagricultural and nonagricultural sectors of theeconomy (SE1)

AFFINITY EFFECTS Affinity effects are intend-ed to capture specific discontinuities (negativeaffinities) or linkages (positive affinities) betweenclasses which either reinforce or offset the over-all effects of hierarchy and sector There are twoaffinity effects The first AF1 identifies the dif-ficulty of moving between the service class (I+II)and the class of farm workers (VIIb) which addsto their hierarchical distance In contrast AF2identifies instances in which mobility is morefrequent than accounted for by hierarchy and sec-tor effects and includes affinities within the non-manual sector (I+II and III) and within the manualsector (V+VI and VIIa) a symmetrical affinitybased on capital possession (IVab and IVc and Iand IVab) and an asymmetrical link between agri-cultural classes of origin (IVc and VIIb) and theunskilled manual class of destination (VIIa)7

TTHHEE MMOOBBIILLIITTYY PPAATTTTEERRNN IINN CCHHIILLEECCOONNVVEERRGGEENNCCEE OORR DDIIVVEERRGGEENNCCEE

Table 2 presents the fit of several models appliedto the Chilean data and Table 3 displays theparameter estimates for each of these modelsModel 1 in Table 2 indicates that the fit of theoriginal core model to the Chilean table is unsat-isfactory under standard statistical criteria butit explains a significant 775 percent of associ-ation under independence To assess the strengthof the different factors driving mobility oppor-tunities I compare the Chilean coefficients withthose obtained for the CASMIN countriesCoefficients from the original core model fit-ted to the Chilean table are reported in row 1 ofTable 3 CASMIN coefficients are reported inrow 2 and differences in coefficientsrsquo magni-tude and the statistical significance of the dif-ferences are reported in row 38

All Chilean coefficients are significant andhave the same sign as for the CASMIN coun-tries Comparison of the magnitude of the coef-ficients however shows significant departuresfrom the core for five effects which revealinteresting differences between Chile and theCASMIN countries The hierarchy coefficientsindicate that whereas the short-range hierar-chical barriers are somewhat weaker in Chile

443344mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

8 CASMIN coefficients were obtained from fittingthe core model to the combined table of France andEngland standardizing both tables to have 10000cases as recommended by Erikson and Goldthorpe(1992a121) The coefficients obtained are virtuallyidentical to those presented by Erikson andGoldthorpe (1992a Table 44)

Table 2 Fit statistics selected models to the Chilean Mobility Table

AssociationL2 BIC df p Explained

Core Modelmdash1 Core Model Original 1375 285 28 000 775mdash2 Core Model EHE 1081 ndash08 28 000 860mdash3 Chilean version of Core Model ([Model 2] + Chilean Parameters) 724 ndash326 27 000 906Association Models 000mdash4 Heterogeneous Quasi-RC(II) Model 734 ndash161 23 000 905mdash5 Homogeneous Quasi-RC(II) Model 803 ndash287 28 000 896mdash6 Quasi-Linear by Linear Association (SES row and column ranking)a 1563 279 33 000 798

Note EHE = empirical hierarchical effects SES = eocioeconomic statusa Linear-by-linear association model using socioeconomic status scores as presented in Appendix Figure A1 torank origin and destination classes

7 For a detailed account of the empirical derivationof the model see Erikson and Goldthorpe (1987a1987b and 1992a chapter 4)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443355

2322-ASR 703 filename70304-torche

Tab

le 3

M

odel

Par

amet

er E

stim

ates

for

the

Chi

lean

Mob

ilit

y Ta

ble

Eff

ect

1 C

ore

mod

el o

rigi

nala

mdash 2 C

ore

mod

el C

AS

MIN

b

mdash 3 D

iffe

renc

e C

ASM

INmdash

Chi

lec

mdash 4 C

ore

mod

el E

HE

d

mdash 5 C

hile

an v

ersi

on o

f co

ree

mdash Not

eD

ata

show

n ar

e m

odel

par

amet

er e

stim

ates

wit

h st

anda

rd e

rror

s in

par

enth

eses

EH

E =

em

piri

cal h

iera

rchi

cal e

ffec

ts

aM

odel

1 f

rom

Tab

le 2

b

Ori

gina

l Cor

e M

odel

fit

ted

to th

e E

ngli

sh a

nd F

renc

h m

obil

ity

tabl

es S

ee te

xt f

or d

etai

ls

cD

iffe

renc

e in

par

amet

er m

agni

tude

bet

wee

n or

igin

al c

ore

mod

el f

itte

d to

the

Chi

lean

tabl

e (r

ow 1

) an

d to

the

CA

SM

IN ta

ble

(row

2)

dM

odel

2 f

rom

Tab

le 2

e

Mod

el 3

fro

m T

able

2

dagger p

lt 1

p

lt 0

5

p

lt 0

1

p

lt 0

01 (

two-

tail

ed)

HI1

ndash12

(

05)

ndash22

(02

)ndash

10 dagger

(06

)ndash

03(

07)

ndash09

dagger(

06)

HI2

ndash65

(11

)ndash

45

(

04)

20

dagger(

12)

ndash74

(09

)ndash

47

(

10)

IN1

41

(

08)

43

(

03)

Insi

g

37

(

07)

34

(

07)

IN2

39

(

10)

84

(

04)

45

(

11)

31

(13

)mdash

IN3

68

(

24)

101

(11

)In

sig

61

(24

)4

8(

23)

SE

1

ndash38

(07

)ndash1

06

(

05)

ndash68

(08

)ndash

42

(

07)

ndash20

(

08)

AF

1

ndash76

(21

)ndash

83

(

13)

Insi

g

ndash84

(20

)ndash

63

(21

)

AF

2

33

(

05)

45

(

02)

ndash12

(

06)

20

(

06)

26

(

06)

AF

1ndashC

mdash mdash mdash mdash

ndash14

9

(46

)

IN2ndash

C

mdash mdash mdash mdash 71

(

14)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(HI1) long-range hierarchical mobility is moredifficult in this country (HI2) This suggestssubstantial inequality between the extremes ofthe class hierarchy combined with smaller dif-ferentiation in the middle of the distribution Asto the inheritance effects the service class theself-employed group and the farmers (classesI+II IVab and IVc respectively) have a sig-nificantly lower propensity to immobility inChile than in the CASMIN countries suggest-ing that class positions usually associated withindependent work are less able to reproducetheir class status intergenerationally than depict-ed by the core Most impressive the barrierbetween the agricultural and nonagriculturalsectors of the economy (SE1) is much weakerin Chile as is the affinity effect linking non-manual classes manual classes and the class-es that own capital among themselves (AF2)

Interestingly all the significant differences incoefficients with the single exception of the bar-rier to long-range hierarchical mobility sug-gest that Chile is more fluid than the core modeldepicts Higher fluidity seems to be driven bythe weakness of horizontal barriers separatingthe agricultural manual and self-employed sec-tors of the economy This is consistent withfindings of weaker sector effects in countriesthat have experienced rapid industrializationsuch as Korea (Park 2004) Brazil (Costa-Ribeiro 2003) Israel (Yaish 2004 Goldthorpeet al 1997) and Japan (Ishida et al 1991)

Additionally in the Chilean case the weak-ness of sector barriers may have been intensi-fied by the deep market transformation of the1970s and 1980s On the one hand the ldquochangeof handsrdquo in the agricultural sector induced bythe agrarian reform and counterreform likelyincreased fluidity between the agricultural andurban classes On the other hand the role of self-employment as an ephemeral refuge against(industrial and public sector) unemploymentmay have reduced barriers between the self-employed group and the rest of the social struc-ture Added to the larger difficulty of long-rangehierarchical mobility these features suggestweak sector cleavages in the Chilean fluiditypattern and high salience of the hierarchicaldimension of mobility

For the countries in which the fit of the coremodel is not adequate Erikson and Goldthorpe(1992a145) suggest adding parameters thataccount for country-specific deviations from

the model I therefore introduce adjustments toreflect salient characteristics of the Chilean flu-idity pattern producing a ldquoChilean versionrdquo ofthe core model

First I modify the hierarchical effects Fittingof the core model suggests that hierarchicalbarriers are a crucial determinant of mobility inChile This conclusion is however obscuredby the fact that the hierarchical strata distin-guished by the core model are not empiricallyobtained and may not accurately represent theChilean ranking of classes in terms of socioe-conomic status (SES) To model hierarchicaleffects in Chile adequately I rank classes usingthe unweighted average of their schooling andearning levels as a proxy for SES I then collapsethe seven classes into three hierarchical stratausing cluster analysis These three empiricallyobtained strata are the following Upper stratum(service class I+II) middle stratum (routinenonmanual and self-employed group III andIVab) and lower stratum (manual and agricul-tural classes V-VI VIIa IVc and VIIb) Detailson the ranking of classes cluster procedures andthe comparison of the CASMIN and empiricalChilean class ranking can be found in theAppendix Comparison of the CASMIN and theempirical class ranking suggests higher differ-entiation at the top than at the bottom of thesocial structure in the Chilean case

Model 2 in Table 2 uses the empiricallyobtained hierarchical strata (EHE) Theimprovement in fit is massive (the difference inBIC is 293 without using any degrees of free-dom) The model now accounts for 86 percentof the association under independence

The large improvement in fit when empiricalhierarchical strata are used challenges theassumption of an internationally homogeneoushierarchical class ranking At the empiricallevel there is no proof that the hierarchicalranking of classes assumed by the core modelfits all CASMIN countries9 Moreover someevidence points to significant cross-country dif-

443366mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

9 Erikson and Goldthorpe(1992a Table 22) pres-ent the average ranking of classes based on standardstatus measures but they do not show the empiricalranking of classes for each country If the occupa-tional composition of classes varies across countriesthere may be significant international differences inthe hierarchical ranking of classes

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ferences in the ranking of classes (Hout andHauser 1992 Sorensen 1992) At the theoreti-cal level the creators of the core model high-light the relevance of highly specific historicallyformed national attributes to explain interna-tional variation in mobility These national par-ticularities may very well result in differenthierarchical positions of classes across coun-tries10 Furthermore even if the homogeneousranking of classes imposed by the CASMINresearchers applied to most CASMIN coun-tries significant departures may be found inthe late industrialized countries currently beingadded to the comparative mobility project TheChilean findings suggest that instead of impos-ing international homogeneity to the hierarchi-cal dimension of mobility the ranking of classesshould be treated as nationally specific param-eters to be estimated as mobility researchersroutinely treat row and column marginals ofthe mobility table The issue is importantbecause the hierarchical dimension is the mostrelevant dimension of mobility (Hout andHauser 1992 Wong 1992) and inadequate mod-eling can bias assessment of not only hierar-chical effects of mobility but also other effects

Two other changes were introduced toaccount for Chilean-specific deviations fromthe core model First the IN2 effect wasrearranged to exclude the self-employed class(IVab) and to include the class of farm workers(VIIb) This rearrangement accounts for thefact that the Chilean self-employed classappeared less likely than depicted by the coremodel to reproduce its class position acrossgenerations and the inheritance level of Chileanfarm workers is higher than in the CASMINcountries These changes are expressed in theChilean-specific inheritance parameter IN2-CSecond an affinity parameter (AF1-C) wasadded to capture the difficulty of long-rangedownward elite mobility This negative asym-metrical affinity links flows from the service

class to the agricultural classes and from theroutine nonmanual class to the class of farmworkers Note that the AF1-C adds to the hier-archical distance between the two extremes ofthe class structure and to the negative affinitycaptured by AF1 to depict the extreme difficultyof long-range downward mobility from theChilean elite The fit of this lsquoChilean versionrsquoof the core model is reported in Table 2 model3 The improvement in BIC from model 2 is sub-stantial Furthermore even if the model fallsshort in terms of standard statistical criteria itaccounts for a large 91 of the associationunder independence thereby supporting theclaim that it adequately represents the Chileanmobility pattern The parameter estimates for theChilean version of the core are reported in row5 Table 3

Because of the core modelrsquos undesirable prop-erties I also use an alternative approach toexamine Chilean mobility patterns Thisapproach uses the now standard associationrowndashcolumn (II) (RC[II]) model (Goodman1979 see also Hout 1983 and Wong 1992) Inthis model both origin and destination classesare scaled so that the association can beexpressed as a linear-by-linear interaction by asingle parameter The multiplicative form ofthe model can be expressed as follows

Fij = iO j

D exp( microi νj)

where i indexes class of origin j indexes classof destination Fij is the expected frequency inthe (ij) cell is the grand mean i

O pertainsto the class of origin marginal effect j

D pertainsto the class of destination marginal effect isa global association parameter microi is the scalevalue for the ith class of origin and νj is the scalescore for the jth class of destination subject tothe following normalization constraints Σmicroi =Σνj = 0 (normalization of the location) and Σmicroi2

= Σνj2 = 1 (normalization of the scale)The class scale scores reflect a latent con-

tinuous variable made manifest by the class cat-egories Empirically obtained from the datathese ldquodistancerdquo scores produce an optimalranking of classes for the purpose of mobilityanalysis The RC(II) model is appropriate formodeling the association of any two ordinalvariables but mobility tables are distinctivebecause of the correspondence between class oforigin and class of destination (Gerber and Hout2004691)

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443377

2322-ASR 703 filename70304-torche

10 For instance during the socialist period inHungary the status of the proletariat may have beenindistinguishable from the status of the routine non-manual class (Szelenyi 199862) In Scotland givento early industrialization the hierarchical barrierbetween the skilled and unskilled working classes thatcharacterizes other CASMIN countries may not exist(Erikson and Goldthorpe 1992a163)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I introduce two adjustments to account forthis correspondence transforming the formu-lation into a ldquohybrid modelrdquo First topologicalparameters will account for the large counts insome diagonal cells reflecting immobility(inheritance) effects These parameters distin-guish two inheritance levels ldquohigh inheritancerdquofor the service class (I+II) farmers (IVc) andfarm workers (VIIb) and ldquolow inheritancerdquo forall other classes with the exception of the rou-tine nonmanual workers

Model 4 in Table 2 reports the fit statistic forthis quasi-RC(II) model The BIC statisticrejects it and prefers the more parsimoniousChilean version of the core model ThereforeI introduce a second adjustment to account forthe correspondence between rows and columnsin mobility tables Origin and destination scoresare constrained to be the same assuming thatthe scaling of classes has not significantlychanged over time Model 5 in Table 2 reportsfit statistics for this homogeneous quasi-RC(II)model On the basis of the BIC statistic the fitof the model is as good as that of the Chileanversion of the core model (Wong [1994] showsthat BIC differences of less than 5 points shouldbe considered indeterminate) Thus under stan-dard statistical criteria the Chilean core modeland the homogeneous quasi-RC(II) model areindistinguishable Given that the Chilean versionof the core model was tailored to fit the Chileantable the fit of model 5 is impressive and it indi-cates that a unidimensional scale can adequatelycapture mobility distances between classes

The logical next question is what this scalerepresents and specifically whether it repre-sents the hierarchical ranking of classes in termsof SES If a close correspondence is foundbetween the mobility distances among classesand their distances in terms of SES this find-ing would suggest that the intragenerationaldistribution of resources and rewards expressedin the SES ranking of classes drives the processof intergenerational mobility In other words thisfinding would support the resource approachrsquosclaim that differences in resources associatedwith diverse class positions determine mobili-ty opportunities across generations

To examine this possibility I compare theclass scores obtained from the homogeneousquasi-RC(II) model with the empiricallyobtained SES ranking of classes based on edu-cation and earnings (Appendix) Figure 4 pres-

ents the comparison between the standardizedvalues of both sets of scores The correspon-dence is striking The mobility distancesbetween classes closely reproduce their dis-tances in terms of SES This close isomorphismsuggests that at least in Chile the intragener-ational inequality in the distribution of socialresources and rewards across classes largelydrives the intergenerational mobility process Totest the isomorphism further I use the SESclass scores as row and column rankings to esti-mate a quasindashlinear-by-linear association model(model 6 Table 2) The fit of model 6 is sig-nificantly worse than that of model 5 almost aspoor as that of the original core model This indi-cates that although the inter- and intragenera-tional dimensions of inequality are very similarin Chile they have a somewhat different struc-ture

In summary the central characteristic of theChilean mobility regime is the predominance oflong-range hierarchical barriers and the weak-ness of sector cleavages separating classes in themiddle and at the lower end of the hierarchicalstructure Findings from the ldquoChilean versionrdquoof the core model and the quasi-RC(II) hybridmodels are consistent depicting a social struc-ture characterized by signif icant barriersbetween the top echelon and the rest of the classstructure The good fit of the homogeneousquasi-RC(II) model provides preliminary sup-port for the resource approach linking inequal-ity and mobility Mobility dynamics seem to bedriven largely by the hierarchical distancesbetween classes and overall Chile seems not tobe less fluid than the CASMIN countries

CCOOMMPPAARRIISSOONN OOFF TTHHEE OORRIIGGIINNAALL AANNDD CCHHIILLEEAANN

VVEERRSSIIOONNSS OOFF TTHHEE CCOORREE MMOODDEELL

Although Chilean mobility dynamics seem to bedriven by inequality across classes in a veryunequal society the fitting of the core modelprovides no indication that Chile is significantlyless fluid than depicted by the core model Asa preliminary analysis of the Chilean level of flu-idity I compare the predicted propensities formobility and immobility between specific pairsof classes in Chile with those in the CASMINcountries Propensities are obtained from theoriginal and Chilean versions of the core modelrespectively (models 2 and 5 of Table 3)

443388mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I begin with the immobility of those at the topof the class structure namely the service class(I+II) The service classrsquo propensity for immo-bility in the original core model is captured bytwo parameters IN1 and IN2 This propensityis more than three times what it would be in theabsence of these effects (e43+84 = 356) InChile the propensity for immobility of class I+IIis captured by the same two parameters and itis lower than what the core model expresses286 times larger than it would be in the absenceof these effects (e34+71) What about the immo-bility of the two agricultural classes In theCASMIN countries immobility of the IVc andVIIb classes is respectively 785 and 154 timeswhat it would be in the absence of these param-eters (e43+84+101ndash22 and e43 respectively) InChile the comparable values are 422 and 286(e34+71+48ndash09 and e43+71 respectively) indi-cating the relatively low inheritance associatedwith land ownership in the Latin Americannation Thus if attention is focused on theimmobility in the upper and lower ends of thesocial hierarchy the comparison does not indi-cate less fluidity in the highly unequal Chileansociety

What about the mobility rates between thetwo most distant classes in the hierarchical rank-ing namely the service class (I+II) and farmworkers (VIIb) In the case of the CASMINcountries the mobility chances between theseclasses are captured by four parameters (HI1HI2 SE1 and AF1) and is 077 times smallerthan it would be in the absence of these effects(endash22 ndash45ndash106ndash83) In Chile the distance fromclass I to class VIIb is expressed by these samefour parameters in addition to the asymmetri-cal disaffinity AF1-C This yields mobilitychances 056 times what would be obtained inthe absence of these effects (endash09 ndash47ndash20ndash63ndash149)Thus the barrier to long-range downwardmobility from the elite is more significant inChile than in the CASMIN countries The sameoccurs with long-range downward mobility fromclass III

If we measure the reciprocal flow (ieupward mobility from class VIIb to class I+II)we obtain the same value for the CASMINcountries because all parameters are symmet-rical in the original core model For Chile how-ever this calculation removes the asymmetricalbarrier to elite downward mobility which rais-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443399

2322-ASR 703 filename70304-torche

Figure 4 Class Scores Obtained from the Homogeneous Quasi-RC(II) Model and from Socioeconomic Status(SES) Ranking

Note (1) = Class standardized socioeconomic status scores based on schooling and earnings (2) = Class stan-dardized scores based on the Homogeneous Quasi-RC(II) model (see text for details)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 7: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

ldquoRevolution in Freedomrdquo (Gazmuri 2000)This progressive government launched redis-tributive policies including an agrarian reformand an educational reform The redistributiveagenda boosted by a socialist administrationthat came to power in 1970 was abruptly halt-ed by a military coup in 1973 The militarytook power and retained it until 1990 Duringthese 17 years the military regime conduct-ed a deep market-oriented transformationNow turned into the ldquoWashington consensusrdquoparadigm (Williamson 1990) this reformincluded macroeconomic stabilization dereg-ulation of prices and markets and the priva-tization of enterprises and social services Ittransformed Chile from a closed economywith heavy state intervention into one of themost open market-based economies in theworld with the productive and welfare role ofthe state reduced to a minimum (Edwards andCox-Edwards 1991 Martinez and Diaz 1999Meller 1996 Velasco 1994)

The depth of the market reform coupledwith a world recession led in the late 1970sand early 1980s to the deepest economic cri-sis since the Great Depression A third of thelabor force was unemployed and povertyafflicted nearly half of Chilean households(Meller 1991) The post-crisis recovery start-ing in the late 1980s was substantial and sus-tained and coincided with theredemocratization of the country The grossdomestic product (GDP) per capita grew morethan 6 percent annually for 15 years a novelrate for Chile only comparable in recent timesto the ldquoEast Asian Miraclerdquo (World Bank1993) The sharp economic growth has trans-formed Chile from one of the poorest coun-tries in Latin America (Hofman 2000) into aldquomiddle-income economyrdquo (World Bank2003a) In the year 2000 the income per capi-ta was approximately US$5000 much lowerthan the United States average of US$31910but the highest in Latin America (World Bank2003a) As a consequence of the sharp eco-nomic growth Chileans have reached levels ofconsumption unthinkable two decades agoand the poverty rate fell from 45 to 21 percentbetween 1985 and 2000 (MIDEPLAN 2000Raczynski 2000)

TTHHEE CCHHIILLEEAANN LLEEVVEELL AANNDD PPAATTTTEERRNN OOFF

IINNEEQQUUAALLIITTYY

The dark side of this ldquosuccess storyrdquo is eco-nomic inequality Structurally rooted in a feu-dal agrarian structure the institutional legacyof the colonial period and in the slow expan-sion of education (Engerman and Sokoloff1997 World Bank 2003b) inequality hasremained persistently high during the 20thcentury The Gini coefficient reached 58 in the1990s which compares with a much lowerGini of 34 among the industrialized coun-tries and is large even in the highly unequalLatin American context with its average Giniof 49 (Deininger and Squire 1996 Marceland Solimano 1994) As Figure 1 indicatesChilean inequality is almost twice that in mostindustrialized countries and 15 times that inthe United States the most unequal nation ofthe industrialized world

Not only the level but also the pattern ofinequality in Chile significantly departs fromthat of the industrialized world With thewealthiest Chilean decile receiving 423 per-cent of the total national income (MIDEPLAN2001) the Chilean pattern of inequality ischaracterized by high ldquoconcentration at thetoprdquo Although inequality is by definition relat-ed to concentration the Chilean case isextreme as compared with the industrializedworld and even with other Latin Americancountries A comparison between the incomeof each decile and the income of the preced-ing (poorer) decile illustrates the point Theratio between the wealthiest and the secondwealthiest decile is twice as large in Chile asin the United States and England and one ofthe largest in Latin America depicting highelite concentration In contrast the ratiobetween the second poorest and the poorestdeciles in Chile is half that of the United Statesand England indicating that inequality at thebottom of the income distribution is muchlower in Chile than in these industrializednations (Szekely and Hilgert 1999)

In fact as Figure 2 indicates Chile is thefourth most unequal country in the mostunequal region of the world However if thewealthiest decile is excluded Chilean inequal-ity is dramatically reduced and Chile becomesthe most equal Latin American country evenmore equal than the United States (Inter-American Development Bank 1999)

442288mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442299

2322-ASR 703 filename70304-torche

Figure 1 Income Inequality in Chile and Industrialized Countries Circa 2000

Note Jap = Japan Swe = Sweden Ita = Italy Ger = Germany Neth = Netherlands Fra = France Ire = Republicof Ireland Eng = England Source World Bank 2001

Figure 2 Gini Coefficient for Total Population and Excluding Wealthiest Decile Chile other Latin AmericanCountries and Unites States in 1998

Note Par = Paraguay Bra = Brazil Ecu = Ecuador Arg = Argentina Bol = Bolivia Ven = Venezuela Uru =Uruguay Source Inter-American Development Bank 1999

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Therefore Chile is unequal largely because theelite concentrates an extremely high proportionof the national income Across the noneliteclasses the distribution of resources is muchmore uniform

Inequality is not a new development inChilean society but rather has deep historicalroots Figure 3 presents the longest availableseries on inequality in Chile depicting earningsdistribution since 1957 in Santiago3 and show-ing persistently high inequality over the lasthalf century However there is significant vari-ation over time a short decline of inequality dur-ing the progressive administrations of themid-1960s and early 1970s an increase since themilitary regime took power in 1973 a peak in1984 and a small decline after the democratictransition to levels still higher than those thatpreceded the military regime

LLAATTEE IINNDDUUSSTTRRIIAALLIIZZAATTIIOONN MMAARRKKEETTRREEFFOORRMM AANNDD TTHHEE CCHHIILLEEAANN CCLLAASSSSSSTTRRUUCCTTUURREE

The ldquocommon social fluidityrdquo hypothesis assertsthat all international variation in mobility pat-terns is attributable to highly specific histori-cally formed national characteristics In theChilean case these factors may be related to themarket reform of the 1970s and 1980s and therapid economic growth since the late 1980sNational mobility studies suggest that coun-tries experiencing rapid industrialization may becharacterized by weaker sector barriers to mobil-ity specif ically barriers between the self-employed and employees and betweenagricultural and nonagricultural classes(Goldthorpe Yaish and Kraus 1997 and Yaish2004 for Israel Ishida Goldthorpe and Erikson1991 for Japan Park 2004 for Korea and Costa-Ribeiro 2003 for Brazil)

In addition two components of the Chileanmarket transformation may have contributed tothe weakness of mobility barriers across sectorsof the economy The first component is theagrarian reform and subsequent counterreformundertaken by the military regime The secondone is the opening of the economy to interna-tional trade and the retrenchment of the statewhich led to a decline in formal employment

443300mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Figure 3 Earnings Inequality in Greatest Santiago 1957ndash2000Source Larranaga 1999 and calculations by the author

3 Santiago the Chilean capital comprises aboutone-third of the countryrsquos population Assessmentby Chilean experts suggests that Santiago trendspresent an unbiased picture of trends at the nationallevel Although earnings inequality is not as com-prehensive a measure as total income inequality it isan adequate much more accurately measured proxy(Galbraith 2002)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

and to the growth and segmentation of the self-employed sector

The main objective of the Chilean agrarianreform (1962ndash1973) was to reduce the extremeland concentration in the country The reformwas very successful In 1955 the top 7 percentof landowners held 65 percent of the land andthe bottom 37 percent held only 1 percent By1973 the last year of the reform 43 percent ofthe land had been expropriated (Kay 2002 Scott1996) An agrarian counterreform was launchedby the military regime as a way of returning theland to its former owners It did not howeverrestore the traditional hacienda order (Gomezand Echenique 1988 Rivera 1988) In fact themilitary government returned only one-third ofthe plots to their old owners sold another thirdand adjudicated the remaining third to the smallproprietors benefited by the agrarian reformUnable to compete with the now cheap foodimports many small proprietors opted to selltheir plots As a consequence as much as two-thirds of the Chilean agricultural land came upfor sale creating an active land market Themain beneficiaries of the ldquomarketization of thecountrysiderdquo were a new group of export-ori-ented entrepreneurs and international investorswho bought large tracts of land (Gwynne 1996Kay 2002) Thus the reform and counterreformtriggered a massive ldquochange of handsrdquo in theChilean countryside likely altering the patternsof land inheritance

The second major effect of the market reformon the class structure was the reduction of for-mal employment caused by the decline of theindustrial working class and by the shrinkage ofthe state Incapable of competing with the nowcheap imports the working class plummetedfrom 34 percent of total employment in theearly 1970s to 20 percent in 1980 while pub-lic employment dropped from 14 to 84 per-cent in the same period (Leon and Martinez2000 Schkolnick 2000 Velasquez 1990) Thedecline in industrial and public employmentled to an increase in self-employment fromabout 15 percent in 1970 to about 28 percent in1980 (Thomas 1996) followed by a declinethrough the economic recovery before stabi-lization at about 22 percent of the total employ-ment in 2000 (MIDEPLAN 2001) In contrastto industrialized countries in which the self-employment rate usually is within the one-digitrange and involves capital ownership in Chile

it encompasses more than one-fifth of the laborforce Furthermore self-employment hadbecome a survival strategy for a large numberof former industrial and public employeesappropriately labeled ldquoforced entrepreneurial-ismrdquo (Infante and Klein 1995 Portes Castellsand Benton 1989) As a consequence theChilean self-employed sector is voluminousand segmented It includes a small segment ofentrepreneurs oriented to capital accumulationand able to hire employees and a large seg-ment of self-employed workers mostly engagedin ldquosurvival activitiesrdquo and unable to hire thelabor of others Differences in economic well-being between these two groups are massivewith the former group earning on averagethree times more than the latter (MIDEPLAN2001)

To be sure a large self-employed sector is nota result of the market transformation Self-employment has been historically high in LatinAmerica because according to some scholarsthe region has experienced ldquodependent inte-grationrdquo into the world capitalistic system(Hopkins and Wallerstein 1982 Luxembourg1951) However the internal heterogeneity ofthe self-employed sector increased during themarket reform in a pattern that foretold whathappened in the rest of Latin America during the1990s (Klein and Tokman 2000 Portes andHoffman 2003) and what may be current devel-opments in the industrialized world (Arum andMuumlller 2004 Noorderhaven et al 2003) Thesecharacteristics of the Chilean class structureare considered in the comparative analysis ofChilean mobility

DDAATTAA AANNDD AANNAALLYYTTIICCAALL AAPPPPRROOAACCHH

This study uses the 2001 Chilean MobilitySurvey (CMS) The CMS is a nationally repre-sentative multistage stratified sample of maleheads of household ages 24 to 69 The samplingstrategy includes the following stages First 87primary sampling units (PSUs) (counties) areselected Then blocks within the PSUs areselected and finally households within blocksare chosen Counties are stratified according tosize (fewer than 20000 20000 to 100000100000 to 200000 and more than 200000inhabitants) and by geographic zone (NorthCenter South) All PSUs in the large size stra-tum are included in the sample to increase effi-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443311

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ciency The fieldwork conducted between Apriland June 2001 consists of face-to-face inter-views in the respondentrsquos household carried outby trained personnel The survey excludesnonndashhead-of-household males which represent17 percent of the male population of the rele-vant age (MIDEPLAN 2001) Among thoseexcluded 865 percent are the sons of the headof the household 124 percent are another rel-ative of the head and 11 percent are other non-related males Their occupational distributionwith control used for age is almost identical tothat of the heads of household The small pro-portion represented by this group and their sim-ilar occupational distribution suggest that theirinclusion would not significantly alter the find-ings presented in this article Excluding thehouseholds not eligible for the survey theresponse rate is 63 percent Although nonre-sponse rates usually are not reported in Chileansurveys exchange with Chilean experts indi-cates that the nonresponse rate is about 20 to 25percent for face-to-face household surveys Thehigher nonresponse rate of the CMS is likelyattributable to the difficulty contacting maleheads of household4 The total sample size is3544 I exclude individuals outside the agerange of 25 to 64 years which is conventional-ly used in comparative mobility research andcases with unusable data After this exclusionthe usable sample size is 3002

CCLLAASSSS SSCCHHEEMMAA

This study uses a class perspective to describethe Chilean social structure It utilizes the seven-category version of the classification devel-oped by the CASMIN project Thisclassification is widely used in internationalcomparative research and describes the basic

stratification of advanced industrial societiesbased on ldquoemployment relationshipsrdquo (Eriksonand Goldthorpe 1992a35ndash47 Goldthorpe andHeath 1992) The seven classes included in theschema are I+II (service class) III (routine non-manual workers) IVab (self-employed work-ers) V+VI (manual supervisors and skilledmanual workers) VIIa (unskilled manual work-ers) IVc (farmers) and VIIb (farm workers)

Use of the 7-class categorization instead ofan alternative detailed 12-class version pro-duced by the CASMIN project (Erikson andGoldthorpe 1992b) induces some loss of infor-mation about the originndashdestination associa-tion A global test of all five aggregationsobtained from the ratio of L2 tests for the inde-pendence model in the collapsed and full tablesindicates that the seven-class schema masks 29percent of the association shown by the full setof classes ([L2 full table ndash L2 collapsed table]L2 full table = [108644 ndash 77191]108644 =29) This proportion is high compared withthat of the CASMIN countries and is surpassedonly by Sweden with 31 percent (Hout andHauser 1992 Table 2) A collapsed sevenfoldclass schema was however preferred because ofthe moderate sample size of the Chilean surveyand because it grants international comparabil-ity of the findings5

Operationalization of this class classifica-tion is based on detailed information concern-ing job title (recoded into the standard ISCO-88classification [ILO 1990]) industry occupa-tional status and supervisory status of workersBecause the CMS is specifically designed toassess class mobility it includes all the infor-mation necessary to produce the CASMIN classschema The CASMIN project has not produceda standard algorithm to generate the class clas-sification but other researchers have producedsuch an algorithm which I use in this analysis(Ganzeboom and Treiman 2003) Table 1 pres-ents the basic intergenerational mobility tableto be used in this analysis including the countsand marginal percentage distribution for originand destination classes

443322mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

4 The nonresponse rate can yield bias if those whowere unreachable or refused to participate differ fromthose included in the sample To estimate the mag-nitude of this bias I compared the distribution of keyvariables with the CASEN 2000 survey CASEN isa large survey (n = 252595) conducted by the Chileangovernment and has a refusal rate lower than 10percent The comparison (available upon request)suggests that the CMS slightly underrepresents agri-cultural workers and the upper class Overall how-ever there is no indication of major nonresponsebias

5 Analyses were replicated using the detailed 12-class table and no difference in substantive find-ings was found

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

MMEETTHHOODDSS

I fit two alternative models to the Chilean mobil-ity table The first is the ldquocore model of social flu-idityrdquo which claims to capture basic internationalsimilarity in mobility Although the core modelhas become a benchmark for mobility researchand currently is the only model that provides acomparative framework for mobility analysisanalysts have drawn attention to a number of itslimitations The model has been criticized forinadequately representing hierarchical mobilityeffects for being tailored to a few industrializedcountries for the ad hoc nature of some effects itincludes (specifically the affinity effects) andfor lacking a criterion to determine how muchdivergence is necessary for it to be rejected(Ganzeboom Luijkx and Treiman 1989 Houtand Hauser 1992 Sorensen 1992 Yamaguchi1987) I therefore also test an alternative ldquohybridmodelrdquo that combines the rowndashcolumn (II) asso-ciation model (Goodman 1979) with parametersaccounting for class immobility The internationaland temporal comparison of Chilean mobilitypatterns uses the uniform-difference (UNIDIFF)model also known as the multiplicative layereffect model (Erikson and Goldthorpe 1992aXie 1992) Model comparison and selection isbased on the Bayesian Information Criterion(BIC) statistic (Raftery 1995)

AANNAALLYYSSIISS OOFF CCHHIILLEEAANN MMOOBBIILLIITTYY IINNCCOOMMPPAARRAATTIIVVEE PPEERRSSPPEECCTTIIVVEE

TTHHEE CCOORREE MMOODDEELL OOFF SSOOCCIIAALL FFLLUUIIDDIITTYY

I start by fitting the core model to the Chileanmobility table The core model represents the

basic international commonality in social flu-idity using a topological formulation In contrastto standard topological models constructed froma single allocation of cells (Hauser 1978) thecore model uses eight matrices each of whichis designed to capture a particular effect thatenhances or reduces mobility between specificclasses (Erikson and Goldthorpe 1987a 1992achapter 4) In this way the model permits thesources of commonality across nations to beidentified clearly and likewise the departuresfrom it The effects are of four types hierarchyinheritance sector and affinity6

HIERARCHY EFFECTS Hierarchy effects reflectthe impact of status distances between classeson fluidity between them To estimate theseeffects the mobility table is divided into threestrata reflecting differences in resources andrewards across classes The strata are the fol-lowing upper stratum (class I+II) middle stra-tum (classes III IVab IVc and V-VI) and lowerstratum (classes VIIa and VIIb) Because thereare three strata two hierarchy effects (HI1 andHI2) can be identified each representing thecrossing of one additional hierarchical barrierThe association between each pair of classes isexpressed as an inverse function of the numberof hierarchical strata crossed

INHERITANCE EFFECTS Inheritance effects aredesigned to capture the propensity for class

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443333

2322-ASR 703 filename70304-torche

Table 1 Seven-Class Intergenerational Mobility Table for Chilean Men

Class of Destination

I-II III IVab IVc V-VI VIIa VIIb N

I-II Service Class 229 49 81 2 37 30 2 430 14III Routine Nonmanual 61 15 34 0 24 27 1 162 5IVab Self-employed 123 37 173 28 102 90 20 573 19IVc Farmers 42 16 68 54 58 54 24 316 11V-VI Skilled Manual 91 33 116 13 155 135 36 579 19VIIa Semiskilled and Unskilled Manual 49 43 115 13 111 123 44 498 17VIIb Farm Workers 28 16 78 17 89 101 115 444 15N 623 209 665 127 576 560 242 3002 21 7 22 4 19 19 8

Note Cell entries represent counts

6 Design matrices in Erikson and Goldthorpe(1992a124ndash129)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

immobility They include three matrices IN1identifies immobility across all classes alike IN2identifies the higher immobility of the serviceclass (I+II) the self-employed group (IVab) andfarmers (IVc) and IN3 accounts for the highestimmobility of farmers (IVc)

SECTOR EFFECTS Sector effects are intended toidentify the difficulty of moving between theagricultural and nonagricultural sectors of theeconomy (SE1)

AFFINITY EFFECTS Affinity effects are intend-ed to capture specific discontinuities (negativeaffinities) or linkages (positive affinities) betweenclasses which either reinforce or offset the over-all effects of hierarchy and sector There are twoaffinity effects The first AF1 identifies the dif-ficulty of moving between the service class (I+II)and the class of farm workers (VIIb) which addsto their hierarchical distance In contrast AF2identifies instances in which mobility is morefrequent than accounted for by hierarchy and sec-tor effects and includes affinities within the non-manual sector (I+II and III) and within the manualsector (V+VI and VIIa) a symmetrical affinitybased on capital possession (IVab and IVc and Iand IVab) and an asymmetrical link between agri-cultural classes of origin (IVc and VIIb) and theunskilled manual class of destination (VIIa)7

TTHHEE MMOOBBIILLIITTYY PPAATTTTEERRNN IINN CCHHIILLEECCOONNVVEERRGGEENNCCEE OORR DDIIVVEERRGGEENNCCEE

Table 2 presents the fit of several models appliedto the Chilean data and Table 3 displays theparameter estimates for each of these modelsModel 1 in Table 2 indicates that the fit of theoriginal core model to the Chilean table is unsat-isfactory under standard statistical criteria butit explains a significant 775 percent of associ-ation under independence To assess the strengthof the different factors driving mobility oppor-tunities I compare the Chilean coefficients withthose obtained for the CASMIN countriesCoefficients from the original core model fit-ted to the Chilean table are reported in row 1 ofTable 3 CASMIN coefficients are reported inrow 2 and differences in coefficientsrsquo magni-tude and the statistical significance of the dif-ferences are reported in row 38

All Chilean coefficients are significant andhave the same sign as for the CASMIN coun-tries Comparison of the magnitude of the coef-ficients however shows significant departuresfrom the core for five effects which revealinteresting differences between Chile and theCASMIN countries The hierarchy coefficientsindicate that whereas the short-range hierar-chical barriers are somewhat weaker in Chile

443344mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

8 CASMIN coefficients were obtained from fittingthe core model to the combined table of France andEngland standardizing both tables to have 10000cases as recommended by Erikson and Goldthorpe(1992a121) The coefficients obtained are virtuallyidentical to those presented by Erikson andGoldthorpe (1992a Table 44)

Table 2 Fit statistics selected models to the Chilean Mobility Table

AssociationL2 BIC df p Explained

Core Modelmdash1 Core Model Original 1375 285 28 000 775mdash2 Core Model EHE 1081 ndash08 28 000 860mdash3 Chilean version of Core Model ([Model 2] + Chilean Parameters) 724 ndash326 27 000 906Association Models 000mdash4 Heterogeneous Quasi-RC(II) Model 734 ndash161 23 000 905mdash5 Homogeneous Quasi-RC(II) Model 803 ndash287 28 000 896mdash6 Quasi-Linear by Linear Association (SES row and column ranking)a 1563 279 33 000 798

Note EHE = empirical hierarchical effects SES = eocioeconomic statusa Linear-by-linear association model using socioeconomic status scores as presented in Appendix Figure A1 torank origin and destination classes

7 For a detailed account of the empirical derivationof the model see Erikson and Goldthorpe (1987a1987b and 1992a chapter 4)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443355

2322-ASR 703 filename70304-torche

Tab

le 3

M

odel

Par

amet

er E

stim

ates

for

the

Chi

lean

Mob

ilit

y Ta

ble

Eff

ect

1 C

ore

mod

el o

rigi

nala

mdash 2 C

ore

mod

el C

AS

MIN

b

mdash 3 D

iffe

renc

e C

ASM

INmdash

Chi

lec

mdash 4 C

ore

mod

el E

HE

d

mdash 5 C

hile

an v

ersi

on o

f co

ree

mdash Not

eD

ata

show

n ar

e m

odel

par

amet

er e

stim

ates

wit

h st

anda

rd e

rror

s in

par

enth

eses

EH

E =

em

piri

cal h

iera

rchi

cal e

ffec

ts

aM

odel

1 f

rom

Tab

le 2

b

Ori

gina

l Cor

e M

odel

fit

ted

to th

e E

ngli

sh a

nd F

renc

h m

obil

ity

tabl

es S

ee te

xt f

or d

etai

ls

cD

iffe

renc

e in

par

amet

er m

agni

tude

bet

wee

n or

igin

al c

ore

mod

el f

itte

d to

the

Chi

lean

tabl

e (r

ow 1

) an

d to

the

CA

SM

IN ta

ble

(row

2)

dM

odel

2 f

rom

Tab

le 2

e

Mod

el 3

fro

m T

able

2

dagger p

lt 1

p

lt 0

5

p

lt 0

1

p

lt 0

01 (

two-

tail

ed)

HI1

ndash12

(

05)

ndash22

(02

)ndash

10 dagger

(06

)ndash

03(

07)

ndash09

dagger(

06)

HI2

ndash65

(11

)ndash

45

(

04)

20

dagger(

12)

ndash74

(09

)ndash

47

(

10)

IN1

41

(

08)

43

(

03)

Insi

g

37

(

07)

34

(

07)

IN2

39

(

10)

84

(

04)

45

(

11)

31

(13

)mdash

IN3

68

(

24)

101

(11

)In

sig

61

(24

)4

8(

23)

SE

1

ndash38

(07

)ndash1

06

(

05)

ndash68

(08

)ndash

42

(

07)

ndash20

(

08)

AF

1

ndash76

(21

)ndash

83

(

13)

Insi

g

ndash84

(20

)ndash

63

(21

)

AF

2

33

(

05)

45

(

02)

ndash12

(

06)

20

(

06)

26

(

06)

AF

1ndashC

mdash mdash mdash mdash

ndash14

9

(46

)

IN2ndash

C

mdash mdash mdash mdash 71

(

14)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(HI1) long-range hierarchical mobility is moredifficult in this country (HI2) This suggestssubstantial inequality between the extremes ofthe class hierarchy combined with smaller dif-ferentiation in the middle of the distribution Asto the inheritance effects the service class theself-employed group and the farmers (classesI+II IVab and IVc respectively) have a sig-nificantly lower propensity to immobility inChile than in the CASMIN countries suggest-ing that class positions usually associated withindependent work are less able to reproducetheir class status intergenerationally than depict-ed by the core Most impressive the barrierbetween the agricultural and nonagriculturalsectors of the economy (SE1) is much weakerin Chile as is the affinity effect linking non-manual classes manual classes and the class-es that own capital among themselves (AF2)

Interestingly all the significant differences incoefficients with the single exception of the bar-rier to long-range hierarchical mobility sug-gest that Chile is more fluid than the core modeldepicts Higher fluidity seems to be driven bythe weakness of horizontal barriers separatingthe agricultural manual and self-employed sec-tors of the economy This is consistent withfindings of weaker sector effects in countriesthat have experienced rapid industrializationsuch as Korea (Park 2004) Brazil (Costa-Ribeiro 2003) Israel (Yaish 2004 Goldthorpeet al 1997) and Japan (Ishida et al 1991)

Additionally in the Chilean case the weak-ness of sector barriers may have been intensi-fied by the deep market transformation of the1970s and 1980s On the one hand the ldquochangeof handsrdquo in the agricultural sector induced bythe agrarian reform and counterreform likelyincreased fluidity between the agricultural andurban classes On the other hand the role of self-employment as an ephemeral refuge against(industrial and public sector) unemploymentmay have reduced barriers between the self-employed group and the rest of the social struc-ture Added to the larger difficulty of long-rangehierarchical mobility these features suggestweak sector cleavages in the Chilean fluiditypattern and high salience of the hierarchicaldimension of mobility

For the countries in which the fit of the coremodel is not adequate Erikson and Goldthorpe(1992a145) suggest adding parameters thataccount for country-specific deviations from

the model I therefore introduce adjustments toreflect salient characteristics of the Chilean flu-idity pattern producing a ldquoChilean versionrdquo ofthe core model

First I modify the hierarchical effects Fittingof the core model suggests that hierarchicalbarriers are a crucial determinant of mobility inChile This conclusion is however obscuredby the fact that the hierarchical strata distin-guished by the core model are not empiricallyobtained and may not accurately represent theChilean ranking of classes in terms of socioe-conomic status (SES) To model hierarchicaleffects in Chile adequately I rank classes usingthe unweighted average of their schooling andearning levels as a proxy for SES I then collapsethe seven classes into three hierarchical stratausing cluster analysis These three empiricallyobtained strata are the following Upper stratum(service class I+II) middle stratum (routinenonmanual and self-employed group III andIVab) and lower stratum (manual and agricul-tural classes V-VI VIIa IVc and VIIb) Detailson the ranking of classes cluster procedures andthe comparison of the CASMIN and empiricalChilean class ranking can be found in theAppendix Comparison of the CASMIN and theempirical class ranking suggests higher differ-entiation at the top than at the bottom of thesocial structure in the Chilean case

Model 2 in Table 2 uses the empiricallyobtained hierarchical strata (EHE) Theimprovement in fit is massive (the difference inBIC is 293 without using any degrees of free-dom) The model now accounts for 86 percentof the association under independence

The large improvement in fit when empiricalhierarchical strata are used challenges theassumption of an internationally homogeneoushierarchical class ranking At the empiricallevel there is no proof that the hierarchicalranking of classes assumed by the core modelfits all CASMIN countries9 Moreover someevidence points to significant cross-country dif-

443366mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

9 Erikson and Goldthorpe(1992a Table 22) pres-ent the average ranking of classes based on standardstatus measures but they do not show the empiricalranking of classes for each country If the occupa-tional composition of classes varies across countriesthere may be significant international differences inthe hierarchical ranking of classes

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ferences in the ranking of classes (Hout andHauser 1992 Sorensen 1992) At the theoreti-cal level the creators of the core model high-light the relevance of highly specific historicallyformed national attributes to explain interna-tional variation in mobility These national par-ticularities may very well result in differenthierarchical positions of classes across coun-tries10 Furthermore even if the homogeneousranking of classes imposed by the CASMINresearchers applied to most CASMIN coun-tries significant departures may be found inthe late industrialized countries currently beingadded to the comparative mobility project TheChilean findings suggest that instead of impos-ing international homogeneity to the hierarchi-cal dimension of mobility the ranking of classesshould be treated as nationally specific param-eters to be estimated as mobility researchersroutinely treat row and column marginals ofthe mobility table The issue is importantbecause the hierarchical dimension is the mostrelevant dimension of mobility (Hout andHauser 1992 Wong 1992) and inadequate mod-eling can bias assessment of not only hierar-chical effects of mobility but also other effects

Two other changes were introduced toaccount for Chilean-specific deviations fromthe core model First the IN2 effect wasrearranged to exclude the self-employed class(IVab) and to include the class of farm workers(VIIb) This rearrangement accounts for thefact that the Chilean self-employed classappeared less likely than depicted by the coremodel to reproduce its class position acrossgenerations and the inheritance level of Chileanfarm workers is higher than in the CASMINcountries These changes are expressed in theChilean-specific inheritance parameter IN2-CSecond an affinity parameter (AF1-C) wasadded to capture the difficulty of long-rangedownward elite mobility This negative asym-metrical affinity links flows from the service

class to the agricultural classes and from theroutine nonmanual class to the class of farmworkers Note that the AF1-C adds to the hier-archical distance between the two extremes ofthe class structure and to the negative affinitycaptured by AF1 to depict the extreme difficultyof long-range downward mobility from theChilean elite The fit of this lsquoChilean versionrsquoof the core model is reported in Table 2 model3 The improvement in BIC from model 2 is sub-stantial Furthermore even if the model fallsshort in terms of standard statistical criteria itaccounts for a large 91 of the associationunder independence thereby supporting theclaim that it adequately represents the Chileanmobility pattern The parameter estimates for theChilean version of the core are reported in row5 Table 3

Because of the core modelrsquos undesirable prop-erties I also use an alternative approach toexamine Chilean mobility patterns Thisapproach uses the now standard associationrowndashcolumn (II) (RC[II]) model (Goodman1979 see also Hout 1983 and Wong 1992) Inthis model both origin and destination classesare scaled so that the association can beexpressed as a linear-by-linear interaction by asingle parameter The multiplicative form ofthe model can be expressed as follows

Fij = iO j

D exp( microi νj)

where i indexes class of origin j indexes classof destination Fij is the expected frequency inthe (ij) cell is the grand mean i

O pertainsto the class of origin marginal effect j

D pertainsto the class of destination marginal effect isa global association parameter microi is the scalevalue for the ith class of origin and νj is the scalescore for the jth class of destination subject tothe following normalization constraints Σmicroi =Σνj = 0 (normalization of the location) and Σmicroi2

= Σνj2 = 1 (normalization of the scale)The class scale scores reflect a latent con-

tinuous variable made manifest by the class cat-egories Empirically obtained from the datathese ldquodistancerdquo scores produce an optimalranking of classes for the purpose of mobilityanalysis The RC(II) model is appropriate formodeling the association of any two ordinalvariables but mobility tables are distinctivebecause of the correspondence between class oforigin and class of destination (Gerber and Hout2004691)

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443377

2322-ASR 703 filename70304-torche

10 For instance during the socialist period inHungary the status of the proletariat may have beenindistinguishable from the status of the routine non-manual class (Szelenyi 199862) In Scotland givento early industrialization the hierarchical barrierbetween the skilled and unskilled working classes thatcharacterizes other CASMIN countries may not exist(Erikson and Goldthorpe 1992a163)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I introduce two adjustments to account forthis correspondence transforming the formu-lation into a ldquohybrid modelrdquo First topologicalparameters will account for the large counts insome diagonal cells reflecting immobility(inheritance) effects These parameters distin-guish two inheritance levels ldquohigh inheritancerdquofor the service class (I+II) farmers (IVc) andfarm workers (VIIb) and ldquolow inheritancerdquo forall other classes with the exception of the rou-tine nonmanual workers

Model 4 in Table 2 reports the fit statistic forthis quasi-RC(II) model The BIC statisticrejects it and prefers the more parsimoniousChilean version of the core model ThereforeI introduce a second adjustment to account forthe correspondence between rows and columnsin mobility tables Origin and destination scoresare constrained to be the same assuming thatthe scaling of classes has not significantlychanged over time Model 5 in Table 2 reportsfit statistics for this homogeneous quasi-RC(II)model On the basis of the BIC statistic the fitof the model is as good as that of the Chileanversion of the core model (Wong [1994] showsthat BIC differences of less than 5 points shouldbe considered indeterminate) Thus under stan-dard statistical criteria the Chilean core modeland the homogeneous quasi-RC(II) model areindistinguishable Given that the Chilean versionof the core model was tailored to fit the Chileantable the fit of model 5 is impressive and it indi-cates that a unidimensional scale can adequatelycapture mobility distances between classes

The logical next question is what this scalerepresents and specifically whether it repre-sents the hierarchical ranking of classes in termsof SES If a close correspondence is foundbetween the mobility distances among classesand their distances in terms of SES this find-ing would suggest that the intragenerationaldistribution of resources and rewards expressedin the SES ranking of classes drives the processof intergenerational mobility In other words thisfinding would support the resource approachrsquosclaim that differences in resources associatedwith diverse class positions determine mobili-ty opportunities across generations

To examine this possibility I compare theclass scores obtained from the homogeneousquasi-RC(II) model with the empiricallyobtained SES ranking of classes based on edu-cation and earnings (Appendix) Figure 4 pres-

ents the comparison between the standardizedvalues of both sets of scores The correspon-dence is striking The mobility distancesbetween classes closely reproduce their dis-tances in terms of SES This close isomorphismsuggests that at least in Chile the intragener-ational inequality in the distribution of socialresources and rewards across classes largelydrives the intergenerational mobility process Totest the isomorphism further I use the SESclass scores as row and column rankings to esti-mate a quasindashlinear-by-linear association model(model 6 Table 2) The fit of model 6 is sig-nificantly worse than that of model 5 almost aspoor as that of the original core model This indi-cates that although the inter- and intragenera-tional dimensions of inequality are very similarin Chile they have a somewhat different struc-ture

In summary the central characteristic of theChilean mobility regime is the predominance oflong-range hierarchical barriers and the weak-ness of sector cleavages separating classes in themiddle and at the lower end of the hierarchicalstructure Findings from the ldquoChilean versionrdquoof the core model and the quasi-RC(II) hybridmodels are consistent depicting a social struc-ture characterized by signif icant barriersbetween the top echelon and the rest of the classstructure The good fit of the homogeneousquasi-RC(II) model provides preliminary sup-port for the resource approach linking inequal-ity and mobility Mobility dynamics seem to bedriven largely by the hierarchical distancesbetween classes and overall Chile seems not tobe less fluid than the CASMIN countries

CCOOMMPPAARRIISSOONN OOFF TTHHEE OORRIIGGIINNAALL AANNDD CCHHIILLEEAANN

VVEERRSSIIOONNSS OOFF TTHHEE CCOORREE MMOODDEELL

Although Chilean mobility dynamics seem to bedriven by inequality across classes in a veryunequal society the fitting of the core modelprovides no indication that Chile is significantlyless fluid than depicted by the core model Asa preliminary analysis of the Chilean level of flu-idity I compare the predicted propensities formobility and immobility between specific pairsof classes in Chile with those in the CASMINcountries Propensities are obtained from theoriginal and Chilean versions of the core modelrespectively (models 2 and 5 of Table 3)

443388mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I begin with the immobility of those at the topof the class structure namely the service class(I+II) The service classrsquo propensity for immo-bility in the original core model is captured bytwo parameters IN1 and IN2 This propensityis more than three times what it would be in theabsence of these effects (e43+84 = 356) InChile the propensity for immobility of class I+IIis captured by the same two parameters and itis lower than what the core model expresses286 times larger than it would be in the absenceof these effects (e34+71) What about the immo-bility of the two agricultural classes In theCASMIN countries immobility of the IVc andVIIb classes is respectively 785 and 154 timeswhat it would be in the absence of these param-eters (e43+84+101ndash22 and e43 respectively) InChile the comparable values are 422 and 286(e34+71+48ndash09 and e43+71 respectively) indi-cating the relatively low inheritance associatedwith land ownership in the Latin Americannation Thus if attention is focused on theimmobility in the upper and lower ends of thesocial hierarchy the comparison does not indi-cate less fluidity in the highly unequal Chileansociety

What about the mobility rates between thetwo most distant classes in the hierarchical rank-ing namely the service class (I+II) and farmworkers (VIIb) In the case of the CASMINcountries the mobility chances between theseclasses are captured by four parameters (HI1HI2 SE1 and AF1) and is 077 times smallerthan it would be in the absence of these effects(endash22 ndash45ndash106ndash83) In Chile the distance fromclass I to class VIIb is expressed by these samefour parameters in addition to the asymmetri-cal disaffinity AF1-C This yields mobilitychances 056 times what would be obtained inthe absence of these effects (endash09 ndash47ndash20ndash63ndash149)Thus the barrier to long-range downwardmobility from the elite is more significant inChile than in the CASMIN countries The sameoccurs with long-range downward mobility fromclass III

If we measure the reciprocal flow (ieupward mobility from class VIIb to class I+II)we obtain the same value for the CASMINcountries because all parameters are symmet-rical in the original core model For Chile how-ever this calculation removes the asymmetricalbarrier to elite downward mobility which rais-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443399

2322-ASR 703 filename70304-torche

Figure 4 Class Scores Obtained from the Homogeneous Quasi-RC(II) Model and from Socioeconomic Status(SES) Ranking

Note (1) = Class standardized socioeconomic status scores based on schooling and earnings (2) = Class stan-dardized scores based on the Homogeneous Quasi-RC(II) model (see text for details)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 8: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash442299

2322-ASR 703 filename70304-torche

Figure 1 Income Inequality in Chile and Industrialized Countries Circa 2000

Note Jap = Japan Swe = Sweden Ita = Italy Ger = Germany Neth = Netherlands Fra = France Ire = Republicof Ireland Eng = England Source World Bank 2001

Figure 2 Gini Coefficient for Total Population and Excluding Wealthiest Decile Chile other Latin AmericanCountries and Unites States in 1998

Note Par = Paraguay Bra = Brazil Ecu = Ecuador Arg = Argentina Bol = Bolivia Ven = Venezuela Uru =Uruguay Source Inter-American Development Bank 1999

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Therefore Chile is unequal largely because theelite concentrates an extremely high proportionof the national income Across the noneliteclasses the distribution of resources is muchmore uniform

Inequality is not a new development inChilean society but rather has deep historicalroots Figure 3 presents the longest availableseries on inequality in Chile depicting earningsdistribution since 1957 in Santiago3 and show-ing persistently high inequality over the lasthalf century However there is significant vari-ation over time a short decline of inequality dur-ing the progressive administrations of themid-1960s and early 1970s an increase since themilitary regime took power in 1973 a peak in1984 and a small decline after the democratictransition to levels still higher than those thatpreceded the military regime

LLAATTEE IINNDDUUSSTTRRIIAALLIIZZAATTIIOONN MMAARRKKEETTRREEFFOORRMM AANNDD TTHHEE CCHHIILLEEAANN CCLLAASSSSSSTTRRUUCCTTUURREE

The ldquocommon social fluidityrdquo hypothesis assertsthat all international variation in mobility pat-terns is attributable to highly specific histori-cally formed national characteristics In theChilean case these factors may be related to themarket reform of the 1970s and 1980s and therapid economic growth since the late 1980sNational mobility studies suggest that coun-tries experiencing rapid industrialization may becharacterized by weaker sector barriers to mobil-ity specif ically barriers between the self-employed and employees and betweenagricultural and nonagricultural classes(Goldthorpe Yaish and Kraus 1997 and Yaish2004 for Israel Ishida Goldthorpe and Erikson1991 for Japan Park 2004 for Korea and Costa-Ribeiro 2003 for Brazil)

In addition two components of the Chileanmarket transformation may have contributed tothe weakness of mobility barriers across sectorsof the economy The first component is theagrarian reform and subsequent counterreformundertaken by the military regime The secondone is the opening of the economy to interna-tional trade and the retrenchment of the statewhich led to a decline in formal employment

443300mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Figure 3 Earnings Inequality in Greatest Santiago 1957ndash2000Source Larranaga 1999 and calculations by the author

3 Santiago the Chilean capital comprises aboutone-third of the countryrsquos population Assessmentby Chilean experts suggests that Santiago trendspresent an unbiased picture of trends at the nationallevel Although earnings inequality is not as com-prehensive a measure as total income inequality it isan adequate much more accurately measured proxy(Galbraith 2002)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

and to the growth and segmentation of the self-employed sector

The main objective of the Chilean agrarianreform (1962ndash1973) was to reduce the extremeland concentration in the country The reformwas very successful In 1955 the top 7 percentof landowners held 65 percent of the land andthe bottom 37 percent held only 1 percent By1973 the last year of the reform 43 percent ofthe land had been expropriated (Kay 2002 Scott1996) An agrarian counterreform was launchedby the military regime as a way of returning theland to its former owners It did not howeverrestore the traditional hacienda order (Gomezand Echenique 1988 Rivera 1988) In fact themilitary government returned only one-third ofthe plots to their old owners sold another thirdand adjudicated the remaining third to the smallproprietors benefited by the agrarian reformUnable to compete with the now cheap foodimports many small proprietors opted to selltheir plots As a consequence as much as two-thirds of the Chilean agricultural land came upfor sale creating an active land market Themain beneficiaries of the ldquomarketization of thecountrysiderdquo were a new group of export-ori-ented entrepreneurs and international investorswho bought large tracts of land (Gwynne 1996Kay 2002) Thus the reform and counterreformtriggered a massive ldquochange of handsrdquo in theChilean countryside likely altering the patternsof land inheritance

The second major effect of the market reformon the class structure was the reduction of for-mal employment caused by the decline of theindustrial working class and by the shrinkage ofthe state Incapable of competing with the nowcheap imports the working class plummetedfrom 34 percent of total employment in theearly 1970s to 20 percent in 1980 while pub-lic employment dropped from 14 to 84 per-cent in the same period (Leon and Martinez2000 Schkolnick 2000 Velasquez 1990) Thedecline in industrial and public employmentled to an increase in self-employment fromabout 15 percent in 1970 to about 28 percent in1980 (Thomas 1996) followed by a declinethrough the economic recovery before stabi-lization at about 22 percent of the total employ-ment in 2000 (MIDEPLAN 2001) In contrastto industrialized countries in which the self-employment rate usually is within the one-digitrange and involves capital ownership in Chile

it encompasses more than one-fifth of the laborforce Furthermore self-employment hadbecome a survival strategy for a large numberof former industrial and public employeesappropriately labeled ldquoforced entrepreneurial-ismrdquo (Infante and Klein 1995 Portes Castellsand Benton 1989) As a consequence theChilean self-employed sector is voluminousand segmented It includes a small segment ofentrepreneurs oriented to capital accumulationand able to hire employees and a large seg-ment of self-employed workers mostly engagedin ldquosurvival activitiesrdquo and unable to hire thelabor of others Differences in economic well-being between these two groups are massivewith the former group earning on averagethree times more than the latter (MIDEPLAN2001)

To be sure a large self-employed sector is nota result of the market transformation Self-employment has been historically high in LatinAmerica because according to some scholarsthe region has experienced ldquodependent inte-grationrdquo into the world capitalistic system(Hopkins and Wallerstein 1982 Luxembourg1951) However the internal heterogeneity ofthe self-employed sector increased during themarket reform in a pattern that foretold whathappened in the rest of Latin America during the1990s (Klein and Tokman 2000 Portes andHoffman 2003) and what may be current devel-opments in the industrialized world (Arum andMuumlller 2004 Noorderhaven et al 2003) Thesecharacteristics of the Chilean class structureare considered in the comparative analysis ofChilean mobility

DDAATTAA AANNDD AANNAALLYYTTIICCAALL AAPPPPRROOAACCHH

This study uses the 2001 Chilean MobilitySurvey (CMS) The CMS is a nationally repre-sentative multistage stratified sample of maleheads of household ages 24 to 69 The samplingstrategy includes the following stages First 87primary sampling units (PSUs) (counties) areselected Then blocks within the PSUs areselected and finally households within blocksare chosen Counties are stratified according tosize (fewer than 20000 20000 to 100000100000 to 200000 and more than 200000inhabitants) and by geographic zone (NorthCenter South) All PSUs in the large size stra-tum are included in the sample to increase effi-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443311

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ciency The fieldwork conducted between Apriland June 2001 consists of face-to-face inter-views in the respondentrsquos household carried outby trained personnel The survey excludesnonndashhead-of-household males which represent17 percent of the male population of the rele-vant age (MIDEPLAN 2001) Among thoseexcluded 865 percent are the sons of the headof the household 124 percent are another rel-ative of the head and 11 percent are other non-related males Their occupational distributionwith control used for age is almost identical tothat of the heads of household The small pro-portion represented by this group and their sim-ilar occupational distribution suggest that theirinclusion would not significantly alter the find-ings presented in this article Excluding thehouseholds not eligible for the survey theresponse rate is 63 percent Although nonre-sponse rates usually are not reported in Chileansurveys exchange with Chilean experts indi-cates that the nonresponse rate is about 20 to 25percent for face-to-face household surveys Thehigher nonresponse rate of the CMS is likelyattributable to the difficulty contacting maleheads of household4 The total sample size is3544 I exclude individuals outside the agerange of 25 to 64 years which is conventional-ly used in comparative mobility research andcases with unusable data After this exclusionthe usable sample size is 3002

CCLLAASSSS SSCCHHEEMMAA

This study uses a class perspective to describethe Chilean social structure It utilizes the seven-category version of the classification devel-oped by the CASMIN project Thisclassification is widely used in internationalcomparative research and describes the basic

stratification of advanced industrial societiesbased on ldquoemployment relationshipsrdquo (Eriksonand Goldthorpe 1992a35ndash47 Goldthorpe andHeath 1992) The seven classes included in theschema are I+II (service class) III (routine non-manual workers) IVab (self-employed work-ers) V+VI (manual supervisors and skilledmanual workers) VIIa (unskilled manual work-ers) IVc (farmers) and VIIb (farm workers)

Use of the 7-class categorization instead ofan alternative detailed 12-class version pro-duced by the CASMIN project (Erikson andGoldthorpe 1992b) induces some loss of infor-mation about the originndashdestination associa-tion A global test of all five aggregationsobtained from the ratio of L2 tests for the inde-pendence model in the collapsed and full tablesindicates that the seven-class schema masks 29percent of the association shown by the full setof classes ([L2 full table ndash L2 collapsed table]L2 full table = [108644 ndash 77191]108644 =29) This proportion is high compared withthat of the CASMIN countries and is surpassedonly by Sweden with 31 percent (Hout andHauser 1992 Table 2) A collapsed sevenfoldclass schema was however preferred because ofthe moderate sample size of the Chilean surveyand because it grants international comparabil-ity of the findings5

Operationalization of this class classifica-tion is based on detailed information concern-ing job title (recoded into the standard ISCO-88classification [ILO 1990]) industry occupa-tional status and supervisory status of workersBecause the CMS is specifically designed toassess class mobility it includes all the infor-mation necessary to produce the CASMIN classschema The CASMIN project has not produceda standard algorithm to generate the class clas-sification but other researchers have producedsuch an algorithm which I use in this analysis(Ganzeboom and Treiman 2003) Table 1 pres-ents the basic intergenerational mobility tableto be used in this analysis including the countsand marginal percentage distribution for originand destination classes

443322mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

4 The nonresponse rate can yield bias if those whowere unreachable or refused to participate differ fromthose included in the sample To estimate the mag-nitude of this bias I compared the distribution of keyvariables with the CASEN 2000 survey CASEN isa large survey (n = 252595) conducted by the Chileangovernment and has a refusal rate lower than 10percent The comparison (available upon request)suggests that the CMS slightly underrepresents agri-cultural workers and the upper class Overall how-ever there is no indication of major nonresponsebias

5 Analyses were replicated using the detailed 12-class table and no difference in substantive find-ings was found

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

MMEETTHHOODDSS

I fit two alternative models to the Chilean mobil-ity table The first is the ldquocore model of social flu-idityrdquo which claims to capture basic internationalsimilarity in mobility Although the core modelhas become a benchmark for mobility researchand currently is the only model that provides acomparative framework for mobility analysisanalysts have drawn attention to a number of itslimitations The model has been criticized forinadequately representing hierarchical mobilityeffects for being tailored to a few industrializedcountries for the ad hoc nature of some effects itincludes (specifically the affinity effects) andfor lacking a criterion to determine how muchdivergence is necessary for it to be rejected(Ganzeboom Luijkx and Treiman 1989 Houtand Hauser 1992 Sorensen 1992 Yamaguchi1987) I therefore also test an alternative ldquohybridmodelrdquo that combines the rowndashcolumn (II) asso-ciation model (Goodman 1979) with parametersaccounting for class immobility The internationaland temporal comparison of Chilean mobilitypatterns uses the uniform-difference (UNIDIFF)model also known as the multiplicative layereffect model (Erikson and Goldthorpe 1992aXie 1992) Model comparison and selection isbased on the Bayesian Information Criterion(BIC) statistic (Raftery 1995)

AANNAALLYYSSIISS OOFF CCHHIILLEEAANN MMOOBBIILLIITTYY IINNCCOOMMPPAARRAATTIIVVEE PPEERRSSPPEECCTTIIVVEE

TTHHEE CCOORREE MMOODDEELL OOFF SSOOCCIIAALL FFLLUUIIDDIITTYY

I start by fitting the core model to the Chileanmobility table The core model represents the

basic international commonality in social flu-idity using a topological formulation In contrastto standard topological models constructed froma single allocation of cells (Hauser 1978) thecore model uses eight matrices each of whichis designed to capture a particular effect thatenhances or reduces mobility between specificclasses (Erikson and Goldthorpe 1987a 1992achapter 4) In this way the model permits thesources of commonality across nations to beidentified clearly and likewise the departuresfrom it The effects are of four types hierarchyinheritance sector and affinity6

HIERARCHY EFFECTS Hierarchy effects reflectthe impact of status distances between classeson fluidity between them To estimate theseeffects the mobility table is divided into threestrata reflecting differences in resources andrewards across classes The strata are the fol-lowing upper stratum (class I+II) middle stra-tum (classes III IVab IVc and V-VI) and lowerstratum (classes VIIa and VIIb) Because thereare three strata two hierarchy effects (HI1 andHI2) can be identified each representing thecrossing of one additional hierarchical barrierThe association between each pair of classes isexpressed as an inverse function of the numberof hierarchical strata crossed

INHERITANCE EFFECTS Inheritance effects aredesigned to capture the propensity for class

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443333

2322-ASR 703 filename70304-torche

Table 1 Seven-Class Intergenerational Mobility Table for Chilean Men

Class of Destination

I-II III IVab IVc V-VI VIIa VIIb N

I-II Service Class 229 49 81 2 37 30 2 430 14III Routine Nonmanual 61 15 34 0 24 27 1 162 5IVab Self-employed 123 37 173 28 102 90 20 573 19IVc Farmers 42 16 68 54 58 54 24 316 11V-VI Skilled Manual 91 33 116 13 155 135 36 579 19VIIa Semiskilled and Unskilled Manual 49 43 115 13 111 123 44 498 17VIIb Farm Workers 28 16 78 17 89 101 115 444 15N 623 209 665 127 576 560 242 3002 21 7 22 4 19 19 8

Note Cell entries represent counts

6 Design matrices in Erikson and Goldthorpe(1992a124ndash129)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

immobility They include three matrices IN1identifies immobility across all classes alike IN2identifies the higher immobility of the serviceclass (I+II) the self-employed group (IVab) andfarmers (IVc) and IN3 accounts for the highestimmobility of farmers (IVc)

SECTOR EFFECTS Sector effects are intended toidentify the difficulty of moving between theagricultural and nonagricultural sectors of theeconomy (SE1)

AFFINITY EFFECTS Affinity effects are intend-ed to capture specific discontinuities (negativeaffinities) or linkages (positive affinities) betweenclasses which either reinforce or offset the over-all effects of hierarchy and sector There are twoaffinity effects The first AF1 identifies the dif-ficulty of moving between the service class (I+II)and the class of farm workers (VIIb) which addsto their hierarchical distance In contrast AF2identifies instances in which mobility is morefrequent than accounted for by hierarchy and sec-tor effects and includes affinities within the non-manual sector (I+II and III) and within the manualsector (V+VI and VIIa) a symmetrical affinitybased on capital possession (IVab and IVc and Iand IVab) and an asymmetrical link between agri-cultural classes of origin (IVc and VIIb) and theunskilled manual class of destination (VIIa)7

TTHHEE MMOOBBIILLIITTYY PPAATTTTEERRNN IINN CCHHIILLEECCOONNVVEERRGGEENNCCEE OORR DDIIVVEERRGGEENNCCEE

Table 2 presents the fit of several models appliedto the Chilean data and Table 3 displays theparameter estimates for each of these modelsModel 1 in Table 2 indicates that the fit of theoriginal core model to the Chilean table is unsat-isfactory under standard statistical criteria butit explains a significant 775 percent of associ-ation under independence To assess the strengthof the different factors driving mobility oppor-tunities I compare the Chilean coefficients withthose obtained for the CASMIN countriesCoefficients from the original core model fit-ted to the Chilean table are reported in row 1 ofTable 3 CASMIN coefficients are reported inrow 2 and differences in coefficientsrsquo magni-tude and the statistical significance of the dif-ferences are reported in row 38

All Chilean coefficients are significant andhave the same sign as for the CASMIN coun-tries Comparison of the magnitude of the coef-ficients however shows significant departuresfrom the core for five effects which revealinteresting differences between Chile and theCASMIN countries The hierarchy coefficientsindicate that whereas the short-range hierar-chical barriers are somewhat weaker in Chile

443344mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

8 CASMIN coefficients were obtained from fittingthe core model to the combined table of France andEngland standardizing both tables to have 10000cases as recommended by Erikson and Goldthorpe(1992a121) The coefficients obtained are virtuallyidentical to those presented by Erikson andGoldthorpe (1992a Table 44)

Table 2 Fit statistics selected models to the Chilean Mobility Table

AssociationL2 BIC df p Explained

Core Modelmdash1 Core Model Original 1375 285 28 000 775mdash2 Core Model EHE 1081 ndash08 28 000 860mdash3 Chilean version of Core Model ([Model 2] + Chilean Parameters) 724 ndash326 27 000 906Association Models 000mdash4 Heterogeneous Quasi-RC(II) Model 734 ndash161 23 000 905mdash5 Homogeneous Quasi-RC(II) Model 803 ndash287 28 000 896mdash6 Quasi-Linear by Linear Association (SES row and column ranking)a 1563 279 33 000 798

Note EHE = empirical hierarchical effects SES = eocioeconomic statusa Linear-by-linear association model using socioeconomic status scores as presented in Appendix Figure A1 torank origin and destination classes

7 For a detailed account of the empirical derivationof the model see Erikson and Goldthorpe (1987a1987b and 1992a chapter 4)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443355

2322-ASR 703 filename70304-torche

Tab

le 3

M

odel

Par

amet

er E

stim

ates

for

the

Chi

lean

Mob

ilit

y Ta

ble

Eff

ect

1 C

ore

mod

el o

rigi

nala

mdash 2 C

ore

mod

el C

AS

MIN

b

mdash 3 D

iffe

renc

e C

ASM

INmdash

Chi

lec

mdash 4 C

ore

mod

el E

HE

d

mdash 5 C

hile

an v

ersi

on o

f co

ree

mdash Not

eD

ata

show

n ar

e m

odel

par

amet

er e

stim

ates

wit

h st

anda

rd e

rror

s in

par

enth

eses

EH

E =

em

piri

cal h

iera

rchi

cal e

ffec

ts

aM

odel

1 f

rom

Tab

le 2

b

Ori

gina

l Cor

e M

odel

fit

ted

to th

e E

ngli

sh a

nd F

renc

h m

obil

ity

tabl

es S

ee te

xt f

or d

etai

ls

cD

iffe

renc

e in

par

amet

er m

agni

tude

bet

wee

n or

igin

al c

ore

mod

el f

itte

d to

the

Chi

lean

tabl

e (r

ow 1

) an

d to

the

CA

SM

IN ta

ble

(row

2)

dM

odel

2 f

rom

Tab

le 2

e

Mod

el 3

fro

m T

able

2

dagger p

lt 1

p

lt 0

5

p

lt 0

1

p

lt 0

01 (

two-

tail

ed)

HI1

ndash12

(

05)

ndash22

(02

)ndash

10 dagger

(06

)ndash

03(

07)

ndash09

dagger(

06)

HI2

ndash65

(11

)ndash

45

(

04)

20

dagger(

12)

ndash74

(09

)ndash

47

(

10)

IN1

41

(

08)

43

(

03)

Insi

g

37

(

07)

34

(

07)

IN2

39

(

10)

84

(

04)

45

(

11)

31

(13

)mdash

IN3

68

(

24)

101

(11

)In

sig

61

(24

)4

8(

23)

SE

1

ndash38

(07

)ndash1

06

(

05)

ndash68

(08

)ndash

42

(

07)

ndash20

(

08)

AF

1

ndash76

(21

)ndash

83

(

13)

Insi

g

ndash84

(20

)ndash

63

(21

)

AF

2

33

(

05)

45

(

02)

ndash12

(

06)

20

(

06)

26

(

06)

AF

1ndashC

mdash mdash mdash mdash

ndash14

9

(46

)

IN2ndash

C

mdash mdash mdash mdash 71

(

14)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(HI1) long-range hierarchical mobility is moredifficult in this country (HI2) This suggestssubstantial inequality between the extremes ofthe class hierarchy combined with smaller dif-ferentiation in the middle of the distribution Asto the inheritance effects the service class theself-employed group and the farmers (classesI+II IVab and IVc respectively) have a sig-nificantly lower propensity to immobility inChile than in the CASMIN countries suggest-ing that class positions usually associated withindependent work are less able to reproducetheir class status intergenerationally than depict-ed by the core Most impressive the barrierbetween the agricultural and nonagriculturalsectors of the economy (SE1) is much weakerin Chile as is the affinity effect linking non-manual classes manual classes and the class-es that own capital among themselves (AF2)

Interestingly all the significant differences incoefficients with the single exception of the bar-rier to long-range hierarchical mobility sug-gest that Chile is more fluid than the core modeldepicts Higher fluidity seems to be driven bythe weakness of horizontal barriers separatingthe agricultural manual and self-employed sec-tors of the economy This is consistent withfindings of weaker sector effects in countriesthat have experienced rapid industrializationsuch as Korea (Park 2004) Brazil (Costa-Ribeiro 2003) Israel (Yaish 2004 Goldthorpeet al 1997) and Japan (Ishida et al 1991)

Additionally in the Chilean case the weak-ness of sector barriers may have been intensi-fied by the deep market transformation of the1970s and 1980s On the one hand the ldquochangeof handsrdquo in the agricultural sector induced bythe agrarian reform and counterreform likelyincreased fluidity between the agricultural andurban classes On the other hand the role of self-employment as an ephemeral refuge against(industrial and public sector) unemploymentmay have reduced barriers between the self-employed group and the rest of the social struc-ture Added to the larger difficulty of long-rangehierarchical mobility these features suggestweak sector cleavages in the Chilean fluiditypattern and high salience of the hierarchicaldimension of mobility

For the countries in which the fit of the coremodel is not adequate Erikson and Goldthorpe(1992a145) suggest adding parameters thataccount for country-specific deviations from

the model I therefore introduce adjustments toreflect salient characteristics of the Chilean flu-idity pattern producing a ldquoChilean versionrdquo ofthe core model

First I modify the hierarchical effects Fittingof the core model suggests that hierarchicalbarriers are a crucial determinant of mobility inChile This conclusion is however obscuredby the fact that the hierarchical strata distin-guished by the core model are not empiricallyobtained and may not accurately represent theChilean ranking of classes in terms of socioe-conomic status (SES) To model hierarchicaleffects in Chile adequately I rank classes usingthe unweighted average of their schooling andearning levels as a proxy for SES I then collapsethe seven classes into three hierarchical stratausing cluster analysis These three empiricallyobtained strata are the following Upper stratum(service class I+II) middle stratum (routinenonmanual and self-employed group III andIVab) and lower stratum (manual and agricul-tural classes V-VI VIIa IVc and VIIb) Detailson the ranking of classes cluster procedures andthe comparison of the CASMIN and empiricalChilean class ranking can be found in theAppendix Comparison of the CASMIN and theempirical class ranking suggests higher differ-entiation at the top than at the bottom of thesocial structure in the Chilean case

Model 2 in Table 2 uses the empiricallyobtained hierarchical strata (EHE) Theimprovement in fit is massive (the difference inBIC is 293 without using any degrees of free-dom) The model now accounts for 86 percentof the association under independence

The large improvement in fit when empiricalhierarchical strata are used challenges theassumption of an internationally homogeneoushierarchical class ranking At the empiricallevel there is no proof that the hierarchicalranking of classes assumed by the core modelfits all CASMIN countries9 Moreover someevidence points to significant cross-country dif-

443366mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

9 Erikson and Goldthorpe(1992a Table 22) pres-ent the average ranking of classes based on standardstatus measures but they do not show the empiricalranking of classes for each country If the occupa-tional composition of classes varies across countriesthere may be significant international differences inthe hierarchical ranking of classes

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ferences in the ranking of classes (Hout andHauser 1992 Sorensen 1992) At the theoreti-cal level the creators of the core model high-light the relevance of highly specific historicallyformed national attributes to explain interna-tional variation in mobility These national par-ticularities may very well result in differenthierarchical positions of classes across coun-tries10 Furthermore even if the homogeneousranking of classes imposed by the CASMINresearchers applied to most CASMIN coun-tries significant departures may be found inthe late industrialized countries currently beingadded to the comparative mobility project TheChilean findings suggest that instead of impos-ing international homogeneity to the hierarchi-cal dimension of mobility the ranking of classesshould be treated as nationally specific param-eters to be estimated as mobility researchersroutinely treat row and column marginals ofthe mobility table The issue is importantbecause the hierarchical dimension is the mostrelevant dimension of mobility (Hout andHauser 1992 Wong 1992) and inadequate mod-eling can bias assessment of not only hierar-chical effects of mobility but also other effects

Two other changes were introduced toaccount for Chilean-specific deviations fromthe core model First the IN2 effect wasrearranged to exclude the self-employed class(IVab) and to include the class of farm workers(VIIb) This rearrangement accounts for thefact that the Chilean self-employed classappeared less likely than depicted by the coremodel to reproduce its class position acrossgenerations and the inheritance level of Chileanfarm workers is higher than in the CASMINcountries These changes are expressed in theChilean-specific inheritance parameter IN2-CSecond an affinity parameter (AF1-C) wasadded to capture the difficulty of long-rangedownward elite mobility This negative asym-metrical affinity links flows from the service

class to the agricultural classes and from theroutine nonmanual class to the class of farmworkers Note that the AF1-C adds to the hier-archical distance between the two extremes ofthe class structure and to the negative affinitycaptured by AF1 to depict the extreme difficultyof long-range downward mobility from theChilean elite The fit of this lsquoChilean versionrsquoof the core model is reported in Table 2 model3 The improvement in BIC from model 2 is sub-stantial Furthermore even if the model fallsshort in terms of standard statistical criteria itaccounts for a large 91 of the associationunder independence thereby supporting theclaim that it adequately represents the Chileanmobility pattern The parameter estimates for theChilean version of the core are reported in row5 Table 3

Because of the core modelrsquos undesirable prop-erties I also use an alternative approach toexamine Chilean mobility patterns Thisapproach uses the now standard associationrowndashcolumn (II) (RC[II]) model (Goodman1979 see also Hout 1983 and Wong 1992) Inthis model both origin and destination classesare scaled so that the association can beexpressed as a linear-by-linear interaction by asingle parameter The multiplicative form ofthe model can be expressed as follows

Fij = iO j

D exp( microi νj)

where i indexes class of origin j indexes classof destination Fij is the expected frequency inthe (ij) cell is the grand mean i

O pertainsto the class of origin marginal effect j

D pertainsto the class of destination marginal effect isa global association parameter microi is the scalevalue for the ith class of origin and νj is the scalescore for the jth class of destination subject tothe following normalization constraints Σmicroi =Σνj = 0 (normalization of the location) and Σmicroi2

= Σνj2 = 1 (normalization of the scale)The class scale scores reflect a latent con-

tinuous variable made manifest by the class cat-egories Empirically obtained from the datathese ldquodistancerdquo scores produce an optimalranking of classes for the purpose of mobilityanalysis The RC(II) model is appropriate formodeling the association of any two ordinalvariables but mobility tables are distinctivebecause of the correspondence between class oforigin and class of destination (Gerber and Hout2004691)

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443377

2322-ASR 703 filename70304-torche

10 For instance during the socialist period inHungary the status of the proletariat may have beenindistinguishable from the status of the routine non-manual class (Szelenyi 199862) In Scotland givento early industrialization the hierarchical barrierbetween the skilled and unskilled working classes thatcharacterizes other CASMIN countries may not exist(Erikson and Goldthorpe 1992a163)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I introduce two adjustments to account forthis correspondence transforming the formu-lation into a ldquohybrid modelrdquo First topologicalparameters will account for the large counts insome diagonal cells reflecting immobility(inheritance) effects These parameters distin-guish two inheritance levels ldquohigh inheritancerdquofor the service class (I+II) farmers (IVc) andfarm workers (VIIb) and ldquolow inheritancerdquo forall other classes with the exception of the rou-tine nonmanual workers

Model 4 in Table 2 reports the fit statistic forthis quasi-RC(II) model The BIC statisticrejects it and prefers the more parsimoniousChilean version of the core model ThereforeI introduce a second adjustment to account forthe correspondence between rows and columnsin mobility tables Origin and destination scoresare constrained to be the same assuming thatthe scaling of classes has not significantlychanged over time Model 5 in Table 2 reportsfit statistics for this homogeneous quasi-RC(II)model On the basis of the BIC statistic the fitof the model is as good as that of the Chileanversion of the core model (Wong [1994] showsthat BIC differences of less than 5 points shouldbe considered indeterminate) Thus under stan-dard statistical criteria the Chilean core modeland the homogeneous quasi-RC(II) model areindistinguishable Given that the Chilean versionof the core model was tailored to fit the Chileantable the fit of model 5 is impressive and it indi-cates that a unidimensional scale can adequatelycapture mobility distances between classes

The logical next question is what this scalerepresents and specifically whether it repre-sents the hierarchical ranking of classes in termsof SES If a close correspondence is foundbetween the mobility distances among classesand their distances in terms of SES this find-ing would suggest that the intragenerationaldistribution of resources and rewards expressedin the SES ranking of classes drives the processof intergenerational mobility In other words thisfinding would support the resource approachrsquosclaim that differences in resources associatedwith diverse class positions determine mobili-ty opportunities across generations

To examine this possibility I compare theclass scores obtained from the homogeneousquasi-RC(II) model with the empiricallyobtained SES ranking of classes based on edu-cation and earnings (Appendix) Figure 4 pres-

ents the comparison between the standardizedvalues of both sets of scores The correspon-dence is striking The mobility distancesbetween classes closely reproduce their dis-tances in terms of SES This close isomorphismsuggests that at least in Chile the intragener-ational inequality in the distribution of socialresources and rewards across classes largelydrives the intergenerational mobility process Totest the isomorphism further I use the SESclass scores as row and column rankings to esti-mate a quasindashlinear-by-linear association model(model 6 Table 2) The fit of model 6 is sig-nificantly worse than that of model 5 almost aspoor as that of the original core model This indi-cates that although the inter- and intragenera-tional dimensions of inequality are very similarin Chile they have a somewhat different struc-ture

In summary the central characteristic of theChilean mobility regime is the predominance oflong-range hierarchical barriers and the weak-ness of sector cleavages separating classes in themiddle and at the lower end of the hierarchicalstructure Findings from the ldquoChilean versionrdquoof the core model and the quasi-RC(II) hybridmodels are consistent depicting a social struc-ture characterized by signif icant barriersbetween the top echelon and the rest of the classstructure The good fit of the homogeneousquasi-RC(II) model provides preliminary sup-port for the resource approach linking inequal-ity and mobility Mobility dynamics seem to bedriven largely by the hierarchical distancesbetween classes and overall Chile seems not tobe less fluid than the CASMIN countries

CCOOMMPPAARRIISSOONN OOFF TTHHEE OORRIIGGIINNAALL AANNDD CCHHIILLEEAANN

VVEERRSSIIOONNSS OOFF TTHHEE CCOORREE MMOODDEELL

Although Chilean mobility dynamics seem to bedriven by inequality across classes in a veryunequal society the fitting of the core modelprovides no indication that Chile is significantlyless fluid than depicted by the core model Asa preliminary analysis of the Chilean level of flu-idity I compare the predicted propensities formobility and immobility between specific pairsof classes in Chile with those in the CASMINcountries Propensities are obtained from theoriginal and Chilean versions of the core modelrespectively (models 2 and 5 of Table 3)

443388mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I begin with the immobility of those at the topof the class structure namely the service class(I+II) The service classrsquo propensity for immo-bility in the original core model is captured bytwo parameters IN1 and IN2 This propensityis more than three times what it would be in theabsence of these effects (e43+84 = 356) InChile the propensity for immobility of class I+IIis captured by the same two parameters and itis lower than what the core model expresses286 times larger than it would be in the absenceof these effects (e34+71) What about the immo-bility of the two agricultural classes In theCASMIN countries immobility of the IVc andVIIb classes is respectively 785 and 154 timeswhat it would be in the absence of these param-eters (e43+84+101ndash22 and e43 respectively) InChile the comparable values are 422 and 286(e34+71+48ndash09 and e43+71 respectively) indi-cating the relatively low inheritance associatedwith land ownership in the Latin Americannation Thus if attention is focused on theimmobility in the upper and lower ends of thesocial hierarchy the comparison does not indi-cate less fluidity in the highly unequal Chileansociety

What about the mobility rates between thetwo most distant classes in the hierarchical rank-ing namely the service class (I+II) and farmworkers (VIIb) In the case of the CASMINcountries the mobility chances between theseclasses are captured by four parameters (HI1HI2 SE1 and AF1) and is 077 times smallerthan it would be in the absence of these effects(endash22 ndash45ndash106ndash83) In Chile the distance fromclass I to class VIIb is expressed by these samefour parameters in addition to the asymmetri-cal disaffinity AF1-C This yields mobilitychances 056 times what would be obtained inthe absence of these effects (endash09 ndash47ndash20ndash63ndash149)Thus the barrier to long-range downwardmobility from the elite is more significant inChile than in the CASMIN countries The sameoccurs with long-range downward mobility fromclass III

If we measure the reciprocal flow (ieupward mobility from class VIIb to class I+II)we obtain the same value for the CASMINcountries because all parameters are symmet-rical in the original core model For Chile how-ever this calculation removes the asymmetricalbarrier to elite downward mobility which rais-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443399

2322-ASR 703 filename70304-torche

Figure 4 Class Scores Obtained from the Homogeneous Quasi-RC(II) Model and from Socioeconomic Status(SES) Ranking

Note (1) = Class standardized socioeconomic status scores based on schooling and earnings (2) = Class stan-dardized scores based on the Homogeneous Quasi-RC(II) model (see text for details)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 9: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

Therefore Chile is unequal largely because theelite concentrates an extremely high proportionof the national income Across the noneliteclasses the distribution of resources is muchmore uniform

Inequality is not a new development inChilean society but rather has deep historicalroots Figure 3 presents the longest availableseries on inequality in Chile depicting earningsdistribution since 1957 in Santiago3 and show-ing persistently high inequality over the lasthalf century However there is significant vari-ation over time a short decline of inequality dur-ing the progressive administrations of themid-1960s and early 1970s an increase since themilitary regime took power in 1973 a peak in1984 and a small decline after the democratictransition to levels still higher than those thatpreceded the military regime

LLAATTEE IINNDDUUSSTTRRIIAALLIIZZAATTIIOONN MMAARRKKEETTRREEFFOORRMM AANNDD TTHHEE CCHHIILLEEAANN CCLLAASSSSSSTTRRUUCCTTUURREE

The ldquocommon social fluidityrdquo hypothesis assertsthat all international variation in mobility pat-terns is attributable to highly specific histori-cally formed national characteristics In theChilean case these factors may be related to themarket reform of the 1970s and 1980s and therapid economic growth since the late 1980sNational mobility studies suggest that coun-tries experiencing rapid industrialization may becharacterized by weaker sector barriers to mobil-ity specif ically barriers between the self-employed and employees and betweenagricultural and nonagricultural classes(Goldthorpe Yaish and Kraus 1997 and Yaish2004 for Israel Ishida Goldthorpe and Erikson1991 for Japan Park 2004 for Korea and Costa-Ribeiro 2003 for Brazil)

In addition two components of the Chileanmarket transformation may have contributed tothe weakness of mobility barriers across sectorsof the economy The first component is theagrarian reform and subsequent counterreformundertaken by the military regime The secondone is the opening of the economy to interna-tional trade and the retrenchment of the statewhich led to a decline in formal employment

443300mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Figure 3 Earnings Inequality in Greatest Santiago 1957ndash2000Source Larranaga 1999 and calculations by the author

3 Santiago the Chilean capital comprises aboutone-third of the countryrsquos population Assessmentby Chilean experts suggests that Santiago trendspresent an unbiased picture of trends at the nationallevel Although earnings inequality is not as com-prehensive a measure as total income inequality it isan adequate much more accurately measured proxy(Galbraith 2002)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

and to the growth and segmentation of the self-employed sector

The main objective of the Chilean agrarianreform (1962ndash1973) was to reduce the extremeland concentration in the country The reformwas very successful In 1955 the top 7 percentof landowners held 65 percent of the land andthe bottom 37 percent held only 1 percent By1973 the last year of the reform 43 percent ofthe land had been expropriated (Kay 2002 Scott1996) An agrarian counterreform was launchedby the military regime as a way of returning theland to its former owners It did not howeverrestore the traditional hacienda order (Gomezand Echenique 1988 Rivera 1988) In fact themilitary government returned only one-third ofthe plots to their old owners sold another thirdand adjudicated the remaining third to the smallproprietors benefited by the agrarian reformUnable to compete with the now cheap foodimports many small proprietors opted to selltheir plots As a consequence as much as two-thirds of the Chilean agricultural land came upfor sale creating an active land market Themain beneficiaries of the ldquomarketization of thecountrysiderdquo were a new group of export-ori-ented entrepreneurs and international investorswho bought large tracts of land (Gwynne 1996Kay 2002) Thus the reform and counterreformtriggered a massive ldquochange of handsrdquo in theChilean countryside likely altering the patternsof land inheritance

The second major effect of the market reformon the class structure was the reduction of for-mal employment caused by the decline of theindustrial working class and by the shrinkage ofthe state Incapable of competing with the nowcheap imports the working class plummetedfrom 34 percent of total employment in theearly 1970s to 20 percent in 1980 while pub-lic employment dropped from 14 to 84 per-cent in the same period (Leon and Martinez2000 Schkolnick 2000 Velasquez 1990) Thedecline in industrial and public employmentled to an increase in self-employment fromabout 15 percent in 1970 to about 28 percent in1980 (Thomas 1996) followed by a declinethrough the economic recovery before stabi-lization at about 22 percent of the total employ-ment in 2000 (MIDEPLAN 2001) In contrastto industrialized countries in which the self-employment rate usually is within the one-digitrange and involves capital ownership in Chile

it encompasses more than one-fifth of the laborforce Furthermore self-employment hadbecome a survival strategy for a large numberof former industrial and public employeesappropriately labeled ldquoforced entrepreneurial-ismrdquo (Infante and Klein 1995 Portes Castellsand Benton 1989) As a consequence theChilean self-employed sector is voluminousand segmented It includes a small segment ofentrepreneurs oriented to capital accumulationand able to hire employees and a large seg-ment of self-employed workers mostly engagedin ldquosurvival activitiesrdquo and unable to hire thelabor of others Differences in economic well-being between these two groups are massivewith the former group earning on averagethree times more than the latter (MIDEPLAN2001)

To be sure a large self-employed sector is nota result of the market transformation Self-employment has been historically high in LatinAmerica because according to some scholarsthe region has experienced ldquodependent inte-grationrdquo into the world capitalistic system(Hopkins and Wallerstein 1982 Luxembourg1951) However the internal heterogeneity ofthe self-employed sector increased during themarket reform in a pattern that foretold whathappened in the rest of Latin America during the1990s (Klein and Tokman 2000 Portes andHoffman 2003) and what may be current devel-opments in the industrialized world (Arum andMuumlller 2004 Noorderhaven et al 2003) Thesecharacteristics of the Chilean class structureare considered in the comparative analysis ofChilean mobility

DDAATTAA AANNDD AANNAALLYYTTIICCAALL AAPPPPRROOAACCHH

This study uses the 2001 Chilean MobilitySurvey (CMS) The CMS is a nationally repre-sentative multistage stratified sample of maleheads of household ages 24 to 69 The samplingstrategy includes the following stages First 87primary sampling units (PSUs) (counties) areselected Then blocks within the PSUs areselected and finally households within blocksare chosen Counties are stratified according tosize (fewer than 20000 20000 to 100000100000 to 200000 and more than 200000inhabitants) and by geographic zone (NorthCenter South) All PSUs in the large size stra-tum are included in the sample to increase effi-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443311

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ciency The fieldwork conducted between Apriland June 2001 consists of face-to-face inter-views in the respondentrsquos household carried outby trained personnel The survey excludesnonndashhead-of-household males which represent17 percent of the male population of the rele-vant age (MIDEPLAN 2001) Among thoseexcluded 865 percent are the sons of the headof the household 124 percent are another rel-ative of the head and 11 percent are other non-related males Their occupational distributionwith control used for age is almost identical tothat of the heads of household The small pro-portion represented by this group and their sim-ilar occupational distribution suggest that theirinclusion would not significantly alter the find-ings presented in this article Excluding thehouseholds not eligible for the survey theresponse rate is 63 percent Although nonre-sponse rates usually are not reported in Chileansurveys exchange with Chilean experts indi-cates that the nonresponse rate is about 20 to 25percent for face-to-face household surveys Thehigher nonresponse rate of the CMS is likelyattributable to the difficulty contacting maleheads of household4 The total sample size is3544 I exclude individuals outside the agerange of 25 to 64 years which is conventional-ly used in comparative mobility research andcases with unusable data After this exclusionthe usable sample size is 3002

CCLLAASSSS SSCCHHEEMMAA

This study uses a class perspective to describethe Chilean social structure It utilizes the seven-category version of the classification devel-oped by the CASMIN project Thisclassification is widely used in internationalcomparative research and describes the basic

stratification of advanced industrial societiesbased on ldquoemployment relationshipsrdquo (Eriksonand Goldthorpe 1992a35ndash47 Goldthorpe andHeath 1992) The seven classes included in theschema are I+II (service class) III (routine non-manual workers) IVab (self-employed work-ers) V+VI (manual supervisors and skilledmanual workers) VIIa (unskilled manual work-ers) IVc (farmers) and VIIb (farm workers)

Use of the 7-class categorization instead ofan alternative detailed 12-class version pro-duced by the CASMIN project (Erikson andGoldthorpe 1992b) induces some loss of infor-mation about the originndashdestination associa-tion A global test of all five aggregationsobtained from the ratio of L2 tests for the inde-pendence model in the collapsed and full tablesindicates that the seven-class schema masks 29percent of the association shown by the full setof classes ([L2 full table ndash L2 collapsed table]L2 full table = [108644 ndash 77191]108644 =29) This proportion is high compared withthat of the CASMIN countries and is surpassedonly by Sweden with 31 percent (Hout andHauser 1992 Table 2) A collapsed sevenfoldclass schema was however preferred because ofthe moderate sample size of the Chilean surveyand because it grants international comparabil-ity of the findings5

Operationalization of this class classifica-tion is based on detailed information concern-ing job title (recoded into the standard ISCO-88classification [ILO 1990]) industry occupa-tional status and supervisory status of workersBecause the CMS is specifically designed toassess class mobility it includes all the infor-mation necessary to produce the CASMIN classschema The CASMIN project has not produceda standard algorithm to generate the class clas-sification but other researchers have producedsuch an algorithm which I use in this analysis(Ganzeboom and Treiman 2003) Table 1 pres-ents the basic intergenerational mobility tableto be used in this analysis including the countsand marginal percentage distribution for originand destination classes

443322mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

4 The nonresponse rate can yield bias if those whowere unreachable or refused to participate differ fromthose included in the sample To estimate the mag-nitude of this bias I compared the distribution of keyvariables with the CASEN 2000 survey CASEN isa large survey (n = 252595) conducted by the Chileangovernment and has a refusal rate lower than 10percent The comparison (available upon request)suggests that the CMS slightly underrepresents agri-cultural workers and the upper class Overall how-ever there is no indication of major nonresponsebias

5 Analyses were replicated using the detailed 12-class table and no difference in substantive find-ings was found

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

MMEETTHHOODDSS

I fit two alternative models to the Chilean mobil-ity table The first is the ldquocore model of social flu-idityrdquo which claims to capture basic internationalsimilarity in mobility Although the core modelhas become a benchmark for mobility researchand currently is the only model that provides acomparative framework for mobility analysisanalysts have drawn attention to a number of itslimitations The model has been criticized forinadequately representing hierarchical mobilityeffects for being tailored to a few industrializedcountries for the ad hoc nature of some effects itincludes (specifically the affinity effects) andfor lacking a criterion to determine how muchdivergence is necessary for it to be rejected(Ganzeboom Luijkx and Treiman 1989 Houtand Hauser 1992 Sorensen 1992 Yamaguchi1987) I therefore also test an alternative ldquohybridmodelrdquo that combines the rowndashcolumn (II) asso-ciation model (Goodman 1979) with parametersaccounting for class immobility The internationaland temporal comparison of Chilean mobilitypatterns uses the uniform-difference (UNIDIFF)model also known as the multiplicative layereffect model (Erikson and Goldthorpe 1992aXie 1992) Model comparison and selection isbased on the Bayesian Information Criterion(BIC) statistic (Raftery 1995)

AANNAALLYYSSIISS OOFF CCHHIILLEEAANN MMOOBBIILLIITTYY IINNCCOOMMPPAARRAATTIIVVEE PPEERRSSPPEECCTTIIVVEE

TTHHEE CCOORREE MMOODDEELL OOFF SSOOCCIIAALL FFLLUUIIDDIITTYY

I start by fitting the core model to the Chileanmobility table The core model represents the

basic international commonality in social flu-idity using a topological formulation In contrastto standard topological models constructed froma single allocation of cells (Hauser 1978) thecore model uses eight matrices each of whichis designed to capture a particular effect thatenhances or reduces mobility between specificclasses (Erikson and Goldthorpe 1987a 1992achapter 4) In this way the model permits thesources of commonality across nations to beidentified clearly and likewise the departuresfrom it The effects are of four types hierarchyinheritance sector and affinity6

HIERARCHY EFFECTS Hierarchy effects reflectthe impact of status distances between classeson fluidity between them To estimate theseeffects the mobility table is divided into threestrata reflecting differences in resources andrewards across classes The strata are the fol-lowing upper stratum (class I+II) middle stra-tum (classes III IVab IVc and V-VI) and lowerstratum (classes VIIa and VIIb) Because thereare three strata two hierarchy effects (HI1 andHI2) can be identified each representing thecrossing of one additional hierarchical barrierThe association between each pair of classes isexpressed as an inverse function of the numberof hierarchical strata crossed

INHERITANCE EFFECTS Inheritance effects aredesigned to capture the propensity for class

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443333

2322-ASR 703 filename70304-torche

Table 1 Seven-Class Intergenerational Mobility Table for Chilean Men

Class of Destination

I-II III IVab IVc V-VI VIIa VIIb N

I-II Service Class 229 49 81 2 37 30 2 430 14III Routine Nonmanual 61 15 34 0 24 27 1 162 5IVab Self-employed 123 37 173 28 102 90 20 573 19IVc Farmers 42 16 68 54 58 54 24 316 11V-VI Skilled Manual 91 33 116 13 155 135 36 579 19VIIa Semiskilled and Unskilled Manual 49 43 115 13 111 123 44 498 17VIIb Farm Workers 28 16 78 17 89 101 115 444 15N 623 209 665 127 576 560 242 3002 21 7 22 4 19 19 8

Note Cell entries represent counts

6 Design matrices in Erikson and Goldthorpe(1992a124ndash129)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

immobility They include three matrices IN1identifies immobility across all classes alike IN2identifies the higher immobility of the serviceclass (I+II) the self-employed group (IVab) andfarmers (IVc) and IN3 accounts for the highestimmobility of farmers (IVc)

SECTOR EFFECTS Sector effects are intended toidentify the difficulty of moving between theagricultural and nonagricultural sectors of theeconomy (SE1)

AFFINITY EFFECTS Affinity effects are intend-ed to capture specific discontinuities (negativeaffinities) or linkages (positive affinities) betweenclasses which either reinforce or offset the over-all effects of hierarchy and sector There are twoaffinity effects The first AF1 identifies the dif-ficulty of moving between the service class (I+II)and the class of farm workers (VIIb) which addsto their hierarchical distance In contrast AF2identifies instances in which mobility is morefrequent than accounted for by hierarchy and sec-tor effects and includes affinities within the non-manual sector (I+II and III) and within the manualsector (V+VI and VIIa) a symmetrical affinitybased on capital possession (IVab and IVc and Iand IVab) and an asymmetrical link between agri-cultural classes of origin (IVc and VIIb) and theunskilled manual class of destination (VIIa)7

TTHHEE MMOOBBIILLIITTYY PPAATTTTEERRNN IINN CCHHIILLEECCOONNVVEERRGGEENNCCEE OORR DDIIVVEERRGGEENNCCEE

Table 2 presents the fit of several models appliedto the Chilean data and Table 3 displays theparameter estimates for each of these modelsModel 1 in Table 2 indicates that the fit of theoriginal core model to the Chilean table is unsat-isfactory under standard statistical criteria butit explains a significant 775 percent of associ-ation under independence To assess the strengthof the different factors driving mobility oppor-tunities I compare the Chilean coefficients withthose obtained for the CASMIN countriesCoefficients from the original core model fit-ted to the Chilean table are reported in row 1 ofTable 3 CASMIN coefficients are reported inrow 2 and differences in coefficientsrsquo magni-tude and the statistical significance of the dif-ferences are reported in row 38

All Chilean coefficients are significant andhave the same sign as for the CASMIN coun-tries Comparison of the magnitude of the coef-ficients however shows significant departuresfrom the core for five effects which revealinteresting differences between Chile and theCASMIN countries The hierarchy coefficientsindicate that whereas the short-range hierar-chical barriers are somewhat weaker in Chile

443344mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

8 CASMIN coefficients were obtained from fittingthe core model to the combined table of France andEngland standardizing both tables to have 10000cases as recommended by Erikson and Goldthorpe(1992a121) The coefficients obtained are virtuallyidentical to those presented by Erikson andGoldthorpe (1992a Table 44)

Table 2 Fit statistics selected models to the Chilean Mobility Table

AssociationL2 BIC df p Explained

Core Modelmdash1 Core Model Original 1375 285 28 000 775mdash2 Core Model EHE 1081 ndash08 28 000 860mdash3 Chilean version of Core Model ([Model 2] + Chilean Parameters) 724 ndash326 27 000 906Association Models 000mdash4 Heterogeneous Quasi-RC(II) Model 734 ndash161 23 000 905mdash5 Homogeneous Quasi-RC(II) Model 803 ndash287 28 000 896mdash6 Quasi-Linear by Linear Association (SES row and column ranking)a 1563 279 33 000 798

Note EHE = empirical hierarchical effects SES = eocioeconomic statusa Linear-by-linear association model using socioeconomic status scores as presented in Appendix Figure A1 torank origin and destination classes

7 For a detailed account of the empirical derivationof the model see Erikson and Goldthorpe (1987a1987b and 1992a chapter 4)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443355

2322-ASR 703 filename70304-torche

Tab

le 3

M

odel

Par

amet

er E

stim

ates

for

the

Chi

lean

Mob

ilit

y Ta

ble

Eff

ect

1 C

ore

mod

el o

rigi

nala

mdash 2 C

ore

mod

el C

AS

MIN

b

mdash 3 D

iffe

renc

e C

ASM

INmdash

Chi

lec

mdash 4 C

ore

mod

el E

HE

d

mdash 5 C

hile

an v

ersi

on o

f co

ree

mdash Not

eD

ata

show

n ar

e m

odel

par

amet

er e

stim

ates

wit

h st

anda

rd e

rror

s in

par

enth

eses

EH

E =

em

piri

cal h

iera

rchi

cal e

ffec

ts

aM

odel

1 f

rom

Tab

le 2

b

Ori

gina

l Cor

e M

odel

fit

ted

to th

e E

ngli

sh a

nd F

renc

h m

obil

ity

tabl

es S

ee te

xt f

or d

etai

ls

cD

iffe

renc

e in

par

amet

er m

agni

tude

bet

wee

n or

igin

al c

ore

mod

el f

itte

d to

the

Chi

lean

tabl

e (r

ow 1

) an

d to

the

CA

SM

IN ta

ble

(row

2)

dM

odel

2 f

rom

Tab

le 2

e

Mod

el 3

fro

m T

able

2

dagger p

lt 1

p

lt 0

5

p

lt 0

1

p

lt 0

01 (

two-

tail

ed)

HI1

ndash12

(

05)

ndash22

(02

)ndash

10 dagger

(06

)ndash

03(

07)

ndash09

dagger(

06)

HI2

ndash65

(11

)ndash

45

(

04)

20

dagger(

12)

ndash74

(09

)ndash

47

(

10)

IN1

41

(

08)

43

(

03)

Insi

g

37

(

07)

34

(

07)

IN2

39

(

10)

84

(

04)

45

(

11)

31

(13

)mdash

IN3

68

(

24)

101

(11

)In

sig

61

(24

)4

8(

23)

SE

1

ndash38

(07

)ndash1

06

(

05)

ndash68

(08

)ndash

42

(

07)

ndash20

(

08)

AF

1

ndash76

(21

)ndash

83

(

13)

Insi

g

ndash84

(20

)ndash

63

(21

)

AF

2

33

(

05)

45

(

02)

ndash12

(

06)

20

(

06)

26

(

06)

AF

1ndashC

mdash mdash mdash mdash

ndash14

9

(46

)

IN2ndash

C

mdash mdash mdash mdash 71

(

14)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(HI1) long-range hierarchical mobility is moredifficult in this country (HI2) This suggestssubstantial inequality between the extremes ofthe class hierarchy combined with smaller dif-ferentiation in the middle of the distribution Asto the inheritance effects the service class theself-employed group and the farmers (classesI+II IVab and IVc respectively) have a sig-nificantly lower propensity to immobility inChile than in the CASMIN countries suggest-ing that class positions usually associated withindependent work are less able to reproducetheir class status intergenerationally than depict-ed by the core Most impressive the barrierbetween the agricultural and nonagriculturalsectors of the economy (SE1) is much weakerin Chile as is the affinity effect linking non-manual classes manual classes and the class-es that own capital among themselves (AF2)

Interestingly all the significant differences incoefficients with the single exception of the bar-rier to long-range hierarchical mobility sug-gest that Chile is more fluid than the core modeldepicts Higher fluidity seems to be driven bythe weakness of horizontal barriers separatingthe agricultural manual and self-employed sec-tors of the economy This is consistent withfindings of weaker sector effects in countriesthat have experienced rapid industrializationsuch as Korea (Park 2004) Brazil (Costa-Ribeiro 2003) Israel (Yaish 2004 Goldthorpeet al 1997) and Japan (Ishida et al 1991)

Additionally in the Chilean case the weak-ness of sector barriers may have been intensi-fied by the deep market transformation of the1970s and 1980s On the one hand the ldquochangeof handsrdquo in the agricultural sector induced bythe agrarian reform and counterreform likelyincreased fluidity between the agricultural andurban classes On the other hand the role of self-employment as an ephemeral refuge against(industrial and public sector) unemploymentmay have reduced barriers between the self-employed group and the rest of the social struc-ture Added to the larger difficulty of long-rangehierarchical mobility these features suggestweak sector cleavages in the Chilean fluiditypattern and high salience of the hierarchicaldimension of mobility

For the countries in which the fit of the coremodel is not adequate Erikson and Goldthorpe(1992a145) suggest adding parameters thataccount for country-specific deviations from

the model I therefore introduce adjustments toreflect salient characteristics of the Chilean flu-idity pattern producing a ldquoChilean versionrdquo ofthe core model

First I modify the hierarchical effects Fittingof the core model suggests that hierarchicalbarriers are a crucial determinant of mobility inChile This conclusion is however obscuredby the fact that the hierarchical strata distin-guished by the core model are not empiricallyobtained and may not accurately represent theChilean ranking of classes in terms of socioe-conomic status (SES) To model hierarchicaleffects in Chile adequately I rank classes usingthe unweighted average of their schooling andearning levels as a proxy for SES I then collapsethe seven classes into three hierarchical stratausing cluster analysis These three empiricallyobtained strata are the following Upper stratum(service class I+II) middle stratum (routinenonmanual and self-employed group III andIVab) and lower stratum (manual and agricul-tural classes V-VI VIIa IVc and VIIb) Detailson the ranking of classes cluster procedures andthe comparison of the CASMIN and empiricalChilean class ranking can be found in theAppendix Comparison of the CASMIN and theempirical class ranking suggests higher differ-entiation at the top than at the bottom of thesocial structure in the Chilean case

Model 2 in Table 2 uses the empiricallyobtained hierarchical strata (EHE) Theimprovement in fit is massive (the difference inBIC is 293 without using any degrees of free-dom) The model now accounts for 86 percentof the association under independence

The large improvement in fit when empiricalhierarchical strata are used challenges theassumption of an internationally homogeneoushierarchical class ranking At the empiricallevel there is no proof that the hierarchicalranking of classes assumed by the core modelfits all CASMIN countries9 Moreover someevidence points to significant cross-country dif-

443366mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

9 Erikson and Goldthorpe(1992a Table 22) pres-ent the average ranking of classes based on standardstatus measures but they do not show the empiricalranking of classes for each country If the occupa-tional composition of classes varies across countriesthere may be significant international differences inthe hierarchical ranking of classes

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ferences in the ranking of classes (Hout andHauser 1992 Sorensen 1992) At the theoreti-cal level the creators of the core model high-light the relevance of highly specific historicallyformed national attributes to explain interna-tional variation in mobility These national par-ticularities may very well result in differenthierarchical positions of classes across coun-tries10 Furthermore even if the homogeneousranking of classes imposed by the CASMINresearchers applied to most CASMIN coun-tries significant departures may be found inthe late industrialized countries currently beingadded to the comparative mobility project TheChilean findings suggest that instead of impos-ing international homogeneity to the hierarchi-cal dimension of mobility the ranking of classesshould be treated as nationally specific param-eters to be estimated as mobility researchersroutinely treat row and column marginals ofthe mobility table The issue is importantbecause the hierarchical dimension is the mostrelevant dimension of mobility (Hout andHauser 1992 Wong 1992) and inadequate mod-eling can bias assessment of not only hierar-chical effects of mobility but also other effects

Two other changes were introduced toaccount for Chilean-specific deviations fromthe core model First the IN2 effect wasrearranged to exclude the self-employed class(IVab) and to include the class of farm workers(VIIb) This rearrangement accounts for thefact that the Chilean self-employed classappeared less likely than depicted by the coremodel to reproduce its class position acrossgenerations and the inheritance level of Chileanfarm workers is higher than in the CASMINcountries These changes are expressed in theChilean-specific inheritance parameter IN2-CSecond an affinity parameter (AF1-C) wasadded to capture the difficulty of long-rangedownward elite mobility This negative asym-metrical affinity links flows from the service

class to the agricultural classes and from theroutine nonmanual class to the class of farmworkers Note that the AF1-C adds to the hier-archical distance between the two extremes ofthe class structure and to the negative affinitycaptured by AF1 to depict the extreme difficultyof long-range downward mobility from theChilean elite The fit of this lsquoChilean versionrsquoof the core model is reported in Table 2 model3 The improvement in BIC from model 2 is sub-stantial Furthermore even if the model fallsshort in terms of standard statistical criteria itaccounts for a large 91 of the associationunder independence thereby supporting theclaim that it adequately represents the Chileanmobility pattern The parameter estimates for theChilean version of the core are reported in row5 Table 3

Because of the core modelrsquos undesirable prop-erties I also use an alternative approach toexamine Chilean mobility patterns Thisapproach uses the now standard associationrowndashcolumn (II) (RC[II]) model (Goodman1979 see also Hout 1983 and Wong 1992) Inthis model both origin and destination classesare scaled so that the association can beexpressed as a linear-by-linear interaction by asingle parameter The multiplicative form ofthe model can be expressed as follows

Fij = iO j

D exp( microi νj)

where i indexes class of origin j indexes classof destination Fij is the expected frequency inthe (ij) cell is the grand mean i

O pertainsto the class of origin marginal effect j

D pertainsto the class of destination marginal effect isa global association parameter microi is the scalevalue for the ith class of origin and νj is the scalescore for the jth class of destination subject tothe following normalization constraints Σmicroi =Σνj = 0 (normalization of the location) and Σmicroi2

= Σνj2 = 1 (normalization of the scale)The class scale scores reflect a latent con-

tinuous variable made manifest by the class cat-egories Empirically obtained from the datathese ldquodistancerdquo scores produce an optimalranking of classes for the purpose of mobilityanalysis The RC(II) model is appropriate formodeling the association of any two ordinalvariables but mobility tables are distinctivebecause of the correspondence between class oforigin and class of destination (Gerber and Hout2004691)

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443377

2322-ASR 703 filename70304-torche

10 For instance during the socialist period inHungary the status of the proletariat may have beenindistinguishable from the status of the routine non-manual class (Szelenyi 199862) In Scotland givento early industrialization the hierarchical barrierbetween the skilled and unskilled working classes thatcharacterizes other CASMIN countries may not exist(Erikson and Goldthorpe 1992a163)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I introduce two adjustments to account forthis correspondence transforming the formu-lation into a ldquohybrid modelrdquo First topologicalparameters will account for the large counts insome diagonal cells reflecting immobility(inheritance) effects These parameters distin-guish two inheritance levels ldquohigh inheritancerdquofor the service class (I+II) farmers (IVc) andfarm workers (VIIb) and ldquolow inheritancerdquo forall other classes with the exception of the rou-tine nonmanual workers

Model 4 in Table 2 reports the fit statistic forthis quasi-RC(II) model The BIC statisticrejects it and prefers the more parsimoniousChilean version of the core model ThereforeI introduce a second adjustment to account forthe correspondence between rows and columnsin mobility tables Origin and destination scoresare constrained to be the same assuming thatthe scaling of classes has not significantlychanged over time Model 5 in Table 2 reportsfit statistics for this homogeneous quasi-RC(II)model On the basis of the BIC statistic the fitof the model is as good as that of the Chileanversion of the core model (Wong [1994] showsthat BIC differences of less than 5 points shouldbe considered indeterminate) Thus under stan-dard statistical criteria the Chilean core modeland the homogeneous quasi-RC(II) model areindistinguishable Given that the Chilean versionof the core model was tailored to fit the Chileantable the fit of model 5 is impressive and it indi-cates that a unidimensional scale can adequatelycapture mobility distances between classes

The logical next question is what this scalerepresents and specifically whether it repre-sents the hierarchical ranking of classes in termsof SES If a close correspondence is foundbetween the mobility distances among classesand their distances in terms of SES this find-ing would suggest that the intragenerationaldistribution of resources and rewards expressedin the SES ranking of classes drives the processof intergenerational mobility In other words thisfinding would support the resource approachrsquosclaim that differences in resources associatedwith diverse class positions determine mobili-ty opportunities across generations

To examine this possibility I compare theclass scores obtained from the homogeneousquasi-RC(II) model with the empiricallyobtained SES ranking of classes based on edu-cation and earnings (Appendix) Figure 4 pres-

ents the comparison between the standardizedvalues of both sets of scores The correspon-dence is striking The mobility distancesbetween classes closely reproduce their dis-tances in terms of SES This close isomorphismsuggests that at least in Chile the intragener-ational inequality in the distribution of socialresources and rewards across classes largelydrives the intergenerational mobility process Totest the isomorphism further I use the SESclass scores as row and column rankings to esti-mate a quasindashlinear-by-linear association model(model 6 Table 2) The fit of model 6 is sig-nificantly worse than that of model 5 almost aspoor as that of the original core model This indi-cates that although the inter- and intragenera-tional dimensions of inequality are very similarin Chile they have a somewhat different struc-ture

In summary the central characteristic of theChilean mobility regime is the predominance oflong-range hierarchical barriers and the weak-ness of sector cleavages separating classes in themiddle and at the lower end of the hierarchicalstructure Findings from the ldquoChilean versionrdquoof the core model and the quasi-RC(II) hybridmodels are consistent depicting a social struc-ture characterized by signif icant barriersbetween the top echelon and the rest of the classstructure The good fit of the homogeneousquasi-RC(II) model provides preliminary sup-port for the resource approach linking inequal-ity and mobility Mobility dynamics seem to bedriven largely by the hierarchical distancesbetween classes and overall Chile seems not tobe less fluid than the CASMIN countries

CCOOMMPPAARRIISSOONN OOFF TTHHEE OORRIIGGIINNAALL AANNDD CCHHIILLEEAANN

VVEERRSSIIOONNSS OOFF TTHHEE CCOORREE MMOODDEELL

Although Chilean mobility dynamics seem to bedriven by inequality across classes in a veryunequal society the fitting of the core modelprovides no indication that Chile is significantlyless fluid than depicted by the core model Asa preliminary analysis of the Chilean level of flu-idity I compare the predicted propensities formobility and immobility between specific pairsof classes in Chile with those in the CASMINcountries Propensities are obtained from theoriginal and Chilean versions of the core modelrespectively (models 2 and 5 of Table 3)

443388mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I begin with the immobility of those at the topof the class structure namely the service class(I+II) The service classrsquo propensity for immo-bility in the original core model is captured bytwo parameters IN1 and IN2 This propensityis more than three times what it would be in theabsence of these effects (e43+84 = 356) InChile the propensity for immobility of class I+IIis captured by the same two parameters and itis lower than what the core model expresses286 times larger than it would be in the absenceof these effects (e34+71) What about the immo-bility of the two agricultural classes In theCASMIN countries immobility of the IVc andVIIb classes is respectively 785 and 154 timeswhat it would be in the absence of these param-eters (e43+84+101ndash22 and e43 respectively) InChile the comparable values are 422 and 286(e34+71+48ndash09 and e43+71 respectively) indi-cating the relatively low inheritance associatedwith land ownership in the Latin Americannation Thus if attention is focused on theimmobility in the upper and lower ends of thesocial hierarchy the comparison does not indi-cate less fluidity in the highly unequal Chileansociety

What about the mobility rates between thetwo most distant classes in the hierarchical rank-ing namely the service class (I+II) and farmworkers (VIIb) In the case of the CASMINcountries the mobility chances between theseclasses are captured by four parameters (HI1HI2 SE1 and AF1) and is 077 times smallerthan it would be in the absence of these effects(endash22 ndash45ndash106ndash83) In Chile the distance fromclass I to class VIIb is expressed by these samefour parameters in addition to the asymmetri-cal disaffinity AF1-C This yields mobilitychances 056 times what would be obtained inthe absence of these effects (endash09 ndash47ndash20ndash63ndash149)Thus the barrier to long-range downwardmobility from the elite is more significant inChile than in the CASMIN countries The sameoccurs with long-range downward mobility fromclass III

If we measure the reciprocal flow (ieupward mobility from class VIIb to class I+II)we obtain the same value for the CASMINcountries because all parameters are symmet-rical in the original core model For Chile how-ever this calculation removes the asymmetricalbarrier to elite downward mobility which rais-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443399

2322-ASR 703 filename70304-torche

Figure 4 Class Scores Obtained from the Homogeneous Quasi-RC(II) Model and from Socioeconomic Status(SES) Ranking

Note (1) = Class standardized socioeconomic status scores based on schooling and earnings (2) = Class stan-dardized scores based on the Homogeneous Quasi-RC(II) model (see text for details)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 10: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

and to the growth and segmentation of the self-employed sector

The main objective of the Chilean agrarianreform (1962ndash1973) was to reduce the extremeland concentration in the country The reformwas very successful In 1955 the top 7 percentof landowners held 65 percent of the land andthe bottom 37 percent held only 1 percent By1973 the last year of the reform 43 percent ofthe land had been expropriated (Kay 2002 Scott1996) An agrarian counterreform was launchedby the military regime as a way of returning theland to its former owners It did not howeverrestore the traditional hacienda order (Gomezand Echenique 1988 Rivera 1988) In fact themilitary government returned only one-third ofthe plots to their old owners sold another thirdand adjudicated the remaining third to the smallproprietors benefited by the agrarian reformUnable to compete with the now cheap foodimports many small proprietors opted to selltheir plots As a consequence as much as two-thirds of the Chilean agricultural land came upfor sale creating an active land market Themain beneficiaries of the ldquomarketization of thecountrysiderdquo were a new group of export-ori-ented entrepreneurs and international investorswho bought large tracts of land (Gwynne 1996Kay 2002) Thus the reform and counterreformtriggered a massive ldquochange of handsrdquo in theChilean countryside likely altering the patternsof land inheritance

The second major effect of the market reformon the class structure was the reduction of for-mal employment caused by the decline of theindustrial working class and by the shrinkage ofthe state Incapable of competing with the nowcheap imports the working class plummetedfrom 34 percent of total employment in theearly 1970s to 20 percent in 1980 while pub-lic employment dropped from 14 to 84 per-cent in the same period (Leon and Martinez2000 Schkolnick 2000 Velasquez 1990) Thedecline in industrial and public employmentled to an increase in self-employment fromabout 15 percent in 1970 to about 28 percent in1980 (Thomas 1996) followed by a declinethrough the economic recovery before stabi-lization at about 22 percent of the total employ-ment in 2000 (MIDEPLAN 2001) In contrastto industrialized countries in which the self-employment rate usually is within the one-digitrange and involves capital ownership in Chile

it encompasses more than one-fifth of the laborforce Furthermore self-employment hadbecome a survival strategy for a large numberof former industrial and public employeesappropriately labeled ldquoforced entrepreneurial-ismrdquo (Infante and Klein 1995 Portes Castellsand Benton 1989) As a consequence theChilean self-employed sector is voluminousand segmented It includes a small segment ofentrepreneurs oriented to capital accumulationand able to hire employees and a large seg-ment of self-employed workers mostly engagedin ldquosurvival activitiesrdquo and unable to hire thelabor of others Differences in economic well-being between these two groups are massivewith the former group earning on averagethree times more than the latter (MIDEPLAN2001)

To be sure a large self-employed sector is nota result of the market transformation Self-employment has been historically high in LatinAmerica because according to some scholarsthe region has experienced ldquodependent inte-grationrdquo into the world capitalistic system(Hopkins and Wallerstein 1982 Luxembourg1951) However the internal heterogeneity ofthe self-employed sector increased during themarket reform in a pattern that foretold whathappened in the rest of Latin America during the1990s (Klein and Tokman 2000 Portes andHoffman 2003) and what may be current devel-opments in the industrialized world (Arum andMuumlller 2004 Noorderhaven et al 2003) Thesecharacteristics of the Chilean class structureare considered in the comparative analysis ofChilean mobility

DDAATTAA AANNDD AANNAALLYYTTIICCAALL AAPPPPRROOAACCHH

This study uses the 2001 Chilean MobilitySurvey (CMS) The CMS is a nationally repre-sentative multistage stratified sample of maleheads of household ages 24 to 69 The samplingstrategy includes the following stages First 87primary sampling units (PSUs) (counties) areselected Then blocks within the PSUs areselected and finally households within blocksare chosen Counties are stratified according tosize (fewer than 20000 20000 to 100000100000 to 200000 and more than 200000inhabitants) and by geographic zone (NorthCenter South) All PSUs in the large size stra-tum are included in the sample to increase effi-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443311

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ciency The fieldwork conducted between Apriland June 2001 consists of face-to-face inter-views in the respondentrsquos household carried outby trained personnel The survey excludesnonndashhead-of-household males which represent17 percent of the male population of the rele-vant age (MIDEPLAN 2001) Among thoseexcluded 865 percent are the sons of the headof the household 124 percent are another rel-ative of the head and 11 percent are other non-related males Their occupational distributionwith control used for age is almost identical tothat of the heads of household The small pro-portion represented by this group and their sim-ilar occupational distribution suggest that theirinclusion would not significantly alter the find-ings presented in this article Excluding thehouseholds not eligible for the survey theresponse rate is 63 percent Although nonre-sponse rates usually are not reported in Chileansurveys exchange with Chilean experts indi-cates that the nonresponse rate is about 20 to 25percent for face-to-face household surveys Thehigher nonresponse rate of the CMS is likelyattributable to the difficulty contacting maleheads of household4 The total sample size is3544 I exclude individuals outside the agerange of 25 to 64 years which is conventional-ly used in comparative mobility research andcases with unusable data After this exclusionthe usable sample size is 3002

CCLLAASSSS SSCCHHEEMMAA

This study uses a class perspective to describethe Chilean social structure It utilizes the seven-category version of the classification devel-oped by the CASMIN project Thisclassification is widely used in internationalcomparative research and describes the basic

stratification of advanced industrial societiesbased on ldquoemployment relationshipsrdquo (Eriksonand Goldthorpe 1992a35ndash47 Goldthorpe andHeath 1992) The seven classes included in theschema are I+II (service class) III (routine non-manual workers) IVab (self-employed work-ers) V+VI (manual supervisors and skilledmanual workers) VIIa (unskilled manual work-ers) IVc (farmers) and VIIb (farm workers)

Use of the 7-class categorization instead ofan alternative detailed 12-class version pro-duced by the CASMIN project (Erikson andGoldthorpe 1992b) induces some loss of infor-mation about the originndashdestination associa-tion A global test of all five aggregationsobtained from the ratio of L2 tests for the inde-pendence model in the collapsed and full tablesindicates that the seven-class schema masks 29percent of the association shown by the full setof classes ([L2 full table ndash L2 collapsed table]L2 full table = [108644 ndash 77191]108644 =29) This proportion is high compared withthat of the CASMIN countries and is surpassedonly by Sweden with 31 percent (Hout andHauser 1992 Table 2) A collapsed sevenfoldclass schema was however preferred because ofthe moderate sample size of the Chilean surveyand because it grants international comparabil-ity of the findings5

Operationalization of this class classifica-tion is based on detailed information concern-ing job title (recoded into the standard ISCO-88classification [ILO 1990]) industry occupa-tional status and supervisory status of workersBecause the CMS is specifically designed toassess class mobility it includes all the infor-mation necessary to produce the CASMIN classschema The CASMIN project has not produceda standard algorithm to generate the class clas-sification but other researchers have producedsuch an algorithm which I use in this analysis(Ganzeboom and Treiman 2003) Table 1 pres-ents the basic intergenerational mobility tableto be used in this analysis including the countsand marginal percentage distribution for originand destination classes

443322mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

4 The nonresponse rate can yield bias if those whowere unreachable or refused to participate differ fromthose included in the sample To estimate the mag-nitude of this bias I compared the distribution of keyvariables with the CASEN 2000 survey CASEN isa large survey (n = 252595) conducted by the Chileangovernment and has a refusal rate lower than 10percent The comparison (available upon request)suggests that the CMS slightly underrepresents agri-cultural workers and the upper class Overall how-ever there is no indication of major nonresponsebias

5 Analyses were replicated using the detailed 12-class table and no difference in substantive find-ings was found

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

MMEETTHHOODDSS

I fit two alternative models to the Chilean mobil-ity table The first is the ldquocore model of social flu-idityrdquo which claims to capture basic internationalsimilarity in mobility Although the core modelhas become a benchmark for mobility researchand currently is the only model that provides acomparative framework for mobility analysisanalysts have drawn attention to a number of itslimitations The model has been criticized forinadequately representing hierarchical mobilityeffects for being tailored to a few industrializedcountries for the ad hoc nature of some effects itincludes (specifically the affinity effects) andfor lacking a criterion to determine how muchdivergence is necessary for it to be rejected(Ganzeboom Luijkx and Treiman 1989 Houtand Hauser 1992 Sorensen 1992 Yamaguchi1987) I therefore also test an alternative ldquohybridmodelrdquo that combines the rowndashcolumn (II) asso-ciation model (Goodman 1979) with parametersaccounting for class immobility The internationaland temporal comparison of Chilean mobilitypatterns uses the uniform-difference (UNIDIFF)model also known as the multiplicative layereffect model (Erikson and Goldthorpe 1992aXie 1992) Model comparison and selection isbased on the Bayesian Information Criterion(BIC) statistic (Raftery 1995)

AANNAALLYYSSIISS OOFF CCHHIILLEEAANN MMOOBBIILLIITTYY IINNCCOOMMPPAARRAATTIIVVEE PPEERRSSPPEECCTTIIVVEE

TTHHEE CCOORREE MMOODDEELL OOFF SSOOCCIIAALL FFLLUUIIDDIITTYY

I start by fitting the core model to the Chileanmobility table The core model represents the

basic international commonality in social flu-idity using a topological formulation In contrastto standard topological models constructed froma single allocation of cells (Hauser 1978) thecore model uses eight matrices each of whichis designed to capture a particular effect thatenhances or reduces mobility between specificclasses (Erikson and Goldthorpe 1987a 1992achapter 4) In this way the model permits thesources of commonality across nations to beidentified clearly and likewise the departuresfrom it The effects are of four types hierarchyinheritance sector and affinity6

HIERARCHY EFFECTS Hierarchy effects reflectthe impact of status distances between classeson fluidity between them To estimate theseeffects the mobility table is divided into threestrata reflecting differences in resources andrewards across classes The strata are the fol-lowing upper stratum (class I+II) middle stra-tum (classes III IVab IVc and V-VI) and lowerstratum (classes VIIa and VIIb) Because thereare three strata two hierarchy effects (HI1 andHI2) can be identified each representing thecrossing of one additional hierarchical barrierThe association between each pair of classes isexpressed as an inverse function of the numberof hierarchical strata crossed

INHERITANCE EFFECTS Inheritance effects aredesigned to capture the propensity for class

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443333

2322-ASR 703 filename70304-torche

Table 1 Seven-Class Intergenerational Mobility Table for Chilean Men

Class of Destination

I-II III IVab IVc V-VI VIIa VIIb N

I-II Service Class 229 49 81 2 37 30 2 430 14III Routine Nonmanual 61 15 34 0 24 27 1 162 5IVab Self-employed 123 37 173 28 102 90 20 573 19IVc Farmers 42 16 68 54 58 54 24 316 11V-VI Skilled Manual 91 33 116 13 155 135 36 579 19VIIa Semiskilled and Unskilled Manual 49 43 115 13 111 123 44 498 17VIIb Farm Workers 28 16 78 17 89 101 115 444 15N 623 209 665 127 576 560 242 3002 21 7 22 4 19 19 8

Note Cell entries represent counts

6 Design matrices in Erikson and Goldthorpe(1992a124ndash129)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

immobility They include three matrices IN1identifies immobility across all classes alike IN2identifies the higher immobility of the serviceclass (I+II) the self-employed group (IVab) andfarmers (IVc) and IN3 accounts for the highestimmobility of farmers (IVc)

SECTOR EFFECTS Sector effects are intended toidentify the difficulty of moving between theagricultural and nonagricultural sectors of theeconomy (SE1)

AFFINITY EFFECTS Affinity effects are intend-ed to capture specific discontinuities (negativeaffinities) or linkages (positive affinities) betweenclasses which either reinforce or offset the over-all effects of hierarchy and sector There are twoaffinity effects The first AF1 identifies the dif-ficulty of moving between the service class (I+II)and the class of farm workers (VIIb) which addsto their hierarchical distance In contrast AF2identifies instances in which mobility is morefrequent than accounted for by hierarchy and sec-tor effects and includes affinities within the non-manual sector (I+II and III) and within the manualsector (V+VI and VIIa) a symmetrical affinitybased on capital possession (IVab and IVc and Iand IVab) and an asymmetrical link between agri-cultural classes of origin (IVc and VIIb) and theunskilled manual class of destination (VIIa)7

TTHHEE MMOOBBIILLIITTYY PPAATTTTEERRNN IINN CCHHIILLEECCOONNVVEERRGGEENNCCEE OORR DDIIVVEERRGGEENNCCEE

Table 2 presents the fit of several models appliedto the Chilean data and Table 3 displays theparameter estimates for each of these modelsModel 1 in Table 2 indicates that the fit of theoriginal core model to the Chilean table is unsat-isfactory under standard statistical criteria butit explains a significant 775 percent of associ-ation under independence To assess the strengthof the different factors driving mobility oppor-tunities I compare the Chilean coefficients withthose obtained for the CASMIN countriesCoefficients from the original core model fit-ted to the Chilean table are reported in row 1 ofTable 3 CASMIN coefficients are reported inrow 2 and differences in coefficientsrsquo magni-tude and the statistical significance of the dif-ferences are reported in row 38

All Chilean coefficients are significant andhave the same sign as for the CASMIN coun-tries Comparison of the magnitude of the coef-ficients however shows significant departuresfrom the core for five effects which revealinteresting differences between Chile and theCASMIN countries The hierarchy coefficientsindicate that whereas the short-range hierar-chical barriers are somewhat weaker in Chile

443344mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

8 CASMIN coefficients were obtained from fittingthe core model to the combined table of France andEngland standardizing both tables to have 10000cases as recommended by Erikson and Goldthorpe(1992a121) The coefficients obtained are virtuallyidentical to those presented by Erikson andGoldthorpe (1992a Table 44)

Table 2 Fit statistics selected models to the Chilean Mobility Table

AssociationL2 BIC df p Explained

Core Modelmdash1 Core Model Original 1375 285 28 000 775mdash2 Core Model EHE 1081 ndash08 28 000 860mdash3 Chilean version of Core Model ([Model 2] + Chilean Parameters) 724 ndash326 27 000 906Association Models 000mdash4 Heterogeneous Quasi-RC(II) Model 734 ndash161 23 000 905mdash5 Homogeneous Quasi-RC(II) Model 803 ndash287 28 000 896mdash6 Quasi-Linear by Linear Association (SES row and column ranking)a 1563 279 33 000 798

Note EHE = empirical hierarchical effects SES = eocioeconomic statusa Linear-by-linear association model using socioeconomic status scores as presented in Appendix Figure A1 torank origin and destination classes

7 For a detailed account of the empirical derivationof the model see Erikson and Goldthorpe (1987a1987b and 1992a chapter 4)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443355

2322-ASR 703 filename70304-torche

Tab

le 3

M

odel

Par

amet

er E

stim

ates

for

the

Chi

lean

Mob

ilit

y Ta

ble

Eff

ect

1 C

ore

mod

el o

rigi

nala

mdash 2 C

ore

mod

el C

AS

MIN

b

mdash 3 D

iffe

renc

e C

ASM

INmdash

Chi

lec

mdash 4 C

ore

mod

el E

HE

d

mdash 5 C

hile

an v

ersi

on o

f co

ree

mdash Not

eD

ata

show

n ar

e m

odel

par

amet

er e

stim

ates

wit

h st

anda

rd e

rror

s in

par

enth

eses

EH

E =

em

piri

cal h

iera

rchi

cal e

ffec

ts

aM

odel

1 f

rom

Tab

le 2

b

Ori

gina

l Cor

e M

odel

fit

ted

to th

e E

ngli

sh a

nd F

renc

h m

obil

ity

tabl

es S

ee te

xt f

or d

etai

ls

cD

iffe

renc

e in

par

amet

er m

agni

tude

bet

wee

n or

igin

al c

ore

mod

el f

itte

d to

the

Chi

lean

tabl

e (r

ow 1

) an

d to

the

CA

SM

IN ta

ble

(row

2)

dM

odel

2 f

rom

Tab

le 2

e

Mod

el 3

fro

m T

able

2

dagger p

lt 1

p

lt 0

5

p

lt 0

1

p

lt 0

01 (

two-

tail

ed)

HI1

ndash12

(

05)

ndash22

(02

)ndash

10 dagger

(06

)ndash

03(

07)

ndash09

dagger(

06)

HI2

ndash65

(11

)ndash

45

(

04)

20

dagger(

12)

ndash74

(09

)ndash

47

(

10)

IN1

41

(

08)

43

(

03)

Insi

g

37

(

07)

34

(

07)

IN2

39

(

10)

84

(

04)

45

(

11)

31

(13

)mdash

IN3

68

(

24)

101

(11

)In

sig

61

(24

)4

8(

23)

SE

1

ndash38

(07

)ndash1

06

(

05)

ndash68

(08

)ndash

42

(

07)

ndash20

(

08)

AF

1

ndash76

(21

)ndash

83

(

13)

Insi

g

ndash84

(20

)ndash

63

(21

)

AF

2

33

(

05)

45

(

02)

ndash12

(

06)

20

(

06)

26

(

06)

AF

1ndashC

mdash mdash mdash mdash

ndash14

9

(46

)

IN2ndash

C

mdash mdash mdash mdash 71

(

14)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(HI1) long-range hierarchical mobility is moredifficult in this country (HI2) This suggestssubstantial inequality between the extremes ofthe class hierarchy combined with smaller dif-ferentiation in the middle of the distribution Asto the inheritance effects the service class theself-employed group and the farmers (classesI+II IVab and IVc respectively) have a sig-nificantly lower propensity to immobility inChile than in the CASMIN countries suggest-ing that class positions usually associated withindependent work are less able to reproducetheir class status intergenerationally than depict-ed by the core Most impressive the barrierbetween the agricultural and nonagriculturalsectors of the economy (SE1) is much weakerin Chile as is the affinity effect linking non-manual classes manual classes and the class-es that own capital among themselves (AF2)

Interestingly all the significant differences incoefficients with the single exception of the bar-rier to long-range hierarchical mobility sug-gest that Chile is more fluid than the core modeldepicts Higher fluidity seems to be driven bythe weakness of horizontal barriers separatingthe agricultural manual and self-employed sec-tors of the economy This is consistent withfindings of weaker sector effects in countriesthat have experienced rapid industrializationsuch as Korea (Park 2004) Brazil (Costa-Ribeiro 2003) Israel (Yaish 2004 Goldthorpeet al 1997) and Japan (Ishida et al 1991)

Additionally in the Chilean case the weak-ness of sector barriers may have been intensi-fied by the deep market transformation of the1970s and 1980s On the one hand the ldquochangeof handsrdquo in the agricultural sector induced bythe agrarian reform and counterreform likelyincreased fluidity between the agricultural andurban classes On the other hand the role of self-employment as an ephemeral refuge against(industrial and public sector) unemploymentmay have reduced barriers between the self-employed group and the rest of the social struc-ture Added to the larger difficulty of long-rangehierarchical mobility these features suggestweak sector cleavages in the Chilean fluiditypattern and high salience of the hierarchicaldimension of mobility

For the countries in which the fit of the coremodel is not adequate Erikson and Goldthorpe(1992a145) suggest adding parameters thataccount for country-specific deviations from

the model I therefore introduce adjustments toreflect salient characteristics of the Chilean flu-idity pattern producing a ldquoChilean versionrdquo ofthe core model

First I modify the hierarchical effects Fittingof the core model suggests that hierarchicalbarriers are a crucial determinant of mobility inChile This conclusion is however obscuredby the fact that the hierarchical strata distin-guished by the core model are not empiricallyobtained and may not accurately represent theChilean ranking of classes in terms of socioe-conomic status (SES) To model hierarchicaleffects in Chile adequately I rank classes usingthe unweighted average of their schooling andearning levels as a proxy for SES I then collapsethe seven classes into three hierarchical stratausing cluster analysis These three empiricallyobtained strata are the following Upper stratum(service class I+II) middle stratum (routinenonmanual and self-employed group III andIVab) and lower stratum (manual and agricul-tural classes V-VI VIIa IVc and VIIb) Detailson the ranking of classes cluster procedures andthe comparison of the CASMIN and empiricalChilean class ranking can be found in theAppendix Comparison of the CASMIN and theempirical class ranking suggests higher differ-entiation at the top than at the bottom of thesocial structure in the Chilean case

Model 2 in Table 2 uses the empiricallyobtained hierarchical strata (EHE) Theimprovement in fit is massive (the difference inBIC is 293 without using any degrees of free-dom) The model now accounts for 86 percentof the association under independence

The large improvement in fit when empiricalhierarchical strata are used challenges theassumption of an internationally homogeneoushierarchical class ranking At the empiricallevel there is no proof that the hierarchicalranking of classes assumed by the core modelfits all CASMIN countries9 Moreover someevidence points to significant cross-country dif-

443366mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

9 Erikson and Goldthorpe(1992a Table 22) pres-ent the average ranking of classes based on standardstatus measures but they do not show the empiricalranking of classes for each country If the occupa-tional composition of classes varies across countriesthere may be significant international differences inthe hierarchical ranking of classes

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ferences in the ranking of classes (Hout andHauser 1992 Sorensen 1992) At the theoreti-cal level the creators of the core model high-light the relevance of highly specific historicallyformed national attributes to explain interna-tional variation in mobility These national par-ticularities may very well result in differenthierarchical positions of classes across coun-tries10 Furthermore even if the homogeneousranking of classes imposed by the CASMINresearchers applied to most CASMIN coun-tries significant departures may be found inthe late industrialized countries currently beingadded to the comparative mobility project TheChilean findings suggest that instead of impos-ing international homogeneity to the hierarchi-cal dimension of mobility the ranking of classesshould be treated as nationally specific param-eters to be estimated as mobility researchersroutinely treat row and column marginals ofthe mobility table The issue is importantbecause the hierarchical dimension is the mostrelevant dimension of mobility (Hout andHauser 1992 Wong 1992) and inadequate mod-eling can bias assessment of not only hierar-chical effects of mobility but also other effects

Two other changes were introduced toaccount for Chilean-specific deviations fromthe core model First the IN2 effect wasrearranged to exclude the self-employed class(IVab) and to include the class of farm workers(VIIb) This rearrangement accounts for thefact that the Chilean self-employed classappeared less likely than depicted by the coremodel to reproduce its class position acrossgenerations and the inheritance level of Chileanfarm workers is higher than in the CASMINcountries These changes are expressed in theChilean-specific inheritance parameter IN2-CSecond an affinity parameter (AF1-C) wasadded to capture the difficulty of long-rangedownward elite mobility This negative asym-metrical affinity links flows from the service

class to the agricultural classes and from theroutine nonmanual class to the class of farmworkers Note that the AF1-C adds to the hier-archical distance between the two extremes ofthe class structure and to the negative affinitycaptured by AF1 to depict the extreme difficultyof long-range downward mobility from theChilean elite The fit of this lsquoChilean versionrsquoof the core model is reported in Table 2 model3 The improvement in BIC from model 2 is sub-stantial Furthermore even if the model fallsshort in terms of standard statistical criteria itaccounts for a large 91 of the associationunder independence thereby supporting theclaim that it adequately represents the Chileanmobility pattern The parameter estimates for theChilean version of the core are reported in row5 Table 3

Because of the core modelrsquos undesirable prop-erties I also use an alternative approach toexamine Chilean mobility patterns Thisapproach uses the now standard associationrowndashcolumn (II) (RC[II]) model (Goodman1979 see also Hout 1983 and Wong 1992) Inthis model both origin and destination classesare scaled so that the association can beexpressed as a linear-by-linear interaction by asingle parameter The multiplicative form ofthe model can be expressed as follows

Fij = iO j

D exp( microi νj)

where i indexes class of origin j indexes classof destination Fij is the expected frequency inthe (ij) cell is the grand mean i

O pertainsto the class of origin marginal effect j

D pertainsto the class of destination marginal effect isa global association parameter microi is the scalevalue for the ith class of origin and νj is the scalescore for the jth class of destination subject tothe following normalization constraints Σmicroi =Σνj = 0 (normalization of the location) and Σmicroi2

= Σνj2 = 1 (normalization of the scale)The class scale scores reflect a latent con-

tinuous variable made manifest by the class cat-egories Empirically obtained from the datathese ldquodistancerdquo scores produce an optimalranking of classes for the purpose of mobilityanalysis The RC(II) model is appropriate formodeling the association of any two ordinalvariables but mobility tables are distinctivebecause of the correspondence between class oforigin and class of destination (Gerber and Hout2004691)

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443377

2322-ASR 703 filename70304-torche

10 For instance during the socialist period inHungary the status of the proletariat may have beenindistinguishable from the status of the routine non-manual class (Szelenyi 199862) In Scotland givento early industrialization the hierarchical barrierbetween the skilled and unskilled working classes thatcharacterizes other CASMIN countries may not exist(Erikson and Goldthorpe 1992a163)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I introduce two adjustments to account forthis correspondence transforming the formu-lation into a ldquohybrid modelrdquo First topologicalparameters will account for the large counts insome diagonal cells reflecting immobility(inheritance) effects These parameters distin-guish two inheritance levels ldquohigh inheritancerdquofor the service class (I+II) farmers (IVc) andfarm workers (VIIb) and ldquolow inheritancerdquo forall other classes with the exception of the rou-tine nonmanual workers

Model 4 in Table 2 reports the fit statistic forthis quasi-RC(II) model The BIC statisticrejects it and prefers the more parsimoniousChilean version of the core model ThereforeI introduce a second adjustment to account forthe correspondence between rows and columnsin mobility tables Origin and destination scoresare constrained to be the same assuming thatthe scaling of classes has not significantlychanged over time Model 5 in Table 2 reportsfit statistics for this homogeneous quasi-RC(II)model On the basis of the BIC statistic the fitof the model is as good as that of the Chileanversion of the core model (Wong [1994] showsthat BIC differences of less than 5 points shouldbe considered indeterminate) Thus under stan-dard statistical criteria the Chilean core modeland the homogeneous quasi-RC(II) model areindistinguishable Given that the Chilean versionof the core model was tailored to fit the Chileantable the fit of model 5 is impressive and it indi-cates that a unidimensional scale can adequatelycapture mobility distances between classes

The logical next question is what this scalerepresents and specifically whether it repre-sents the hierarchical ranking of classes in termsof SES If a close correspondence is foundbetween the mobility distances among classesand their distances in terms of SES this find-ing would suggest that the intragenerationaldistribution of resources and rewards expressedin the SES ranking of classes drives the processof intergenerational mobility In other words thisfinding would support the resource approachrsquosclaim that differences in resources associatedwith diverse class positions determine mobili-ty opportunities across generations

To examine this possibility I compare theclass scores obtained from the homogeneousquasi-RC(II) model with the empiricallyobtained SES ranking of classes based on edu-cation and earnings (Appendix) Figure 4 pres-

ents the comparison between the standardizedvalues of both sets of scores The correspon-dence is striking The mobility distancesbetween classes closely reproduce their dis-tances in terms of SES This close isomorphismsuggests that at least in Chile the intragener-ational inequality in the distribution of socialresources and rewards across classes largelydrives the intergenerational mobility process Totest the isomorphism further I use the SESclass scores as row and column rankings to esti-mate a quasindashlinear-by-linear association model(model 6 Table 2) The fit of model 6 is sig-nificantly worse than that of model 5 almost aspoor as that of the original core model This indi-cates that although the inter- and intragenera-tional dimensions of inequality are very similarin Chile they have a somewhat different struc-ture

In summary the central characteristic of theChilean mobility regime is the predominance oflong-range hierarchical barriers and the weak-ness of sector cleavages separating classes in themiddle and at the lower end of the hierarchicalstructure Findings from the ldquoChilean versionrdquoof the core model and the quasi-RC(II) hybridmodels are consistent depicting a social struc-ture characterized by signif icant barriersbetween the top echelon and the rest of the classstructure The good fit of the homogeneousquasi-RC(II) model provides preliminary sup-port for the resource approach linking inequal-ity and mobility Mobility dynamics seem to bedriven largely by the hierarchical distancesbetween classes and overall Chile seems not tobe less fluid than the CASMIN countries

CCOOMMPPAARRIISSOONN OOFF TTHHEE OORRIIGGIINNAALL AANNDD CCHHIILLEEAANN

VVEERRSSIIOONNSS OOFF TTHHEE CCOORREE MMOODDEELL

Although Chilean mobility dynamics seem to bedriven by inequality across classes in a veryunequal society the fitting of the core modelprovides no indication that Chile is significantlyless fluid than depicted by the core model Asa preliminary analysis of the Chilean level of flu-idity I compare the predicted propensities formobility and immobility between specific pairsof classes in Chile with those in the CASMINcountries Propensities are obtained from theoriginal and Chilean versions of the core modelrespectively (models 2 and 5 of Table 3)

443388mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I begin with the immobility of those at the topof the class structure namely the service class(I+II) The service classrsquo propensity for immo-bility in the original core model is captured bytwo parameters IN1 and IN2 This propensityis more than three times what it would be in theabsence of these effects (e43+84 = 356) InChile the propensity for immobility of class I+IIis captured by the same two parameters and itis lower than what the core model expresses286 times larger than it would be in the absenceof these effects (e34+71) What about the immo-bility of the two agricultural classes In theCASMIN countries immobility of the IVc andVIIb classes is respectively 785 and 154 timeswhat it would be in the absence of these param-eters (e43+84+101ndash22 and e43 respectively) InChile the comparable values are 422 and 286(e34+71+48ndash09 and e43+71 respectively) indi-cating the relatively low inheritance associatedwith land ownership in the Latin Americannation Thus if attention is focused on theimmobility in the upper and lower ends of thesocial hierarchy the comparison does not indi-cate less fluidity in the highly unequal Chileansociety

What about the mobility rates between thetwo most distant classes in the hierarchical rank-ing namely the service class (I+II) and farmworkers (VIIb) In the case of the CASMINcountries the mobility chances between theseclasses are captured by four parameters (HI1HI2 SE1 and AF1) and is 077 times smallerthan it would be in the absence of these effects(endash22 ndash45ndash106ndash83) In Chile the distance fromclass I to class VIIb is expressed by these samefour parameters in addition to the asymmetri-cal disaffinity AF1-C This yields mobilitychances 056 times what would be obtained inthe absence of these effects (endash09 ndash47ndash20ndash63ndash149)Thus the barrier to long-range downwardmobility from the elite is more significant inChile than in the CASMIN countries The sameoccurs with long-range downward mobility fromclass III

If we measure the reciprocal flow (ieupward mobility from class VIIb to class I+II)we obtain the same value for the CASMINcountries because all parameters are symmet-rical in the original core model For Chile how-ever this calculation removes the asymmetricalbarrier to elite downward mobility which rais-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443399

2322-ASR 703 filename70304-torche

Figure 4 Class Scores Obtained from the Homogeneous Quasi-RC(II) Model and from Socioeconomic Status(SES) Ranking

Note (1) = Class standardized socioeconomic status scores based on schooling and earnings (2) = Class stan-dardized scores based on the Homogeneous Quasi-RC(II) model (see text for details)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 11: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

ciency The fieldwork conducted between Apriland June 2001 consists of face-to-face inter-views in the respondentrsquos household carried outby trained personnel The survey excludesnonndashhead-of-household males which represent17 percent of the male population of the rele-vant age (MIDEPLAN 2001) Among thoseexcluded 865 percent are the sons of the headof the household 124 percent are another rel-ative of the head and 11 percent are other non-related males Their occupational distributionwith control used for age is almost identical tothat of the heads of household The small pro-portion represented by this group and their sim-ilar occupational distribution suggest that theirinclusion would not significantly alter the find-ings presented in this article Excluding thehouseholds not eligible for the survey theresponse rate is 63 percent Although nonre-sponse rates usually are not reported in Chileansurveys exchange with Chilean experts indi-cates that the nonresponse rate is about 20 to 25percent for face-to-face household surveys Thehigher nonresponse rate of the CMS is likelyattributable to the difficulty contacting maleheads of household4 The total sample size is3544 I exclude individuals outside the agerange of 25 to 64 years which is conventional-ly used in comparative mobility research andcases with unusable data After this exclusionthe usable sample size is 3002

CCLLAASSSS SSCCHHEEMMAA

This study uses a class perspective to describethe Chilean social structure It utilizes the seven-category version of the classification devel-oped by the CASMIN project Thisclassification is widely used in internationalcomparative research and describes the basic

stratification of advanced industrial societiesbased on ldquoemployment relationshipsrdquo (Eriksonand Goldthorpe 1992a35ndash47 Goldthorpe andHeath 1992) The seven classes included in theschema are I+II (service class) III (routine non-manual workers) IVab (self-employed work-ers) V+VI (manual supervisors and skilledmanual workers) VIIa (unskilled manual work-ers) IVc (farmers) and VIIb (farm workers)

Use of the 7-class categorization instead ofan alternative detailed 12-class version pro-duced by the CASMIN project (Erikson andGoldthorpe 1992b) induces some loss of infor-mation about the originndashdestination associa-tion A global test of all five aggregationsobtained from the ratio of L2 tests for the inde-pendence model in the collapsed and full tablesindicates that the seven-class schema masks 29percent of the association shown by the full setof classes ([L2 full table ndash L2 collapsed table]L2 full table = [108644 ndash 77191]108644 =29) This proportion is high compared withthat of the CASMIN countries and is surpassedonly by Sweden with 31 percent (Hout andHauser 1992 Table 2) A collapsed sevenfoldclass schema was however preferred because ofthe moderate sample size of the Chilean surveyand because it grants international comparabil-ity of the findings5

Operationalization of this class classifica-tion is based on detailed information concern-ing job title (recoded into the standard ISCO-88classification [ILO 1990]) industry occupa-tional status and supervisory status of workersBecause the CMS is specifically designed toassess class mobility it includes all the infor-mation necessary to produce the CASMIN classschema The CASMIN project has not produceda standard algorithm to generate the class clas-sification but other researchers have producedsuch an algorithm which I use in this analysis(Ganzeboom and Treiman 2003) Table 1 pres-ents the basic intergenerational mobility tableto be used in this analysis including the countsand marginal percentage distribution for originand destination classes

443322mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

4 The nonresponse rate can yield bias if those whowere unreachable or refused to participate differ fromthose included in the sample To estimate the mag-nitude of this bias I compared the distribution of keyvariables with the CASEN 2000 survey CASEN isa large survey (n = 252595) conducted by the Chileangovernment and has a refusal rate lower than 10percent The comparison (available upon request)suggests that the CMS slightly underrepresents agri-cultural workers and the upper class Overall how-ever there is no indication of major nonresponsebias

5 Analyses were replicated using the detailed 12-class table and no difference in substantive find-ings was found

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

MMEETTHHOODDSS

I fit two alternative models to the Chilean mobil-ity table The first is the ldquocore model of social flu-idityrdquo which claims to capture basic internationalsimilarity in mobility Although the core modelhas become a benchmark for mobility researchand currently is the only model that provides acomparative framework for mobility analysisanalysts have drawn attention to a number of itslimitations The model has been criticized forinadequately representing hierarchical mobilityeffects for being tailored to a few industrializedcountries for the ad hoc nature of some effects itincludes (specifically the affinity effects) andfor lacking a criterion to determine how muchdivergence is necessary for it to be rejected(Ganzeboom Luijkx and Treiman 1989 Houtand Hauser 1992 Sorensen 1992 Yamaguchi1987) I therefore also test an alternative ldquohybridmodelrdquo that combines the rowndashcolumn (II) asso-ciation model (Goodman 1979) with parametersaccounting for class immobility The internationaland temporal comparison of Chilean mobilitypatterns uses the uniform-difference (UNIDIFF)model also known as the multiplicative layereffect model (Erikson and Goldthorpe 1992aXie 1992) Model comparison and selection isbased on the Bayesian Information Criterion(BIC) statistic (Raftery 1995)

AANNAALLYYSSIISS OOFF CCHHIILLEEAANN MMOOBBIILLIITTYY IINNCCOOMMPPAARRAATTIIVVEE PPEERRSSPPEECCTTIIVVEE

TTHHEE CCOORREE MMOODDEELL OOFF SSOOCCIIAALL FFLLUUIIDDIITTYY

I start by fitting the core model to the Chileanmobility table The core model represents the

basic international commonality in social flu-idity using a topological formulation In contrastto standard topological models constructed froma single allocation of cells (Hauser 1978) thecore model uses eight matrices each of whichis designed to capture a particular effect thatenhances or reduces mobility between specificclasses (Erikson and Goldthorpe 1987a 1992achapter 4) In this way the model permits thesources of commonality across nations to beidentified clearly and likewise the departuresfrom it The effects are of four types hierarchyinheritance sector and affinity6

HIERARCHY EFFECTS Hierarchy effects reflectthe impact of status distances between classeson fluidity between them To estimate theseeffects the mobility table is divided into threestrata reflecting differences in resources andrewards across classes The strata are the fol-lowing upper stratum (class I+II) middle stra-tum (classes III IVab IVc and V-VI) and lowerstratum (classes VIIa and VIIb) Because thereare three strata two hierarchy effects (HI1 andHI2) can be identified each representing thecrossing of one additional hierarchical barrierThe association between each pair of classes isexpressed as an inverse function of the numberof hierarchical strata crossed

INHERITANCE EFFECTS Inheritance effects aredesigned to capture the propensity for class

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443333

2322-ASR 703 filename70304-torche

Table 1 Seven-Class Intergenerational Mobility Table for Chilean Men

Class of Destination

I-II III IVab IVc V-VI VIIa VIIb N

I-II Service Class 229 49 81 2 37 30 2 430 14III Routine Nonmanual 61 15 34 0 24 27 1 162 5IVab Self-employed 123 37 173 28 102 90 20 573 19IVc Farmers 42 16 68 54 58 54 24 316 11V-VI Skilled Manual 91 33 116 13 155 135 36 579 19VIIa Semiskilled and Unskilled Manual 49 43 115 13 111 123 44 498 17VIIb Farm Workers 28 16 78 17 89 101 115 444 15N 623 209 665 127 576 560 242 3002 21 7 22 4 19 19 8

Note Cell entries represent counts

6 Design matrices in Erikson and Goldthorpe(1992a124ndash129)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

immobility They include three matrices IN1identifies immobility across all classes alike IN2identifies the higher immobility of the serviceclass (I+II) the self-employed group (IVab) andfarmers (IVc) and IN3 accounts for the highestimmobility of farmers (IVc)

SECTOR EFFECTS Sector effects are intended toidentify the difficulty of moving between theagricultural and nonagricultural sectors of theeconomy (SE1)

AFFINITY EFFECTS Affinity effects are intend-ed to capture specific discontinuities (negativeaffinities) or linkages (positive affinities) betweenclasses which either reinforce or offset the over-all effects of hierarchy and sector There are twoaffinity effects The first AF1 identifies the dif-ficulty of moving between the service class (I+II)and the class of farm workers (VIIb) which addsto their hierarchical distance In contrast AF2identifies instances in which mobility is morefrequent than accounted for by hierarchy and sec-tor effects and includes affinities within the non-manual sector (I+II and III) and within the manualsector (V+VI and VIIa) a symmetrical affinitybased on capital possession (IVab and IVc and Iand IVab) and an asymmetrical link between agri-cultural classes of origin (IVc and VIIb) and theunskilled manual class of destination (VIIa)7

TTHHEE MMOOBBIILLIITTYY PPAATTTTEERRNN IINN CCHHIILLEECCOONNVVEERRGGEENNCCEE OORR DDIIVVEERRGGEENNCCEE

Table 2 presents the fit of several models appliedto the Chilean data and Table 3 displays theparameter estimates for each of these modelsModel 1 in Table 2 indicates that the fit of theoriginal core model to the Chilean table is unsat-isfactory under standard statistical criteria butit explains a significant 775 percent of associ-ation under independence To assess the strengthof the different factors driving mobility oppor-tunities I compare the Chilean coefficients withthose obtained for the CASMIN countriesCoefficients from the original core model fit-ted to the Chilean table are reported in row 1 ofTable 3 CASMIN coefficients are reported inrow 2 and differences in coefficientsrsquo magni-tude and the statistical significance of the dif-ferences are reported in row 38

All Chilean coefficients are significant andhave the same sign as for the CASMIN coun-tries Comparison of the magnitude of the coef-ficients however shows significant departuresfrom the core for five effects which revealinteresting differences between Chile and theCASMIN countries The hierarchy coefficientsindicate that whereas the short-range hierar-chical barriers are somewhat weaker in Chile

443344mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

8 CASMIN coefficients were obtained from fittingthe core model to the combined table of France andEngland standardizing both tables to have 10000cases as recommended by Erikson and Goldthorpe(1992a121) The coefficients obtained are virtuallyidentical to those presented by Erikson andGoldthorpe (1992a Table 44)

Table 2 Fit statistics selected models to the Chilean Mobility Table

AssociationL2 BIC df p Explained

Core Modelmdash1 Core Model Original 1375 285 28 000 775mdash2 Core Model EHE 1081 ndash08 28 000 860mdash3 Chilean version of Core Model ([Model 2] + Chilean Parameters) 724 ndash326 27 000 906Association Models 000mdash4 Heterogeneous Quasi-RC(II) Model 734 ndash161 23 000 905mdash5 Homogeneous Quasi-RC(II) Model 803 ndash287 28 000 896mdash6 Quasi-Linear by Linear Association (SES row and column ranking)a 1563 279 33 000 798

Note EHE = empirical hierarchical effects SES = eocioeconomic statusa Linear-by-linear association model using socioeconomic status scores as presented in Appendix Figure A1 torank origin and destination classes

7 For a detailed account of the empirical derivationof the model see Erikson and Goldthorpe (1987a1987b and 1992a chapter 4)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443355

2322-ASR 703 filename70304-torche

Tab

le 3

M

odel

Par

amet

er E

stim

ates

for

the

Chi

lean

Mob

ilit

y Ta

ble

Eff

ect

1 C

ore

mod

el o

rigi

nala

mdash 2 C

ore

mod

el C

AS

MIN

b

mdash 3 D

iffe

renc

e C

ASM

INmdash

Chi

lec

mdash 4 C

ore

mod

el E

HE

d

mdash 5 C

hile

an v

ersi

on o

f co

ree

mdash Not

eD

ata

show

n ar

e m

odel

par

amet

er e

stim

ates

wit

h st

anda

rd e

rror

s in

par

enth

eses

EH

E =

em

piri

cal h

iera

rchi

cal e

ffec

ts

aM

odel

1 f

rom

Tab

le 2

b

Ori

gina

l Cor

e M

odel

fit

ted

to th

e E

ngli

sh a

nd F

renc

h m

obil

ity

tabl

es S

ee te

xt f

or d

etai

ls

cD

iffe

renc

e in

par

amet

er m

agni

tude

bet

wee

n or

igin

al c

ore

mod

el f

itte

d to

the

Chi

lean

tabl

e (r

ow 1

) an

d to

the

CA

SM

IN ta

ble

(row

2)

dM

odel

2 f

rom

Tab

le 2

e

Mod

el 3

fro

m T

able

2

dagger p

lt 1

p

lt 0

5

p

lt 0

1

p

lt 0

01 (

two-

tail

ed)

HI1

ndash12

(

05)

ndash22

(02

)ndash

10 dagger

(06

)ndash

03(

07)

ndash09

dagger(

06)

HI2

ndash65

(11

)ndash

45

(

04)

20

dagger(

12)

ndash74

(09

)ndash

47

(

10)

IN1

41

(

08)

43

(

03)

Insi

g

37

(

07)

34

(

07)

IN2

39

(

10)

84

(

04)

45

(

11)

31

(13

)mdash

IN3

68

(

24)

101

(11

)In

sig

61

(24

)4

8(

23)

SE

1

ndash38

(07

)ndash1

06

(

05)

ndash68

(08

)ndash

42

(

07)

ndash20

(

08)

AF

1

ndash76

(21

)ndash

83

(

13)

Insi

g

ndash84

(20

)ndash

63

(21

)

AF

2

33

(

05)

45

(

02)

ndash12

(

06)

20

(

06)

26

(

06)

AF

1ndashC

mdash mdash mdash mdash

ndash14

9

(46

)

IN2ndash

C

mdash mdash mdash mdash 71

(

14)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(HI1) long-range hierarchical mobility is moredifficult in this country (HI2) This suggestssubstantial inequality between the extremes ofthe class hierarchy combined with smaller dif-ferentiation in the middle of the distribution Asto the inheritance effects the service class theself-employed group and the farmers (classesI+II IVab and IVc respectively) have a sig-nificantly lower propensity to immobility inChile than in the CASMIN countries suggest-ing that class positions usually associated withindependent work are less able to reproducetheir class status intergenerationally than depict-ed by the core Most impressive the barrierbetween the agricultural and nonagriculturalsectors of the economy (SE1) is much weakerin Chile as is the affinity effect linking non-manual classes manual classes and the class-es that own capital among themselves (AF2)

Interestingly all the significant differences incoefficients with the single exception of the bar-rier to long-range hierarchical mobility sug-gest that Chile is more fluid than the core modeldepicts Higher fluidity seems to be driven bythe weakness of horizontal barriers separatingthe agricultural manual and self-employed sec-tors of the economy This is consistent withfindings of weaker sector effects in countriesthat have experienced rapid industrializationsuch as Korea (Park 2004) Brazil (Costa-Ribeiro 2003) Israel (Yaish 2004 Goldthorpeet al 1997) and Japan (Ishida et al 1991)

Additionally in the Chilean case the weak-ness of sector barriers may have been intensi-fied by the deep market transformation of the1970s and 1980s On the one hand the ldquochangeof handsrdquo in the agricultural sector induced bythe agrarian reform and counterreform likelyincreased fluidity between the agricultural andurban classes On the other hand the role of self-employment as an ephemeral refuge against(industrial and public sector) unemploymentmay have reduced barriers between the self-employed group and the rest of the social struc-ture Added to the larger difficulty of long-rangehierarchical mobility these features suggestweak sector cleavages in the Chilean fluiditypattern and high salience of the hierarchicaldimension of mobility

For the countries in which the fit of the coremodel is not adequate Erikson and Goldthorpe(1992a145) suggest adding parameters thataccount for country-specific deviations from

the model I therefore introduce adjustments toreflect salient characteristics of the Chilean flu-idity pattern producing a ldquoChilean versionrdquo ofthe core model

First I modify the hierarchical effects Fittingof the core model suggests that hierarchicalbarriers are a crucial determinant of mobility inChile This conclusion is however obscuredby the fact that the hierarchical strata distin-guished by the core model are not empiricallyobtained and may not accurately represent theChilean ranking of classes in terms of socioe-conomic status (SES) To model hierarchicaleffects in Chile adequately I rank classes usingthe unweighted average of their schooling andearning levels as a proxy for SES I then collapsethe seven classes into three hierarchical stratausing cluster analysis These three empiricallyobtained strata are the following Upper stratum(service class I+II) middle stratum (routinenonmanual and self-employed group III andIVab) and lower stratum (manual and agricul-tural classes V-VI VIIa IVc and VIIb) Detailson the ranking of classes cluster procedures andthe comparison of the CASMIN and empiricalChilean class ranking can be found in theAppendix Comparison of the CASMIN and theempirical class ranking suggests higher differ-entiation at the top than at the bottom of thesocial structure in the Chilean case

Model 2 in Table 2 uses the empiricallyobtained hierarchical strata (EHE) Theimprovement in fit is massive (the difference inBIC is 293 without using any degrees of free-dom) The model now accounts for 86 percentof the association under independence

The large improvement in fit when empiricalhierarchical strata are used challenges theassumption of an internationally homogeneoushierarchical class ranking At the empiricallevel there is no proof that the hierarchicalranking of classes assumed by the core modelfits all CASMIN countries9 Moreover someevidence points to significant cross-country dif-

443366mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

9 Erikson and Goldthorpe(1992a Table 22) pres-ent the average ranking of classes based on standardstatus measures but they do not show the empiricalranking of classes for each country If the occupa-tional composition of classes varies across countriesthere may be significant international differences inthe hierarchical ranking of classes

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ferences in the ranking of classes (Hout andHauser 1992 Sorensen 1992) At the theoreti-cal level the creators of the core model high-light the relevance of highly specific historicallyformed national attributes to explain interna-tional variation in mobility These national par-ticularities may very well result in differenthierarchical positions of classes across coun-tries10 Furthermore even if the homogeneousranking of classes imposed by the CASMINresearchers applied to most CASMIN coun-tries significant departures may be found inthe late industrialized countries currently beingadded to the comparative mobility project TheChilean findings suggest that instead of impos-ing international homogeneity to the hierarchi-cal dimension of mobility the ranking of classesshould be treated as nationally specific param-eters to be estimated as mobility researchersroutinely treat row and column marginals ofthe mobility table The issue is importantbecause the hierarchical dimension is the mostrelevant dimension of mobility (Hout andHauser 1992 Wong 1992) and inadequate mod-eling can bias assessment of not only hierar-chical effects of mobility but also other effects

Two other changes were introduced toaccount for Chilean-specific deviations fromthe core model First the IN2 effect wasrearranged to exclude the self-employed class(IVab) and to include the class of farm workers(VIIb) This rearrangement accounts for thefact that the Chilean self-employed classappeared less likely than depicted by the coremodel to reproduce its class position acrossgenerations and the inheritance level of Chileanfarm workers is higher than in the CASMINcountries These changes are expressed in theChilean-specific inheritance parameter IN2-CSecond an affinity parameter (AF1-C) wasadded to capture the difficulty of long-rangedownward elite mobility This negative asym-metrical affinity links flows from the service

class to the agricultural classes and from theroutine nonmanual class to the class of farmworkers Note that the AF1-C adds to the hier-archical distance between the two extremes ofthe class structure and to the negative affinitycaptured by AF1 to depict the extreme difficultyof long-range downward mobility from theChilean elite The fit of this lsquoChilean versionrsquoof the core model is reported in Table 2 model3 The improvement in BIC from model 2 is sub-stantial Furthermore even if the model fallsshort in terms of standard statistical criteria itaccounts for a large 91 of the associationunder independence thereby supporting theclaim that it adequately represents the Chileanmobility pattern The parameter estimates for theChilean version of the core are reported in row5 Table 3

Because of the core modelrsquos undesirable prop-erties I also use an alternative approach toexamine Chilean mobility patterns Thisapproach uses the now standard associationrowndashcolumn (II) (RC[II]) model (Goodman1979 see also Hout 1983 and Wong 1992) Inthis model both origin and destination classesare scaled so that the association can beexpressed as a linear-by-linear interaction by asingle parameter The multiplicative form ofthe model can be expressed as follows

Fij = iO j

D exp( microi νj)

where i indexes class of origin j indexes classof destination Fij is the expected frequency inthe (ij) cell is the grand mean i

O pertainsto the class of origin marginal effect j

D pertainsto the class of destination marginal effect isa global association parameter microi is the scalevalue for the ith class of origin and νj is the scalescore for the jth class of destination subject tothe following normalization constraints Σmicroi =Σνj = 0 (normalization of the location) and Σmicroi2

= Σνj2 = 1 (normalization of the scale)The class scale scores reflect a latent con-

tinuous variable made manifest by the class cat-egories Empirically obtained from the datathese ldquodistancerdquo scores produce an optimalranking of classes for the purpose of mobilityanalysis The RC(II) model is appropriate formodeling the association of any two ordinalvariables but mobility tables are distinctivebecause of the correspondence between class oforigin and class of destination (Gerber and Hout2004691)

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443377

2322-ASR 703 filename70304-torche

10 For instance during the socialist period inHungary the status of the proletariat may have beenindistinguishable from the status of the routine non-manual class (Szelenyi 199862) In Scotland givento early industrialization the hierarchical barrierbetween the skilled and unskilled working classes thatcharacterizes other CASMIN countries may not exist(Erikson and Goldthorpe 1992a163)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I introduce two adjustments to account forthis correspondence transforming the formu-lation into a ldquohybrid modelrdquo First topologicalparameters will account for the large counts insome diagonal cells reflecting immobility(inheritance) effects These parameters distin-guish two inheritance levels ldquohigh inheritancerdquofor the service class (I+II) farmers (IVc) andfarm workers (VIIb) and ldquolow inheritancerdquo forall other classes with the exception of the rou-tine nonmanual workers

Model 4 in Table 2 reports the fit statistic forthis quasi-RC(II) model The BIC statisticrejects it and prefers the more parsimoniousChilean version of the core model ThereforeI introduce a second adjustment to account forthe correspondence between rows and columnsin mobility tables Origin and destination scoresare constrained to be the same assuming thatthe scaling of classes has not significantlychanged over time Model 5 in Table 2 reportsfit statistics for this homogeneous quasi-RC(II)model On the basis of the BIC statistic the fitof the model is as good as that of the Chileanversion of the core model (Wong [1994] showsthat BIC differences of less than 5 points shouldbe considered indeterminate) Thus under stan-dard statistical criteria the Chilean core modeland the homogeneous quasi-RC(II) model areindistinguishable Given that the Chilean versionof the core model was tailored to fit the Chileantable the fit of model 5 is impressive and it indi-cates that a unidimensional scale can adequatelycapture mobility distances between classes

The logical next question is what this scalerepresents and specifically whether it repre-sents the hierarchical ranking of classes in termsof SES If a close correspondence is foundbetween the mobility distances among classesand their distances in terms of SES this find-ing would suggest that the intragenerationaldistribution of resources and rewards expressedin the SES ranking of classes drives the processof intergenerational mobility In other words thisfinding would support the resource approachrsquosclaim that differences in resources associatedwith diverse class positions determine mobili-ty opportunities across generations

To examine this possibility I compare theclass scores obtained from the homogeneousquasi-RC(II) model with the empiricallyobtained SES ranking of classes based on edu-cation and earnings (Appendix) Figure 4 pres-

ents the comparison between the standardizedvalues of both sets of scores The correspon-dence is striking The mobility distancesbetween classes closely reproduce their dis-tances in terms of SES This close isomorphismsuggests that at least in Chile the intragener-ational inequality in the distribution of socialresources and rewards across classes largelydrives the intergenerational mobility process Totest the isomorphism further I use the SESclass scores as row and column rankings to esti-mate a quasindashlinear-by-linear association model(model 6 Table 2) The fit of model 6 is sig-nificantly worse than that of model 5 almost aspoor as that of the original core model This indi-cates that although the inter- and intragenera-tional dimensions of inequality are very similarin Chile they have a somewhat different struc-ture

In summary the central characteristic of theChilean mobility regime is the predominance oflong-range hierarchical barriers and the weak-ness of sector cleavages separating classes in themiddle and at the lower end of the hierarchicalstructure Findings from the ldquoChilean versionrdquoof the core model and the quasi-RC(II) hybridmodels are consistent depicting a social struc-ture characterized by signif icant barriersbetween the top echelon and the rest of the classstructure The good fit of the homogeneousquasi-RC(II) model provides preliminary sup-port for the resource approach linking inequal-ity and mobility Mobility dynamics seem to bedriven largely by the hierarchical distancesbetween classes and overall Chile seems not tobe less fluid than the CASMIN countries

CCOOMMPPAARRIISSOONN OOFF TTHHEE OORRIIGGIINNAALL AANNDD CCHHIILLEEAANN

VVEERRSSIIOONNSS OOFF TTHHEE CCOORREE MMOODDEELL

Although Chilean mobility dynamics seem to bedriven by inequality across classes in a veryunequal society the fitting of the core modelprovides no indication that Chile is significantlyless fluid than depicted by the core model Asa preliminary analysis of the Chilean level of flu-idity I compare the predicted propensities formobility and immobility between specific pairsof classes in Chile with those in the CASMINcountries Propensities are obtained from theoriginal and Chilean versions of the core modelrespectively (models 2 and 5 of Table 3)

443388mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I begin with the immobility of those at the topof the class structure namely the service class(I+II) The service classrsquo propensity for immo-bility in the original core model is captured bytwo parameters IN1 and IN2 This propensityis more than three times what it would be in theabsence of these effects (e43+84 = 356) InChile the propensity for immobility of class I+IIis captured by the same two parameters and itis lower than what the core model expresses286 times larger than it would be in the absenceof these effects (e34+71) What about the immo-bility of the two agricultural classes In theCASMIN countries immobility of the IVc andVIIb classes is respectively 785 and 154 timeswhat it would be in the absence of these param-eters (e43+84+101ndash22 and e43 respectively) InChile the comparable values are 422 and 286(e34+71+48ndash09 and e43+71 respectively) indi-cating the relatively low inheritance associatedwith land ownership in the Latin Americannation Thus if attention is focused on theimmobility in the upper and lower ends of thesocial hierarchy the comparison does not indi-cate less fluidity in the highly unequal Chileansociety

What about the mobility rates between thetwo most distant classes in the hierarchical rank-ing namely the service class (I+II) and farmworkers (VIIb) In the case of the CASMINcountries the mobility chances between theseclasses are captured by four parameters (HI1HI2 SE1 and AF1) and is 077 times smallerthan it would be in the absence of these effects(endash22 ndash45ndash106ndash83) In Chile the distance fromclass I to class VIIb is expressed by these samefour parameters in addition to the asymmetri-cal disaffinity AF1-C This yields mobilitychances 056 times what would be obtained inthe absence of these effects (endash09 ndash47ndash20ndash63ndash149)Thus the barrier to long-range downwardmobility from the elite is more significant inChile than in the CASMIN countries The sameoccurs with long-range downward mobility fromclass III

If we measure the reciprocal flow (ieupward mobility from class VIIb to class I+II)we obtain the same value for the CASMINcountries because all parameters are symmet-rical in the original core model For Chile how-ever this calculation removes the asymmetricalbarrier to elite downward mobility which rais-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443399

2322-ASR 703 filename70304-torche

Figure 4 Class Scores Obtained from the Homogeneous Quasi-RC(II) Model and from Socioeconomic Status(SES) Ranking

Note (1) = Class standardized socioeconomic status scores based on schooling and earnings (2) = Class stan-dardized scores based on the Homogeneous Quasi-RC(II) model (see text for details)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 12: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

MMEETTHHOODDSS

I fit two alternative models to the Chilean mobil-ity table The first is the ldquocore model of social flu-idityrdquo which claims to capture basic internationalsimilarity in mobility Although the core modelhas become a benchmark for mobility researchand currently is the only model that provides acomparative framework for mobility analysisanalysts have drawn attention to a number of itslimitations The model has been criticized forinadequately representing hierarchical mobilityeffects for being tailored to a few industrializedcountries for the ad hoc nature of some effects itincludes (specifically the affinity effects) andfor lacking a criterion to determine how muchdivergence is necessary for it to be rejected(Ganzeboom Luijkx and Treiman 1989 Houtand Hauser 1992 Sorensen 1992 Yamaguchi1987) I therefore also test an alternative ldquohybridmodelrdquo that combines the rowndashcolumn (II) asso-ciation model (Goodman 1979) with parametersaccounting for class immobility The internationaland temporal comparison of Chilean mobilitypatterns uses the uniform-difference (UNIDIFF)model also known as the multiplicative layereffect model (Erikson and Goldthorpe 1992aXie 1992) Model comparison and selection isbased on the Bayesian Information Criterion(BIC) statistic (Raftery 1995)

AANNAALLYYSSIISS OOFF CCHHIILLEEAANN MMOOBBIILLIITTYY IINNCCOOMMPPAARRAATTIIVVEE PPEERRSSPPEECCTTIIVVEE

TTHHEE CCOORREE MMOODDEELL OOFF SSOOCCIIAALL FFLLUUIIDDIITTYY

I start by fitting the core model to the Chileanmobility table The core model represents the

basic international commonality in social flu-idity using a topological formulation In contrastto standard topological models constructed froma single allocation of cells (Hauser 1978) thecore model uses eight matrices each of whichis designed to capture a particular effect thatenhances or reduces mobility between specificclasses (Erikson and Goldthorpe 1987a 1992achapter 4) In this way the model permits thesources of commonality across nations to beidentified clearly and likewise the departuresfrom it The effects are of four types hierarchyinheritance sector and affinity6

HIERARCHY EFFECTS Hierarchy effects reflectthe impact of status distances between classeson fluidity between them To estimate theseeffects the mobility table is divided into threestrata reflecting differences in resources andrewards across classes The strata are the fol-lowing upper stratum (class I+II) middle stra-tum (classes III IVab IVc and V-VI) and lowerstratum (classes VIIa and VIIb) Because thereare three strata two hierarchy effects (HI1 andHI2) can be identified each representing thecrossing of one additional hierarchical barrierThe association between each pair of classes isexpressed as an inverse function of the numberof hierarchical strata crossed

INHERITANCE EFFECTS Inheritance effects aredesigned to capture the propensity for class

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443333

2322-ASR 703 filename70304-torche

Table 1 Seven-Class Intergenerational Mobility Table for Chilean Men

Class of Destination

I-II III IVab IVc V-VI VIIa VIIb N

I-II Service Class 229 49 81 2 37 30 2 430 14III Routine Nonmanual 61 15 34 0 24 27 1 162 5IVab Self-employed 123 37 173 28 102 90 20 573 19IVc Farmers 42 16 68 54 58 54 24 316 11V-VI Skilled Manual 91 33 116 13 155 135 36 579 19VIIa Semiskilled and Unskilled Manual 49 43 115 13 111 123 44 498 17VIIb Farm Workers 28 16 78 17 89 101 115 444 15N 623 209 665 127 576 560 242 3002 21 7 22 4 19 19 8

Note Cell entries represent counts

6 Design matrices in Erikson and Goldthorpe(1992a124ndash129)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

immobility They include three matrices IN1identifies immobility across all classes alike IN2identifies the higher immobility of the serviceclass (I+II) the self-employed group (IVab) andfarmers (IVc) and IN3 accounts for the highestimmobility of farmers (IVc)

SECTOR EFFECTS Sector effects are intended toidentify the difficulty of moving between theagricultural and nonagricultural sectors of theeconomy (SE1)

AFFINITY EFFECTS Affinity effects are intend-ed to capture specific discontinuities (negativeaffinities) or linkages (positive affinities) betweenclasses which either reinforce or offset the over-all effects of hierarchy and sector There are twoaffinity effects The first AF1 identifies the dif-ficulty of moving between the service class (I+II)and the class of farm workers (VIIb) which addsto their hierarchical distance In contrast AF2identifies instances in which mobility is morefrequent than accounted for by hierarchy and sec-tor effects and includes affinities within the non-manual sector (I+II and III) and within the manualsector (V+VI and VIIa) a symmetrical affinitybased on capital possession (IVab and IVc and Iand IVab) and an asymmetrical link between agri-cultural classes of origin (IVc and VIIb) and theunskilled manual class of destination (VIIa)7

TTHHEE MMOOBBIILLIITTYY PPAATTTTEERRNN IINN CCHHIILLEECCOONNVVEERRGGEENNCCEE OORR DDIIVVEERRGGEENNCCEE

Table 2 presents the fit of several models appliedto the Chilean data and Table 3 displays theparameter estimates for each of these modelsModel 1 in Table 2 indicates that the fit of theoriginal core model to the Chilean table is unsat-isfactory under standard statistical criteria butit explains a significant 775 percent of associ-ation under independence To assess the strengthof the different factors driving mobility oppor-tunities I compare the Chilean coefficients withthose obtained for the CASMIN countriesCoefficients from the original core model fit-ted to the Chilean table are reported in row 1 ofTable 3 CASMIN coefficients are reported inrow 2 and differences in coefficientsrsquo magni-tude and the statistical significance of the dif-ferences are reported in row 38

All Chilean coefficients are significant andhave the same sign as for the CASMIN coun-tries Comparison of the magnitude of the coef-ficients however shows significant departuresfrom the core for five effects which revealinteresting differences between Chile and theCASMIN countries The hierarchy coefficientsindicate that whereas the short-range hierar-chical barriers are somewhat weaker in Chile

443344mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

8 CASMIN coefficients were obtained from fittingthe core model to the combined table of France andEngland standardizing both tables to have 10000cases as recommended by Erikson and Goldthorpe(1992a121) The coefficients obtained are virtuallyidentical to those presented by Erikson andGoldthorpe (1992a Table 44)

Table 2 Fit statistics selected models to the Chilean Mobility Table

AssociationL2 BIC df p Explained

Core Modelmdash1 Core Model Original 1375 285 28 000 775mdash2 Core Model EHE 1081 ndash08 28 000 860mdash3 Chilean version of Core Model ([Model 2] + Chilean Parameters) 724 ndash326 27 000 906Association Models 000mdash4 Heterogeneous Quasi-RC(II) Model 734 ndash161 23 000 905mdash5 Homogeneous Quasi-RC(II) Model 803 ndash287 28 000 896mdash6 Quasi-Linear by Linear Association (SES row and column ranking)a 1563 279 33 000 798

Note EHE = empirical hierarchical effects SES = eocioeconomic statusa Linear-by-linear association model using socioeconomic status scores as presented in Appendix Figure A1 torank origin and destination classes

7 For a detailed account of the empirical derivationof the model see Erikson and Goldthorpe (1987a1987b and 1992a chapter 4)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443355

2322-ASR 703 filename70304-torche

Tab

le 3

M

odel

Par

amet

er E

stim

ates

for

the

Chi

lean

Mob

ilit

y Ta

ble

Eff

ect

1 C

ore

mod

el o

rigi

nala

mdash 2 C

ore

mod

el C

AS

MIN

b

mdash 3 D

iffe

renc

e C

ASM

INmdash

Chi

lec

mdash 4 C

ore

mod

el E

HE

d

mdash 5 C

hile

an v

ersi

on o

f co

ree

mdash Not

eD

ata

show

n ar

e m

odel

par

amet

er e

stim

ates

wit

h st

anda

rd e

rror

s in

par

enth

eses

EH

E =

em

piri

cal h

iera

rchi

cal e

ffec

ts

aM

odel

1 f

rom

Tab

le 2

b

Ori

gina

l Cor

e M

odel

fit

ted

to th

e E

ngli

sh a

nd F

renc

h m

obil

ity

tabl

es S

ee te

xt f

or d

etai

ls

cD

iffe

renc

e in

par

amet

er m

agni

tude

bet

wee

n or

igin

al c

ore

mod

el f

itte

d to

the

Chi

lean

tabl

e (r

ow 1

) an

d to

the

CA

SM

IN ta

ble

(row

2)

dM

odel

2 f

rom

Tab

le 2

e

Mod

el 3

fro

m T

able

2

dagger p

lt 1

p

lt 0

5

p

lt 0

1

p

lt 0

01 (

two-

tail

ed)

HI1

ndash12

(

05)

ndash22

(02

)ndash

10 dagger

(06

)ndash

03(

07)

ndash09

dagger(

06)

HI2

ndash65

(11

)ndash

45

(

04)

20

dagger(

12)

ndash74

(09

)ndash

47

(

10)

IN1

41

(

08)

43

(

03)

Insi

g

37

(

07)

34

(

07)

IN2

39

(

10)

84

(

04)

45

(

11)

31

(13

)mdash

IN3

68

(

24)

101

(11

)In

sig

61

(24

)4

8(

23)

SE

1

ndash38

(07

)ndash1

06

(

05)

ndash68

(08

)ndash

42

(

07)

ndash20

(

08)

AF

1

ndash76

(21

)ndash

83

(

13)

Insi

g

ndash84

(20

)ndash

63

(21

)

AF

2

33

(

05)

45

(

02)

ndash12

(

06)

20

(

06)

26

(

06)

AF

1ndashC

mdash mdash mdash mdash

ndash14

9

(46

)

IN2ndash

C

mdash mdash mdash mdash 71

(

14)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(HI1) long-range hierarchical mobility is moredifficult in this country (HI2) This suggestssubstantial inequality between the extremes ofthe class hierarchy combined with smaller dif-ferentiation in the middle of the distribution Asto the inheritance effects the service class theself-employed group and the farmers (classesI+II IVab and IVc respectively) have a sig-nificantly lower propensity to immobility inChile than in the CASMIN countries suggest-ing that class positions usually associated withindependent work are less able to reproducetheir class status intergenerationally than depict-ed by the core Most impressive the barrierbetween the agricultural and nonagriculturalsectors of the economy (SE1) is much weakerin Chile as is the affinity effect linking non-manual classes manual classes and the class-es that own capital among themselves (AF2)

Interestingly all the significant differences incoefficients with the single exception of the bar-rier to long-range hierarchical mobility sug-gest that Chile is more fluid than the core modeldepicts Higher fluidity seems to be driven bythe weakness of horizontal barriers separatingthe agricultural manual and self-employed sec-tors of the economy This is consistent withfindings of weaker sector effects in countriesthat have experienced rapid industrializationsuch as Korea (Park 2004) Brazil (Costa-Ribeiro 2003) Israel (Yaish 2004 Goldthorpeet al 1997) and Japan (Ishida et al 1991)

Additionally in the Chilean case the weak-ness of sector barriers may have been intensi-fied by the deep market transformation of the1970s and 1980s On the one hand the ldquochangeof handsrdquo in the agricultural sector induced bythe agrarian reform and counterreform likelyincreased fluidity between the agricultural andurban classes On the other hand the role of self-employment as an ephemeral refuge against(industrial and public sector) unemploymentmay have reduced barriers between the self-employed group and the rest of the social struc-ture Added to the larger difficulty of long-rangehierarchical mobility these features suggestweak sector cleavages in the Chilean fluiditypattern and high salience of the hierarchicaldimension of mobility

For the countries in which the fit of the coremodel is not adequate Erikson and Goldthorpe(1992a145) suggest adding parameters thataccount for country-specific deviations from

the model I therefore introduce adjustments toreflect salient characteristics of the Chilean flu-idity pattern producing a ldquoChilean versionrdquo ofthe core model

First I modify the hierarchical effects Fittingof the core model suggests that hierarchicalbarriers are a crucial determinant of mobility inChile This conclusion is however obscuredby the fact that the hierarchical strata distin-guished by the core model are not empiricallyobtained and may not accurately represent theChilean ranking of classes in terms of socioe-conomic status (SES) To model hierarchicaleffects in Chile adequately I rank classes usingthe unweighted average of their schooling andearning levels as a proxy for SES I then collapsethe seven classes into three hierarchical stratausing cluster analysis These three empiricallyobtained strata are the following Upper stratum(service class I+II) middle stratum (routinenonmanual and self-employed group III andIVab) and lower stratum (manual and agricul-tural classes V-VI VIIa IVc and VIIb) Detailson the ranking of classes cluster procedures andthe comparison of the CASMIN and empiricalChilean class ranking can be found in theAppendix Comparison of the CASMIN and theempirical class ranking suggests higher differ-entiation at the top than at the bottom of thesocial structure in the Chilean case

Model 2 in Table 2 uses the empiricallyobtained hierarchical strata (EHE) Theimprovement in fit is massive (the difference inBIC is 293 without using any degrees of free-dom) The model now accounts for 86 percentof the association under independence

The large improvement in fit when empiricalhierarchical strata are used challenges theassumption of an internationally homogeneoushierarchical class ranking At the empiricallevel there is no proof that the hierarchicalranking of classes assumed by the core modelfits all CASMIN countries9 Moreover someevidence points to significant cross-country dif-

443366mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

9 Erikson and Goldthorpe(1992a Table 22) pres-ent the average ranking of classes based on standardstatus measures but they do not show the empiricalranking of classes for each country If the occupa-tional composition of classes varies across countriesthere may be significant international differences inthe hierarchical ranking of classes

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ferences in the ranking of classes (Hout andHauser 1992 Sorensen 1992) At the theoreti-cal level the creators of the core model high-light the relevance of highly specific historicallyformed national attributes to explain interna-tional variation in mobility These national par-ticularities may very well result in differenthierarchical positions of classes across coun-tries10 Furthermore even if the homogeneousranking of classes imposed by the CASMINresearchers applied to most CASMIN coun-tries significant departures may be found inthe late industrialized countries currently beingadded to the comparative mobility project TheChilean findings suggest that instead of impos-ing international homogeneity to the hierarchi-cal dimension of mobility the ranking of classesshould be treated as nationally specific param-eters to be estimated as mobility researchersroutinely treat row and column marginals ofthe mobility table The issue is importantbecause the hierarchical dimension is the mostrelevant dimension of mobility (Hout andHauser 1992 Wong 1992) and inadequate mod-eling can bias assessment of not only hierar-chical effects of mobility but also other effects

Two other changes were introduced toaccount for Chilean-specific deviations fromthe core model First the IN2 effect wasrearranged to exclude the self-employed class(IVab) and to include the class of farm workers(VIIb) This rearrangement accounts for thefact that the Chilean self-employed classappeared less likely than depicted by the coremodel to reproduce its class position acrossgenerations and the inheritance level of Chileanfarm workers is higher than in the CASMINcountries These changes are expressed in theChilean-specific inheritance parameter IN2-CSecond an affinity parameter (AF1-C) wasadded to capture the difficulty of long-rangedownward elite mobility This negative asym-metrical affinity links flows from the service

class to the agricultural classes and from theroutine nonmanual class to the class of farmworkers Note that the AF1-C adds to the hier-archical distance between the two extremes ofthe class structure and to the negative affinitycaptured by AF1 to depict the extreme difficultyof long-range downward mobility from theChilean elite The fit of this lsquoChilean versionrsquoof the core model is reported in Table 2 model3 The improvement in BIC from model 2 is sub-stantial Furthermore even if the model fallsshort in terms of standard statistical criteria itaccounts for a large 91 of the associationunder independence thereby supporting theclaim that it adequately represents the Chileanmobility pattern The parameter estimates for theChilean version of the core are reported in row5 Table 3

Because of the core modelrsquos undesirable prop-erties I also use an alternative approach toexamine Chilean mobility patterns Thisapproach uses the now standard associationrowndashcolumn (II) (RC[II]) model (Goodman1979 see also Hout 1983 and Wong 1992) Inthis model both origin and destination classesare scaled so that the association can beexpressed as a linear-by-linear interaction by asingle parameter The multiplicative form ofthe model can be expressed as follows

Fij = iO j

D exp( microi νj)

where i indexes class of origin j indexes classof destination Fij is the expected frequency inthe (ij) cell is the grand mean i

O pertainsto the class of origin marginal effect j

D pertainsto the class of destination marginal effect isa global association parameter microi is the scalevalue for the ith class of origin and νj is the scalescore for the jth class of destination subject tothe following normalization constraints Σmicroi =Σνj = 0 (normalization of the location) and Σmicroi2

= Σνj2 = 1 (normalization of the scale)The class scale scores reflect a latent con-

tinuous variable made manifest by the class cat-egories Empirically obtained from the datathese ldquodistancerdquo scores produce an optimalranking of classes for the purpose of mobilityanalysis The RC(II) model is appropriate formodeling the association of any two ordinalvariables but mobility tables are distinctivebecause of the correspondence between class oforigin and class of destination (Gerber and Hout2004691)

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443377

2322-ASR 703 filename70304-torche

10 For instance during the socialist period inHungary the status of the proletariat may have beenindistinguishable from the status of the routine non-manual class (Szelenyi 199862) In Scotland givento early industrialization the hierarchical barrierbetween the skilled and unskilled working classes thatcharacterizes other CASMIN countries may not exist(Erikson and Goldthorpe 1992a163)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I introduce two adjustments to account forthis correspondence transforming the formu-lation into a ldquohybrid modelrdquo First topologicalparameters will account for the large counts insome diagonal cells reflecting immobility(inheritance) effects These parameters distin-guish two inheritance levels ldquohigh inheritancerdquofor the service class (I+II) farmers (IVc) andfarm workers (VIIb) and ldquolow inheritancerdquo forall other classes with the exception of the rou-tine nonmanual workers

Model 4 in Table 2 reports the fit statistic forthis quasi-RC(II) model The BIC statisticrejects it and prefers the more parsimoniousChilean version of the core model ThereforeI introduce a second adjustment to account forthe correspondence between rows and columnsin mobility tables Origin and destination scoresare constrained to be the same assuming thatthe scaling of classes has not significantlychanged over time Model 5 in Table 2 reportsfit statistics for this homogeneous quasi-RC(II)model On the basis of the BIC statistic the fitof the model is as good as that of the Chileanversion of the core model (Wong [1994] showsthat BIC differences of less than 5 points shouldbe considered indeterminate) Thus under stan-dard statistical criteria the Chilean core modeland the homogeneous quasi-RC(II) model areindistinguishable Given that the Chilean versionof the core model was tailored to fit the Chileantable the fit of model 5 is impressive and it indi-cates that a unidimensional scale can adequatelycapture mobility distances between classes

The logical next question is what this scalerepresents and specifically whether it repre-sents the hierarchical ranking of classes in termsof SES If a close correspondence is foundbetween the mobility distances among classesand their distances in terms of SES this find-ing would suggest that the intragenerationaldistribution of resources and rewards expressedin the SES ranking of classes drives the processof intergenerational mobility In other words thisfinding would support the resource approachrsquosclaim that differences in resources associatedwith diverse class positions determine mobili-ty opportunities across generations

To examine this possibility I compare theclass scores obtained from the homogeneousquasi-RC(II) model with the empiricallyobtained SES ranking of classes based on edu-cation and earnings (Appendix) Figure 4 pres-

ents the comparison between the standardizedvalues of both sets of scores The correspon-dence is striking The mobility distancesbetween classes closely reproduce their dis-tances in terms of SES This close isomorphismsuggests that at least in Chile the intragener-ational inequality in the distribution of socialresources and rewards across classes largelydrives the intergenerational mobility process Totest the isomorphism further I use the SESclass scores as row and column rankings to esti-mate a quasindashlinear-by-linear association model(model 6 Table 2) The fit of model 6 is sig-nificantly worse than that of model 5 almost aspoor as that of the original core model This indi-cates that although the inter- and intragenera-tional dimensions of inequality are very similarin Chile they have a somewhat different struc-ture

In summary the central characteristic of theChilean mobility regime is the predominance oflong-range hierarchical barriers and the weak-ness of sector cleavages separating classes in themiddle and at the lower end of the hierarchicalstructure Findings from the ldquoChilean versionrdquoof the core model and the quasi-RC(II) hybridmodels are consistent depicting a social struc-ture characterized by signif icant barriersbetween the top echelon and the rest of the classstructure The good fit of the homogeneousquasi-RC(II) model provides preliminary sup-port for the resource approach linking inequal-ity and mobility Mobility dynamics seem to bedriven largely by the hierarchical distancesbetween classes and overall Chile seems not tobe less fluid than the CASMIN countries

CCOOMMPPAARRIISSOONN OOFF TTHHEE OORRIIGGIINNAALL AANNDD CCHHIILLEEAANN

VVEERRSSIIOONNSS OOFF TTHHEE CCOORREE MMOODDEELL

Although Chilean mobility dynamics seem to bedriven by inequality across classes in a veryunequal society the fitting of the core modelprovides no indication that Chile is significantlyless fluid than depicted by the core model Asa preliminary analysis of the Chilean level of flu-idity I compare the predicted propensities formobility and immobility between specific pairsof classes in Chile with those in the CASMINcountries Propensities are obtained from theoriginal and Chilean versions of the core modelrespectively (models 2 and 5 of Table 3)

443388mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I begin with the immobility of those at the topof the class structure namely the service class(I+II) The service classrsquo propensity for immo-bility in the original core model is captured bytwo parameters IN1 and IN2 This propensityis more than three times what it would be in theabsence of these effects (e43+84 = 356) InChile the propensity for immobility of class I+IIis captured by the same two parameters and itis lower than what the core model expresses286 times larger than it would be in the absenceof these effects (e34+71) What about the immo-bility of the two agricultural classes In theCASMIN countries immobility of the IVc andVIIb classes is respectively 785 and 154 timeswhat it would be in the absence of these param-eters (e43+84+101ndash22 and e43 respectively) InChile the comparable values are 422 and 286(e34+71+48ndash09 and e43+71 respectively) indi-cating the relatively low inheritance associatedwith land ownership in the Latin Americannation Thus if attention is focused on theimmobility in the upper and lower ends of thesocial hierarchy the comparison does not indi-cate less fluidity in the highly unequal Chileansociety

What about the mobility rates between thetwo most distant classes in the hierarchical rank-ing namely the service class (I+II) and farmworkers (VIIb) In the case of the CASMINcountries the mobility chances between theseclasses are captured by four parameters (HI1HI2 SE1 and AF1) and is 077 times smallerthan it would be in the absence of these effects(endash22 ndash45ndash106ndash83) In Chile the distance fromclass I to class VIIb is expressed by these samefour parameters in addition to the asymmetri-cal disaffinity AF1-C This yields mobilitychances 056 times what would be obtained inthe absence of these effects (endash09 ndash47ndash20ndash63ndash149)Thus the barrier to long-range downwardmobility from the elite is more significant inChile than in the CASMIN countries The sameoccurs with long-range downward mobility fromclass III

If we measure the reciprocal flow (ieupward mobility from class VIIb to class I+II)we obtain the same value for the CASMINcountries because all parameters are symmet-rical in the original core model For Chile how-ever this calculation removes the asymmetricalbarrier to elite downward mobility which rais-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443399

2322-ASR 703 filename70304-torche

Figure 4 Class Scores Obtained from the Homogeneous Quasi-RC(II) Model and from Socioeconomic Status(SES) Ranking

Note (1) = Class standardized socioeconomic status scores based on schooling and earnings (2) = Class stan-dardized scores based on the Homogeneous Quasi-RC(II) model (see text for details)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 13: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

immobility They include three matrices IN1identifies immobility across all classes alike IN2identifies the higher immobility of the serviceclass (I+II) the self-employed group (IVab) andfarmers (IVc) and IN3 accounts for the highestimmobility of farmers (IVc)

SECTOR EFFECTS Sector effects are intended toidentify the difficulty of moving between theagricultural and nonagricultural sectors of theeconomy (SE1)

AFFINITY EFFECTS Affinity effects are intend-ed to capture specific discontinuities (negativeaffinities) or linkages (positive affinities) betweenclasses which either reinforce or offset the over-all effects of hierarchy and sector There are twoaffinity effects The first AF1 identifies the dif-ficulty of moving between the service class (I+II)and the class of farm workers (VIIb) which addsto their hierarchical distance In contrast AF2identifies instances in which mobility is morefrequent than accounted for by hierarchy and sec-tor effects and includes affinities within the non-manual sector (I+II and III) and within the manualsector (V+VI and VIIa) a symmetrical affinitybased on capital possession (IVab and IVc and Iand IVab) and an asymmetrical link between agri-cultural classes of origin (IVc and VIIb) and theunskilled manual class of destination (VIIa)7

TTHHEE MMOOBBIILLIITTYY PPAATTTTEERRNN IINN CCHHIILLEECCOONNVVEERRGGEENNCCEE OORR DDIIVVEERRGGEENNCCEE

Table 2 presents the fit of several models appliedto the Chilean data and Table 3 displays theparameter estimates for each of these modelsModel 1 in Table 2 indicates that the fit of theoriginal core model to the Chilean table is unsat-isfactory under standard statistical criteria butit explains a significant 775 percent of associ-ation under independence To assess the strengthof the different factors driving mobility oppor-tunities I compare the Chilean coefficients withthose obtained for the CASMIN countriesCoefficients from the original core model fit-ted to the Chilean table are reported in row 1 ofTable 3 CASMIN coefficients are reported inrow 2 and differences in coefficientsrsquo magni-tude and the statistical significance of the dif-ferences are reported in row 38

All Chilean coefficients are significant andhave the same sign as for the CASMIN coun-tries Comparison of the magnitude of the coef-ficients however shows significant departuresfrom the core for five effects which revealinteresting differences between Chile and theCASMIN countries The hierarchy coefficientsindicate that whereas the short-range hierar-chical barriers are somewhat weaker in Chile

443344mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

8 CASMIN coefficients were obtained from fittingthe core model to the combined table of France andEngland standardizing both tables to have 10000cases as recommended by Erikson and Goldthorpe(1992a121) The coefficients obtained are virtuallyidentical to those presented by Erikson andGoldthorpe (1992a Table 44)

Table 2 Fit statistics selected models to the Chilean Mobility Table

AssociationL2 BIC df p Explained

Core Modelmdash1 Core Model Original 1375 285 28 000 775mdash2 Core Model EHE 1081 ndash08 28 000 860mdash3 Chilean version of Core Model ([Model 2] + Chilean Parameters) 724 ndash326 27 000 906Association Models 000mdash4 Heterogeneous Quasi-RC(II) Model 734 ndash161 23 000 905mdash5 Homogeneous Quasi-RC(II) Model 803 ndash287 28 000 896mdash6 Quasi-Linear by Linear Association (SES row and column ranking)a 1563 279 33 000 798

Note EHE = empirical hierarchical effects SES = eocioeconomic statusa Linear-by-linear association model using socioeconomic status scores as presented in Appendix Figure A1 torank origin and destination classes

7 For a detailed account of the empirical derivationof the model see Erikson and Goldthorpe (1987a1987b and 1992a chapter 4)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443355

2322-ASR 703 filename70304-torche

Tab

le 3

M

odel

Par

amet

er E

stim

ates

for

the

Chi

lean

Mob

ilit

y Ta

ble

Eff

ect

1 C

ore

mod

el o

rigi

nala

mdash 2 C

ore

mod

el C

AS

MIN

b

mdash 3 D

iffe

renc

e C

ASM

INmdash

Chi

lec

mdash 4 C

ore

mod

el E

HE

d

mdash 5 C

hile

an v

ersi

on o

f co

ree

mdash Not

eD

ata

show

n ar

e m

odel

par

amet

er e

stim

ates

wit

h st

anda

rd e

rror

s in

par

enth

eses

EH

E =

em

piri

cal h

iera

rchi

cal e

ffec

ts

aM

odel

1 f

rom

Tab

le 2

b

Ori

gina

l Cor

e M

odel

fit

ted

to th

e E

ngli

sh a

nd F

renc

h m

obil

ity

tabl

es S

ee te

xt f

or d

etai

ls

cD

iffe

renc

e in

par

amet

er m

agni

tude

bet

wee

n or

igin

al c

ore

mod

el f

itte

d to

the

Chi

lean

tabl

e (r

ow 1

) an

d to

the

CA

SM

IN ta

ble

(row

2)

dM

odel

2 f

rom

Tab

le 2

e

Mod

el 3

fro

m T

able

2

dagger p

lt 1

p

lt 0

5

p

lt 0

1

p

lt 0

01 (

two-

tail

ed)

HI1

ndash12

(

05)

ndash22

(02

)ndash

10 dagger

(06

)ndash

03(

07)

ndash09

dagger(

06)

HI2

ndash65

(11

)ndash

45

(

04)

20

dagger(

12)

ndash74

(09

)ndash

47

(

10)

IN1

41

(

08)

43

(

03)

Insi

g

37

(

07)

34

(

07)

IN2

39

(

10)

84

(

04)

45

(

11)

31

(13

)mdash

IN3

68

(

24)

101

(11

)In

sig

61

(24

)4

8(

23)

SE

1

ndash38

(07

)ndash1

06

(

05)

ndash68

(08

)ndash

42

(

07)

ndash20

(

08)

AF

1

ndash76

(21

)ndash

83

(

13)

Insi

g

ndash84

(20

)ndash

63

(21

)

AF

2

33

(

05)

45

(

02)

ndash12

(

06)

20

(

06)

26

(

06)

AF

1ndashC

mdash mdash mdash mdash

ndash14

9

(46

)

IN2ndash

C

mdash mdash mdash mdash 71

(

14)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(HI1) long-range hierarchical mobility is moredifficult in this country (HI2) This suggestssubstantial inequality between the extremes ofthe class hierarchy combined with smaller dif-ferentiation in the middle of the distribution Asto the inheritance effects the service class theself-employed group and the farmers (classesI+II IVab and IVc respectively) have a sig-nificantly lower propensity to immobility inChile than in the CASMIN countries suggest-ing that class positions usually associated withindependent work are less able to reproducetheir class status intergenerationally than depict-ed by the core Most impressive the barrierbetween the agricultural and nonagriculturalsectors of the economy (SE1) is much weakerin Chile as is the affinity effect linking non-manual classes manual classes and the class-es that own capital among themselves (AF2)

Interestingly all the significant differences incoefficients with the single exception of the bar-rier to long-range hierarchical mobility sug-gest that Chile is more fluid than the core modeldepicts Higher fluidity seems to be driven bythe weakness of horizontal barriers separatingthe agricultural manual and self-employed sec-tors of the economy This is consistent withfindings of weaker sector effects in countriesthat have experienced rapid industrializationsuch as Korea (Park 2004) Brazil (Costa-Ribeiro 2003) Israel (Yaish 2004 Goldthorpeet al 1997) and Japan (Ishida et al 1991)

Additionally in the Chilean case the weak-ness of sector barriers may have been intensi-fied by the deep market transformation of the1970s and 1980s On the one hand the ldquochangeof handsrdquo in the agricultural sector induced bythe agrarian reform and counterreform likelyincreased fluidity between the agricultural andurban classes On the other hand the role of self-employment as an ephemeral refuge against(industrial and public sector) unemploymentmay have reduced barriers between the self-employed group and the rest of the social struc-ture Added to the larger difficulty of long-rangehierarchical mobility these features suggestweak sector cleavages in the Chilean fluiditypattern and high salience of the hierarchicaldimension of mobility

For the countries in which the fit of the coremodel is not adequate Erikson and Goldthorpe(1992a145) suggest adding parameters thataccount for country-specific deviations from

the model I therefore introduce adjustments toreflect salient characteristics of the Chilean flu-idity pattern producing a ldquoChilean versionrdquo ofthe core model

First I modify the hierarchical effects Fittingof the core model suggests that hierarchicalbarriers are a crucial determinant of mobility inChile This conclusion is however obscuredby the fact that the hierarchical strata distin-guished by the core model are not empiricallyobtained and may not accurately represent theChilean ranking of classes in terms of socioe-conomic status (SES) To model hierarchicaleffects in Chile adequately I rank classes usingthe unweighted average of their schooling andearning levels as a proxy for SES I then collapsethe seven classes into three hierarchical stratausing cluster analysis These three empiricallyobtained strata are the following Upper stratum(service class I+II) middle stratum (routinenonmanual and self-employed group III andIVab) and lower stratum (manual and agricul-tural classes V-VI VIIa IVc and VIIb) Detailson the ranking of classes cluster procedures andthe comparison of the CASMIN and empiricalChilean class ranking can be found in theAppendix Comparison of the CASMIN and theempirical class ranking suggests higher differ-entiation at the top than at the bottom of thesocial structure in the Chilean case

Model 2 in Table 2 uses the empiricallyobtained hierarchical strata (EHE) Theimprovement in fit is massive (the difference inBIC is 293 without using any degrees of free-dom) The model now accounts for 86 percentof the association under independence

The large improvement in fit when empiricalhierarchical strata are used challenges theassumption of an internationally homogeneoushierarchical class ranking At the empiricallevel there is no proof that the hierarchicalranking of classes assumed by the core modelfits all CASMIN countries9 Moreover someevidence points to significant cross-country dif-

443366mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

9 Erikson and Goldthorpe(1992a Table 22) pres-ent the average ranking of classes based on standardstatus measures but they do not show the empiricalranking of classes for each country If the occupa-tional composition of classes varies across countriesthere may be significant international differences inthe hierarchical ranking of classes

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ferences in the ranking of classes (Hout andHauser 1992 Sorensen 1992) At the theoreti-cal level the creators of the core model high-light the relevance of highly specific historicallyformed national attributes to explain interna-tional variation in mobility These national par-ticularities may very well result in differenthierarchical positions of classes across coun-tries10 Furthermore even if the homogeneousranking of classes imposed by the CASMINresearchers applied to most CASMIN coun-tries significant departures may be found inthe late industrialized countries currently beingadded to the comparative mobility project TheChilean findings suggest that instead of impos-ing international homogeneity to the hierarchi-cal dimension of mobility the ranking of classesshould be treated as nationally specific param-eters to be estimated as mobility researchersroutinely treat row and column marginals ofthe mobility table The issue is importantbecause the hierarchical dimension is the mostrelevant dimension of mobility (Hout andHauser 1992 Wong 1992) and inadequate mod-eling can bias assessment of not only hierar-chical effects of mobility but also other effects

Two other changes were introduced toaccount for Chilean-specific deviations fromthe core model First the IN2 effect wasrearranged to exclude the self-employed class(IVab) and to include the class of farm workers(VIIb) This rearrangement accounts for thefact that the Chilean self-employed classappeared less likely than depicted by the coremodel to reproduce its class position acrossgenerations and the inheritance level of Chileanfarm workers is higher than in the CASMINcountries These changes are expressed in theChilean-specific inheritance parameter IN2-CSecond an affinity parameter (AF1-C) wasadded to capture the difficulty of long-rangedownward elite mobility This negative asym-metrical affinity links flows from the service

class to the agricultural classes and from theroutine nonmanual class to the class of farmworkers Note that the AF1-C adds to the hier-archical distance between the two extremes ofthe class structure and to the negative affinitycaptured by AF1 to depict the extreme difficultyof long-range downward mobility from theChilean elite The fit of this lsquoChilean versionrsquoof the core model is reported in Table 2 model3 The improvement in BIC from model 2 is sub-stantial Furthermore even if the model fallsshort in terms of standard statistical criteria itaccounts for a large 91 of the associationunder independence thereby supporting theclaim that it adequately represents the Chileanmobility pattern The parameter estimates for theChilean version of the core are reported in row5 Table 3

Because of the core modelrsquos undesirable prop-erties I also use an alternative approach toexamine Chilean mobility patterns Thisapproach uses the now standard associationrowndashcolumn (II) (RC[II]) model (Goodman1979 see also Hout 1983 and Wong 1992) Inthis model both origin and destination classesare scaled so that the association can beexpressed as a linear-by-linear interaction by asingle parameter The multiplicative form ofthe model can be expressed as follows

Fij = iO j

D exp( microi νj)

where i indexes class of origin j indexes classof destination Fij is the expected frequency inthe (ij) cell is the grand mean i

O pertainsto the class of origin marginal effect j

D pertainsto the class of destination marginal effect isa global association parameter microi is the scalevalue for the ith class of origin and νj is the scalescore for the jth class of destination subject tothe following normalization constraints Σmicroi =Σνj = 0 (normalization of the location) and Σmicroi2

= Σνj2 = 1 (normalization of the scale)The class scale scores reflect a latent con-

tinuous variable made manifest by the class cat-egories Empirically obtained from the datathese ldquodistancerdquo scores produce an optimalranking of classes for the purpose of mobilityanalysis The RC(II) model is appropriate formodeling the association of any two ordinalvariables but mobility tables are distinctivebecause of the correspondence between class oforigin and class of destination (Gerber and Hout2004691)

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443377

2322-ASR 703 filename70304-torche

10 For instance during the socialist period inHungary the status of the proletariat may have beenindistinguishable from the status of the routine non-manual class (Szelenyi 199862) In Scotland givento early industrialization the hierarchical barrierbetween the skilled and unskilled working classes thatcharacterizes other CASMIN countries may not exist(Erikson and Goldthorpe 1992a163)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I introduce two adjustments to account forthis correspondence transforming the formu-lation into a ldquohybrid modelrdquo First topologicalparameters will account for the large counts insome diagonal cells reflecting immobility(inheritance) effects These parameters distin-guish two inheritance levels ldquohigh inheritancerdquofor the service class (I+II) farmers (IVc) andfarm workers (VIIb) and ldquolow inheritancerdquo forall other classes with the exception of the rou-tine nonmanual workers

Model 4 in Table 2 reports the fit statistic forthis quasi-RC(II) model The BIC statisticrejects it and prefers the more parsimoniousChilean version of the core model ThereforeI introduce a second adjustment to account forthe correspondence between rows and columnsin mobility tables Origin and destination scoresare constrained to be the same assuming thatthe scaling of classes has not significantlychanged over time Model 5 in Table 2 reportsfit statistics for this homogeneous quasi-RC(II)model On the basis of the BIC statistic the fitof the model is as good as that of the Chileanversion of the core model (Wong [1994] showsthat BIC differences of less than 5 points shouldbe considered indeterminate) Thus under stan-dard statistical criteria the Chilean core modeland the homogeneous quasi-RC(II) model areindistinguishable Given that the Chilean versionof the core model was tailored to fit the Chileantable the fit of model 5 is impressive and it indi-cates that a unidimensional scale can adequatelycapture mobility distances between classes

The logical next question is what this scalerepresents and specifically whether it repre-sents the hierarchical ranking of classes in termsof SES If a close correspondence is foundbetween the mobility distances among classesand their distances in terms of SES this find-ing would suggest that the intragenerationaldistribution of resources and rewards expressedin the SES ranking of classes drives the processof intergenerational mobility In other words thisfinding would support the resource approachrsquosclaim that differences in resources associatedwith diverse class positions determine mobili-ty opportunities across generations

To examine this possibility I compare theclass scores obtained from the homogeneousquasi-RC(II) model with the empiricallyobtained SES ranking of classes based on edu-cation and earnings (Appendix) Figure 4 pres-

ents the comparison between the standardizedvalues of both sets of scores The correspon-dence is striking The mobility distancesbetween classes closely reproduce their dis-tances in terms of SES This close isomorphismsuggests that at least in Chile the intragener-ational inequality in the distribution of socialresources and rewards across classes largelydrives the intergenerational mobility process Totest the isomorphism further I use the SESclass scores as row and column rankings to esti-mate a quasindashlinear-by-linear association model(model 6 Table 2) The fit of model 6 is sig-nificantly worse than that of model 5 almost aspoor as that of the original core model This indi-cates that although the inter- and intragenera-tional dimensions of inequality are very similarin Chile they have a somewhat different struc-ture

In summary the central characteristic of theChilean mobility regime is the predominance oflong-range hierarchical barriers and the weak-ness of sector cleavages separating classes in themiddle and at the lower end of the hierarchicalstructure Findings from the ldquoChilean versionrdquoof the core model and the quasi-RC(II) hybridmodels are consistent depicting a social struc-ture characterized by signif icant barriersbetween the top echelon and the rest of the classstructure The good fit of the homogeneousquasi-RC(II) model provides preliminary sup-port for the resource approach linking inequal-ity and mobility Mobility dynamics seem to bedriven largely by the hierarchical distancesbetween classes and overall Chile seems not tobe less fluid than the CASMIN countries

CCOOMMPPAARRIISSOONN OOFF TTHHEE OORRIIGGIINNAALL AANNDD CCHHIILLEEAANN

VVEERRSSIIOONNSS OOFF TTHHEE CCOORREE MMOODDEELL

Although Chilean mobility dynamics seem to bedriven by inequality across classes in a veryunequal society the fitting of the core modelprovides no indication that Chile is significantlyless fluid than depicted by the core model Asa preliminary analysis of the Chilean level of flu-idity I compare the predicted propensities formobility and immobility between specific pairsof classes in Chile with those in the CASMINcountries Propensities are obtained from theoriginal and Chilean versions of the core modelrespectively (models 2 and 5 of Table 3)

443388mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I begin with the immobility of those at the topof the class structure namely the service class(I+II) The service classrsquo propensity for immo-bility in the original core model is captured bytwo parameters IN1 and IN2 This propensityis more than three times what it would be in theabsence of these effects (e43+84 = 356) InChile the propensity for immobility of class I+IIis captured by the same two parameters and itis lower than what the core model expresses286 times larger than it would be in the absenceof these effects (e34+71) What about the immo-bility of the two agricultural classes In theCASMIN countries immobility of the IVc andVIIb classes is respectively 785 and 154 timeswhat it would be in the absence of these param-eters (e43+84+101ndash22 and e43 respectively) InChile the comparable values are 422 and 286(e34+71+48ndash09 and e43+71 respectively) indi-cating the relatively low inheritance associatedwith land ownership in the Latin Americannation Thus if attention is focused on theimmobility in the upper and lower ends of thesocial hierarchy the comparison does not indi-cate less fluidity in the highly unequal Chileansociety

What about the mobility rates between thetwo most distant classes in the hierarchical rank-ing namely the service class (I+II) and farmworkers (VIIb) In the case of the CASMINcountries the mobility chances between theseclasses are captured by four parameters (HI1HI2 SE1 and AF1) and is 077 times smallerthan it would be in the absence of these effects(endash22 ndash45ndash106ndash83) In Chile the distance fromclass I to class VIIb is expressed by these samefour parameters in addition to the asymmetri-cal disaffinity AF1-C This yields mobilitychances 056 times what would be obtained inthe absence of these effects (endash09 ndash47ndash20ndash63ndash149)Thus the barrier to long-range downwardmobility from the elite is more significant inChile than in the CASMIN countries The sameoccurs with long-range downward mobility fromclass III

If we measure the reciprocal flow (ieupward mobility from class VIIb to class I+II)we obtain the same value for the CASMINcountries because all parameters are symmet-rical in the original core model For Chile how-ever this calculation removes the asymmetricalbarrier to elite downward mobility which rais-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443399

2322-ASR 703 filename70304-torche

Figure 4 Class Scores Obtained from the Homogeneous Quasi-RC(II) Model and from Socioeconomic Status(SES) Ranking

Note (1) = Class standardized socioeconomic status scores based on schooling and earnings (2) = Class stan-dardized scores based on the Homogeneous Quasi-RC(II) model (see text for details)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 14: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443355

2322-ASR 703 filename70304-torche

Tab

le 3

M

odel

Par

amet

er E

stim

ates

for

the

Chi

lean

Mob

ilit

y Ta

ble

Eff

ect

1 C

ore

mod

el o

rigi

nala

mdash 2 C

ore

mod

el C

AS

MIN

b

mdash 3 D

iffe

renc

e C

ASM

INmdash

Chi

lec

mdash 4 C

ore

mod

el E

HE

d

mdash 5 C

hile

an v

ersi

on o

f co

ree

mdash Not

eD

ata

show

n ar

e m

odel

par

amet

er e

stim

ates

wit

h st

anda

rd e

rror

s in

par

enth

eses

EH

E =

em

piri

cal h

iera

rchi

cal e

ffec

ts

aM

odel

1 f

rom

Tab

le 2

b

Ori

gina

l Cor

e M

odel

fit

ted

to th

e E

ngli

sh a

nd F

renc

h m

obil

ity

tabl

es S

ee te

xt f

or d

etai

ls

cD

iffe

renc

e in

par

amet

er m

agni

tude

bet

wee

n or

igin

al c

ore

mod

el f

itte

d to

the

Chi

lean

tabl

e (r

ow 1

) an

d to

the

CA

SM

IN ta

ble

(row

2)

dM

odel

2 f

rom

Tab

le 2

e

Mod

el 3

fro

m T

able

2

dagger p

lt 1

p

lt 0

5

p

lt 0

1

p

lt 0

01 (

two-

tail

ed)

HI1

ndash12

(

05)

ndash22

(02

)ndash

10 dagger

(06

)ndash

03(

07)

ndash09

dagger(

06)

HI2

ndash65

(11

)ndash

45

(

04)

20

dagger(

12)

ndash74

(09

)ndash

47

(

10)

IN1

41

(

08)

43

(

03)

Insi

g

37

(

07)

34

(

07)

IN2

39

(

10)

84

(

04)

45

(

11)

31

(13

)mdash

IN3

68

(

24)

101

(11

)In

sig

61

(24

)4

8(

23)

SE

1

ndash38

(07

)ndash1

06

(

05)

ndash68

(08

)ndash

42

(

07)

ndash20

(

08)

AF

1

ndash76

(21

)ndash

83

(

13)

Insi

g

ndash84

(20

)ndash

63

(21

)

AF

2

33

(

05)

45

(

02)

ndash12

(

06)

20

(

06)

26

(

06)

AF

1ndashC

mdash mdash mdash mdash

ndash14

9

(46

)

IN2ndash

C

mdash mdash mdash mdash 71

(

14)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(HI1) long-range hierarchical mobility is moredifficult in this country (HI2) This suggestssubstantial inequality between the extremes ofthe class hierarchy combined with smaller dif-ferentiation in the middle of the distribution Asto the inheritance effects the service class theself-employed group and the farmers (classesI+II IVab and IVc respectively) have a sig-nificantly lower propensity to immobility inChile than in the CASMIN countries suggest-ing that class positions usually associated withindependent work are less able to reproducetheir class status intergenerationally than depict-ed by the core Most impressive the barrierbetween the agricultural and nonagriculturalsectors of the economy (SE1) is much weakerin Chile as is the affinity effect linking non-manual classes manual classes and the class-es that own capital among themselves (AF2)

Interestingly all the significant differences incoefficients with the single exception of the bar-rier to long-range hierarchical mobility sug-gest that Chile is more fluid than the core modeldepicts Higher fluidity seems to be driven bythe weakness of horizontal barriers separatingthe agricultural manual and self-employed sec-tors of the economy This is consistent withfindings of weaker sector effects in countriesthat have experienced rapid industrializationsuch as Korea (Park 2004) Brazil (Costa-Ribeiro 2003) Israel (Yaish 2004 Goldthorpeet al 1997) and Japan (Ishida et al 1991)

Additionally in the Chilean case the weak-ness of sector barriers may have been intensi-fied by the deep market transformation of the1970s and 1980s On the one hand the ldquochangeof handsrdquo in the agricultural sector induced bythe agrarian reform and counterreform likelyincreased fluidity between the agricultural andurban classes On the other hand the role of self-employment as an ephemeral refuge against(industrial and public sector) unemploymentmay have reduced barriers between the self-employed group and the rest of the social struc-ture Added to the larger difficulty of long-rangehierarchical mobility these features suggestweak sector cleavages in the Chilean fluiditypattern and high salience of the hierarchicaldimension of mobility

For the countries in which the fit of the coremodel is not adequate Erikson and Goldthorpe(1992a145) suggest adding parameters thataccount for country-specific deviations from

the model I therefore introduce adjustments toreflect salient characteristics of the Chilean flu-idity pattern producing a ldquoChilean versionrdquo ofthe core model

First I modify the hierarchical effects Fittingof the core model suggests that hierarchicalbarriers are a crucial determinant of mobility inChile This conclusion is however obscuredby the fact that the hierarchical strata distin-guished by the core model are not empiricallyobtained and may not accurately represent theChilean ranking of classes in terms of socioe-conomic status (SES) To model hierarchicaleffects in Chile adequately I rank classes usingthe unweighted average of their schooling andearning levels as a proxy for SES I then collapsethe seven classes into three hierarchical stratausing cluster analysis These three empiricallyobtained strata are the following Upper stratum(service class I+II) middle stratum (routinenonmanual and self-employed group III andIVab) and lower stratum (manual and agricul-tural classes V-VI VIIa IVc and VIIb) Detailson the ranking of classes cluster procedures andthe comparison of the CASMIN and empiricalChilean class ranking can be found in theAppendix Comparison of the CASMIN and theempirical class ranking suggests higher differ-entiation at the top than at the bottom of thesocial structure in the Chilean case

Model 2 in Table 2 uses the empiricallyobtained hierarchical strata (EHE) Theimprovement in fit is massive (the difference inBIC is 293 without using any degrees of free-dom) The model now accounts for 86 percentof the association under independence

The large improvement in fit when empiricalhierarchical strata are used challenges theassumption of an internationally homogeneoushierarchical class ranking At the empiricallevel there is no proof that the hierarchicalranking of classes assumed by the core modelfits all CASMIN countries9 Moreover someevidence points to significant cross-country dif-

443366mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

9 Erikson and Goldthorpe(1992a Table 22) pres-ent the average ranking of classes based on standardstatus measures but they do not show the empiricalranking of classes for each country If the occupa-tional composition of classes varies across countriesthere may be significant international differences inthe hierarchical ranking of classes

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ferences in the ranking of classes (Hout andHauser 1992 Sorensen 1992) At the theoreti-cal level the creators of the core model high-light the relevance of highly specific historicallyformed national attributes to explain interna-tional variation in mobility These national par-ticularities may very well result in differenthierarchical positions of classes across coun-tries10 Furthermore even if the homogeneousranking of classes imposed by the CASMINresearchers applied to most CASMIN coun-tries significant departures may be found inthe late industrialized countries currently beingadded to the comparative mobility project TheChilean findings suggest that instead of impos-ing international homogeneity to the hierarchi-cal dimension of mobility the ranking of classesshould be treated as nationally specific param-eters to be estimated as mobility researchersroutinely treat row and column marginals ofthe mobility table The issue is importantbecause the hierarchical dimension is the mostrelevant dimension of mobility (Hout andHauser 1992 Wong 1992) and inadequate mod-eling can bias assessment of not only hierar-chical effects of mobility but also other effects

Two other changes were introduced toaccount for Chilean-specific deviations fromthe core model First the IN2 effect wasrearranged to exclude the self-employed class(IVab) and to include the class of farm workers(VIIb) This rearrangement accounts for thefact that the Chilean self-employed classappeared less likely than depicted by the coremodel to reproduce its class position acrossgenerations and the inheritance level of Chileanfarm workers is higher than in the CASMINcountries These changes are expressed in theChilean-specific inheritance parameter IN2-CSecond an affinity parameter (AF1-C) wasadded to capture the difficulty of long-rangedownward elite mobility This negative asym-metrical affinity links flows from the service

class to the agricultural classes and from theroutine nonmanual class to the class of farmworkers Note that the AF1-C adds to the hier-archical distance between the two extremes ofthe class structure and to the negative affinitycaptured by AF1 to depict the extreme difficultyof long-range downward mobility from theChilean elite The fit of this lsquoChilean versionrsquoof the core model is reported in Table 2 model3 The improvement in BIC from model 2 is sub-stantial Furthermore even if the model fallsshort in terms of standard statistical criteria itaccounts for a large 91 of the associationunder independence thereby supporting theclaim that it adequately represents the Chileanmobility pattern The parameter estimates for theChilean version of the core are reported in row5 Table 3

Because of the core modelrsquos undesirable prop-erties I also use an alternative approach toexamine Chilean mobility patterns Thisapproach uses the now standard associationrowndashcolumn (II) (RC[II]) model (Goodman1979 see also Hout 1983 and Wong 1992) Inthis model both origin and destination classesare scaled so that the association can beexpressed as a linear-by-linear interaction by asingle parameter The multiplicative form ofthe model can be expressed as follows

Fij = iO j

D exp( microi νj)

where i indexes class of origin j indexes classof destination Fij is the expected frequency inthe (ij) cell is the grand mean i

O pertainsto the class of origin marginal effect j

D pertainsto the class of destination marginal effect isa global association parameter microi is the scalevalue for the ith class of origin and νj is the scalescore for the jth class of destination subject tothe following normalization constraints Σmicroi =Σνj = 0 (normalization of the location) and Σmicroi2

= Σνj2 = 1 (normalization of the scale)The class scale scores reflect a latent con-

tinuous variable made manifest by the class cat-egories Empirically obtained from the datathese ldquodistancerdquo scores produce an optimalranking of classes for the purpose of mobilityanalysis The RC(II) model is appropriate formodeling the association of any two ordinalvariables but mobility tables are distinctivebecause of the correspondence between class oforigin and class of destination (Gerber and Hout2004691)

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443377

2322-ASR 703 filename70304-torche

10 For instance during the socialist period inHungary the status of the proletariat may have beenindistinguishable from the status of the routine non-manual class (Szelenyi 199862) In Scotland givento early industrialization the hierarchical barrierbetween the skilled and unskilled working classes thatcharacterizes other CASMIN countries may not exist(Erikson and Goldthorpe 1992a163)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I introduce two adjustments to account forthis correspondence transforming the formu-lation into a ldquohybrid modelrdquo First topologicalparameters will account for the large counts insome diagonal cells reflecting immobility(inheritance) effects These parameters distin-guish two inheritance levels ldquohigh inheritancerdquofor the service class (I+II) farmers (IVc) andfarm workers (VIIb) and ldquolow inheritancerdquo forall other classes with the exception of the rou-tine nonmanual workers

Model 4 in Table 2 reports the fit statistic forthis quasi-RC(II) model The BIC statisticrejects it and prefers the more parsimoniousChilean version of the core model ThereforeI introduce a second adjustment to account forthe correspondence between rows and columnsin mobility tables Origin and destination scoresare constrained to be the same assuming thatthe scaling of classes has not significantlychanged over time Model 5 in Table 2 reportsfit statistics for this homogeneous quasi-RC(II)model On the basis of the BIC statistic the fitof the model is as good as that of the Chileanversion of the core model (Wong [1994] showsthat BIC differences of less than 5 points shouldbe considered indeterminate) Thus under stan-dard statistical criteria the Chilean core modeland the homogeneous quasi-RC(II) model areindistinguishable Given that the Chilean versionof the core model was tailored to fit the Chileantable the fit of model 5 is impressive and it indi-cates that a unidimensional scale can adequatelycapture mobility distances between classes

The logical next question is what this scalerepresents and specifically whether it repre-sents the hierarchical ranking of classes in termsof SES If a close correspondence is foundbetween the mobility distances among classesand their distances in terms of SES this find-ing would suggest that the intragenerationaldistribution of resources and rewards expressedin the SES ranking of classes drives the processof intergenerational mobility In other words thisfinding would support the resource approachrsquosclaim that differences in resources associatedwith diverse class positions determine mobili-ty opportunities across generations

To examine this possibility I compare theclass scores obtained from the homogeneousquasi-RC(II) model with the empiricallyobtained SES ranking of classes based on edu-cation and earnings (Appendix) Figure 4 pres-

ents the comparison between the standardizedvalues of both sets of scores The correspon-dence is striking The mobility distancesbetween classes closely reproduce their dis-tances in terms of SES This close isomorphismsuggests that at least in Chile the intragener-ational inequality in the distribution of socialresources and rewards across classes largelydrives the intergenerational mobility process Totest the isomorphism further I use the SESclass scores as row and column rankings to esti-mate a quasindashlinear-by-linear association model(model 6 Table 2) The fit of model 6 is sig-nificantly worse than that of model 5 almost aspoor as that of the original core model This indi-cates that although the inter- and intragenera-tional dimensions of inequality are very similarin Chile they have a somewhat different struc-ture

In summary the central characteristic of theChilean mobility regime is the predominance oflong-range hierarchical barriers and the weak-ness of sector cleavages separating classes in themiddle and at the lower end of the hierarchicalstructure Findings from the ldquoChilean versionrdquoof the core model and the quasi-RC(II) hybridmodels are consistent depicting a social struc-ture characterized by signif icant barriersbetween the top echelon and the rest of the classstructure The good fit of the homogeneousquasi-RC(II) model provides preliminary sup-port for the resource approach linking inequal-ity and mobility Mobility dynamics seem to bedriven largely by the hierarchical distancesbetween classes and overall Chile seems not tobe less fluid than the CASMIN countries

CCOOMMPPAARRIISSOONN OOFF TTHHEE OORRIIGGIINNAALL AANNDD CCHHIILLEEAANN

VVEERRSSIIOONNSS OOFF TTHHEE CCOORREE MMOODDEELL

Although Chilean mobility dynamics seem to bedriven by inequality across classes in a veryunequal society the fitting of the core modelprovides no indication that Chile is significantlyless fluid than depicted by the core model Asa preliminary analysis of the Chilean level of flu-idity I compare the predicted propensities formobility and immobility between specific pairsof classes in Chile with those in the CASMINcountries Propensities are obtained from theoriginal and Chilean versions of the core modelrespectively (models 2 and 5 of Table 3)

443388mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I begin with the immobility of those at the topof the class structure namely the service class(I+II) The service classrsquo propensity for immo-bility in the original core model is captured bytwo parameters IN1 and IN2 This propensityis more than three times what it would be in theabsence of these effects (e43+84 = 356) InChile the propensity for immobility of class I+IIis captured by the same two parameters and itis lower than what the core model expresses286 times larger than it would be in the absenceof these effects (e34+71) What about the immo-bility of the two agricultural classes In theCASMIN countries immobility of the IVc andVIIb classes is respectively 785 and 154 timeswhat it would be in the absence of these param-eters (e43+84+101ndash22 and e43 respectively) InChile the comparable values are 422 and 286(e34+71+48ndash09 and e43+71 respectively) indi-cating the relatively low inheritance associatedwith land ownership in the Latin Americannation Thus if attention is focused on theimmobility in the upper and lower ends of thesocial hierarchy the comparison does not indi-cate less fluidity in the highly unequal Chileansociety

What about the mobility rates between thetwo most distant classes in the hierarchical rank-ing namely the service class (I+II) and farmworkers (VIIb) In the case of the CASMINcountries the mobility chances between theseclasses are captured by four parameters (HI1HI2 SE1 and AF1) and is 077 times smallerthan it would be in the absence of these effects(endash22 ndash45ndash106ndash83) In Chile the distance fromclass I to class VIIb is expressed by these samefour parameters in addition to the asymmetri-cal disaffinity AF1-C This yields mobilitychances 056 times what would be obtained inthe absence of these effects (endash09 ndash47ndash20ndash63ndash149)Thus the barrier to long-range downwardmobility from the elite is more significant inChile than in the CASMIN countries The sameoccurs with long-range downward mobility fromclass III

If we measure the reciprocal flow (ieupward mobility from class VIIb to class I+II)we obtain the same value for the CASMINcountries because all parameters are symmet-rical in the original core model For Chile how-ever this calculation removes the asymmetricalbarrier to elite downward mobility which rais-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443399

2322-ASR 703 filename70304-torche

Figure 4 Class Scores Obtained from the Homogeneous Quasi-RC(II) Model and from Socioeconomic Status(SES) Ranking

Note (1) = Class standardized socioeconomic status scores based on schooling and earnings (2) = Class stan-dardized scores based on the Homogeneous Quasi-RC(II) model (see text for details)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 15: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

(HI1) long-range hierarchical mobility is moredifficult in this country (HI2) This suggestssubstantial inequality between the extremes ofthe class hierarchy combined with smaller dif-ferentiation in the middle of the distribution Asto the inheritance effects the service class theself-employed group and the farmers (classesI+II IVab and IVc respectively) have a sig-nificantly lower propensity to immobility inChile than in the CASMIN countries suggest-ing that class positions usually associated withindependent work are less able to reproducetheir class status intergenerationally than depict-ed by the core Most impressive the barrierbetween the agricultural and nonagriculturalsectors of the economy (SE1) is much weakerin Chile as is the affinity effect linking non-manual classes manual classes and the class-es that own capital among themselves (AF2)

Interestingly all the significant differences incoefficients with the single exception of the bar-rier to long-range hierarchical mobility sug-gest that Chile is more fluid than the core modeldepicts Higher fluidity seems to be driven bythe weakness of horizontal barriers separatingthe agricultural manual and self-employed sec-tors of the economy This is consistent withfindings of weaker sector effects in countriesthat have experienced rapid industrializationsuch as Korea (Park 2004) Brazil (Costa-Ribeiro 2003) Israel (Yaish 2004 Goldthorpeet al 1997) and Japan (Ishida et al 1991)

Additionally in the Chilean case the weak-ness of sector barriers may have been intensi-fied by the deep market transformation of the1970s and 1980s On the one hand the ldquochangeof handsrdquo in the agricultural sector induced bythe agrarian reform and counterreform likelyincreased fluidity between the agricultural andurban classes On the other hand the role of self-employment as an ephemeral refuge against(industrial and public sector) unemploymentmay have reduced barriers between the self-employed group and the rest of the social struc-ture Added to the larger difficulty of long-rangehierarchical mobility these features suggestweak sector cleavages in the Chilean fluiditypattern and high salience of the hierarchicaldimension of mobility

For the countries in which the fit of the coremodel is not adequate Erikson and Goldthorpe(1992a145) suggest adding parameters thataccount for country-specific deviations from

the model I therefore introduce adjustments toreflect salient characteristics of the Chilean flu-idity pattern producing a ldquoChilean versionrdquo ofthe core model

First I modify the hierarchical effects Fittingof the core model suggests that hierarchicalbarriers are a crucial determinant of mobility inChile This conclusion is however obscuredby the fact that the hierarchical strata distin-guished by the core model are not empiricallyobtained and may not accurately represent theChilean ranking of classes in terms of socioe-conomic status (SES) To model hierarchicaleffects in Chile adequately I rank classes usingthe unweighted average of their schooling andearning levels as a proxy for SES I then collapsethe seven classes into three hierarchical stratausing cluster analysis These three empiricallyobtained strata are the following Upper stratum(service class I+II) middle stratum (routinenonmanual and self-employed group III andIVab) and lower stratum (manual and agricul-tural classes V-VI VIIa IVc and VIIb) Detailson the ranking of classes cluster procedures andthe comparison of the CASMIN and empiricalChilean class ranking can be found in theAppendix Comparison of the CASMIN and theempirical class ranking suggests higher differ-entiation at the top than at the bottom of thesocial structure in the Chilean case

Model 2 in Table 2 uses the empiricallyobtained hierarchical strata (EHE) Theimprovement in fit is massive (the difference inBIC is 293 without using any degrees of free-dom) The model now accounts for 86 percentof the association under independence

The large improvement in fit when empiricalhierarchical strata are used challenges theassumption of an internationally homogeneoushierarchical class ranking At the empiricallevel there is no proof that the hierarchicalranking of classes assumed by the core modelfits all CASMIN countries9 Moreover someevidence points to significant cross-country dif-

443366mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

9 Erikson and Goldthorpe(1992a Table 22) pres-ent the average ranking of classes based on standardstatus measures but they do not show the empiricalranking of classes for each country If the occupa-tional composition of classes varies across countriesthere may be significant international differences inthe hierarchical ranking of classes

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ferences in the ranking of classes (Hout andHauser 1992 Sorensen 1992) At the theoreti-cal level the creators of the core model high-light the relevance of highly specific historicallyformed national attributes to explain interna-tional variation in mobility These national par-ticularities may very well result in differenthierarchical positions of classes across coun-tries10 Furthermore even if the homogeneousranking of classes imposed by the CASMINresearchers applied to most CASMIN coun-tries significant departures may be found inthe late industrialized countries currently beingadded to the comparative mobility project TheChilean findings suggest that instead of impos-ing international homogeneity to the hierarchi-cal dimension of mobility the ranking of classesshould be treated as nationally specific param-eters to be estimated as mobility researchersroutinely treat row and column marginals ofthe mobility table The issue is importantbecause the hierarchical dimension is the mostrelevant dimension of mobility (Hout andHauser 1992 Wong 1992) and inadequate mod-eling can bias assessment of not only hierar-chical effects of mobility but also other effects

Two other changes were introduced toaccount for Chilean-specific deviations fromthe core model First the IN2 effect wasrearranged to exclude the self-employed class(IVab) and to include the class of farm workers(VIIb) This rearrangement accounts for thefact that the Chilean self-employed classappeared less likely than depicted by the coremodel to reproduce its class position acrossgenerations and the inheritance level of Chileanfarm workers is higher than in the CASMINcountries These changes are expressed in theChilean-specific inheritance parameter IN2-CSecond an affinity parameter (AF1-C) wasadded to capture the difficulty of long-rangedownward elite mobility This negative asym-metrical affinity links flows from the service

class to the agricultural classes and from theroutine nonmanual class to the class of farmworkers Note that the AF1-C adds to the hier-archical distance between the two extremes ofthe class structure and to the negative affinitycaptured by AF1 to depict the extreme difficultyof long-range downward mobility from theChilean elite The fit of this lsquoChilean versionrsquoof the core model is reported in Table 2 model3 The improvement in BIC from model 2 is sub-stantial Furthermore even if the model fallsshort in terms of standard statistical criteria itaccounts for a large 91 of the associationunder independence thereby supporting theclaim that it adequately represents the Chileanmobility pattern The parameter estimates for theChilean version of the core are reported in row5 Table 3

Because of the core modelrsquos undesirable prop-erties I also use an alternative approach toexamine Chilean mobility patterns Thisapproach uses the now standard associationrowndashcolumn (II) (RC[II]) model (Goodman1979 see also Hout 1983 and Wong 1992) Inthis model both origin and destination classesare scaled so that the association can beexpressed as a linear-by-linear interaction by asingle parameter The multiplicative form ofthe model can be expressed as follows

Fij = iO j

D exp( microi νj)

where i indexes class of origin j indexes classof destination Fij is the expected frequency inthe (ij) cell is the grand mean i

O pertainsto the class of origin marginal effect j

D pertainsto the class of destination marginal effect isa global association parameter microi is the scalevalue for the ith class of origin and νj is the scalescore for the jth class of destination subject tothe following normalization constraints Σmicroi =Σνj = 0 (normalization of the location) and Σmicroi2

= Σνj2 = 1 (normalization of the scale)The class scale scores reflect a latent con-

tinuous variable made manifest by the class cat-egories Empirically obtained from the datathese ldquodistancerdquo scores produce an optimalranking of classes for the purpose of mobilityanalysis The RC(II) model is appropriate formodeling the association of any two ordinalvariables but mobility tables are distinctivebecause of the correspondence between class oforigin and class of destination (Gerber and Hout2004691)

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443377

2322-ASR 703 filename70304-torche

10 For instance during the socialist period inHungary the status of the proletariat may have beenindistinguishable from the status of the routine non-manual class (Szelenyi 199862) In Scotland givento early industrialization the hierarchical barrierbetween the skilled and unskilled working classes thatcharacterizes other CASMIN countries may not exist(Erikson and Goldthorpe 1992a163)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I introduce two adjustments to account forthis correspondence transforming the formu-lation into a ldquohybrid modelrdquo First topologicalparameters will account for the large counts insome diagonal cells reflecting immobility(inheritance) effects These parameters distin-guish two inheritance levels ldquohigh inheritancerdquofor the service class (I+II) farmers (IVc) andfarm workers (VIIb) and ldquolow inheritancerdquo forall other classes with the exception of the rou-tine nonmanual workers

Model 4 in Table 2 reports the fit statistic forthis quasi-RC(II) model The BIC statisticrejects it and prefers the more parsimoniousChilean version of the core model ThereforeI introduce a second adjustment to account forthe correspondence between rows and columnsin mobility tables Origin and destination scoresare constrained to be the same assuming thatthe scaling of classes has not significantlychanged over time Model 5 in Table 2 reportsfit statistics for this homogeneous quasi-RC(II)model On the basis of the BIC statistic the fitof the model is as good as that of the Chileanversion of the core model (Wong [1994] showsthat BIC differences of less than 5 points shouldbe considered indeterminate) Thus under stan-dard statistical criteria the Chilean core modeland the homogeneous quasi-RC(II) model areindistinguishable Given that the Chilean versionof the core model was tailored to fit the Chileantable the fit of model 5 is impressive and it indi-cates that a unidimensional scale can adequatelycapture mobility distances between classes

The logical next question is what this scalerepresents and specifically whether it repre-sents the hierarchical ranking of classes in termsof SES If a close correspondence is foundbetween the mobility distances among classesand their distances in terms of SES this find-ing would suggest that the intragenerationaldistribution of resources and rewards expressedin the SES ranking of classes drives the processof intergenerational mobility In other words thisfinding would support the resource approachrsquosclaim that differences in resources associatedwith diverse class positions determine mobili-ty opportunities across generations

To examine this possibility I compare theclass scores obtained from the homogeneousquasi-RC(II) model with the empiricallyobtained SES ranking of classes based on edu-cation and earnings (Appendix) Figure 4 pres-

ents the comparison between the standardizedvalues of both sets of scores The correspon-dence is striking The mobility distancesbetween classes closely reproduce their dis-tances in terms of SES This close isomorphismsuggests that at least in Chile the intragener-ational inequality in the distribution of socialresources and rewards across classes largelydrives the intergenerational mobility process Totest the isomorphism further I use the SESclass scores as row and column rankings to esti-mate a quasindashlinear-by-linear association model(model 6 Table 2) The fit of model 6 is sig-nificantly worse than that of model 5 almost aspoor as that of the original core model This indi-cates that although the inter- and intragenera-tional dimensions of inequality are very similarin Chile they have a somewhat different struc-ture

In summary the central characteristic of theChilean mobility regime is the predominance oflong-range hierarchical barriers and the weak-ness of sector cleavages separating classes in themiddle and at the lower end of the hierarchicalstructure Findings from the ldquoChilean versionrdquoof the core model and the quasi-RC(II) hybridmodels are consistent depicting a social struc-ture characterized by signif icant barriersbetween the top echelon and the rest of the classstructure The good fit of the homogeneousquasi-RC(II) model provides preliminary sup-port for the resource approach linking inequal-ity and mobility Mobility dynamics seem to bedriven largely by the hierarchical distancesbetween classes and overall Chile seems not tobe less fluid than the CASMIN countries

CCOOMMPPAARRIISSOONN OOFF TTHHEE OORRIIGGIINNAALL AANNDD CCHHIILLEEAANN

VVEERRSSIIOONNSS OOFF TTHHEE CCOORREE MMOODDEELL

Although Chilean mobility dynamics seem to bedriven by inequality across classes in a veryunequal society the fitting of the core modelprovides no indication that Chile is significantlyless fluid than depicted by the core model Asa preliminary analysis of the Chilean level of flu-idity I compare the predicted propensities formobility and immobility between specific pairsof classes in Chile with those in the CASMINcountries Propensities are obtained from theoriginal and Chilean versions of the core modelrespectively (models 2 and 5 of Table 3)

443388mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I begin with the immobility of those at the topof the class structure namely the service class(I+II) The service classrsquo propensity for immo-bility in the original core model is captured bytwo parameters IN1 and IN2 This propensityis more than three times what it would be in theabsence of these effects (e43+84 = 356) InChile the propensity for immobility of class I+IIis captured by the same two parameters and itis lower than what the core model expresses286 times larger than it would be in the absenceof these effects (e34+71) What about the immo-bility of the two agricultural classes In theCASMIN countries immobility of the IVc andVIIb classes is respectively 785 and 154 timeswhat it would be in the absence of these param-eters (e43+84+101ndash22 and e43 respectively) InChile the comparable values are 422 and 286(e34+71+48ndash09 and e43+71 respectively) indi-cating the relatively low inheritance associatedwith land ownership in the Latin Americannation Thus if attention is focused on theimmobility in the upper and lower ends of thesocial hierarchy the comparison does not indi-cate less fluidity in the highly unequal Chileansociety

What about the mobility rates between thetwo most distant classes in the hierarchical rank-ing namely the service class (I+II) and farmworkers (VIIb) In the case of the CASMINcountries the mobility chances between theseclasses are captured by four parameters (HI1HI2 SE1 and AF1) and is 077 times smallerthan it would be in the absence of these effects(endash22 ndash45ndash106ndash83) In Chile the distance fromclass I to class VIIb is expressed by these samefour parameters in addition to the asymmetri-cal disaffinity AF1-C This yields mobilitychances 056 times what would be obtained inthe absence of these effects (endash09 ndash47ndash20ndash63ndash149)Thus the barrier to long-range downwardmobility from the elite is more significant inChile than in the CASMIN countries The sameoccurs with long-range downward mobility fromclass III

If we measure the reciprocal flow (ieupward mobility from class VIIb to class I+II)we obtain the same value for the CASMINcountries because all parameters are symmet-rical in the original core model For Chile how-ever this calculation removes the asymmetricalbarrier to elite downward mobility which rais-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443399

2322-ASR 703 filename70304-torche

Figure 4 Class Scores Obtained from the Homogeneous Quasi-RC(II) Model and from Socioeconomic Status(SES) Ranking

Note (1) = Class standardized socioeconomic status scores based on schooling and earnings (2) = Class stan-dardized scores based on the Homogeneous Quasi-RC(II) model (see text for details)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 16: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

ferences in the ranking of classes (Hout andHauser 1992 Sorensen 1992) At the theoreti-cal level the creators of the core model high-light the relevance of highly specific historicallyformed national attributes to explain interna-tional variation in mobility These national par-ticularities may very well result in differenthierarchical positions of classes across coun-tries10 Furthermore even if the homogeneousranking of classes imposed by the CASMINresearchers applied to most CASMIN coun-tries significant departures may be found inthe late industrialized countries currently beingadded to the comparative mobility project TheChilean findings suggest that instead of impos-ing international homogeneity to the hierarchi-cal dimension of mobility the ranking of classesshould be treated as nationally specific param-eters to be estimated as mobility researchersroutinely treat row and column marginals ofthe mobility table The issue is importantbecause the hierarchical dimension is the mostrelevant dimension of mobility (Hout andHauser 1992 Wong 1992) and inadequate mod-eling can bias assessment of not only hierar-chical effects of mobility but also other effects

Two other changes were introduced toaccount for Chilean-specific deviations fromthe core model First the IN2 effect wasrearranged to exclude the self-employed class(IVab) and to include the class of farm workers(VIIb) This rearrangement accounts for thefact that the Chilean self-employed classappeared less likely than depicted by the coremodel to reproduce its class position acrossgenerations and the inheritance level of Chileanfarm workers is higher than in the CASMINcountries These changes are expressed in theChilean-specific inheritance parameter IN2-CSecond an affinity parameter (AF1-C) wasadded to capture the difficulty of long-rangedownward elite mobility This negative asym-metrical affinity links flows from the service

class to the agricultural classes and from theroutine nonmanual class to the class of farmworkers Note that the AF1-C adds to the hier-archical distance between the two extremes ofthe class structure and to the negative affinitycaptured by AF1 to depict the extreme difficultyof long-range downward mobility from theChilean elite The fit of this lsquoChilean versionrsquoof the core model is reported in Table 2 model3 The improvement in BIC from model 2 is sub-stantial Furthermore even if the model fallsshort in terms of standard statistical criteria itaccounts for a large 91 of the associationunder independence thereby supporting theclaim that it adequately represents the Chileanmobility pattern The parameter estimates for theChilean version of the core are reported in row5 Table 3

Because of the core modelrsquos undesirable prop-erties I also use an alternative approach toexamine Chilean mobility patterns Thisapproach uses the now standard associationrowndashcolumn (II) (RC[II]) model (Goodman1979 see also Hout 1983 and Wong 1992) Inthis model both origin and destination classesare scaled so that the association can beexpressed as a linear-by-linear interaction by asingle parameter The multiplicative form ofthe model can be expressed as follows

Fij = iO j

D exp( microi νj)

where i indexes class of origin j indexes classof destination Fij is the expected frequency inthe (ij) cell is the grand mean i

O pertainsto the class of origin marginal effect j

D pertainsto the class of destination marginal effect isa global association parameter microi is the scalevalue for the ith class of origin and νj is the scalescore for the jth class of destination subject tothe following normalization constraints Σmicroi =Σνj = 0 (normalization of the location) and Σmicroi2

= Σνj2 = 1 (normalization of the scale)The class scale scores reflect a latent con-

tinuous variable made manifest by the class cat-egories Empirically obtained from the datathese ldquodistancerdquo scores produce an optimalranking of classes for the purpose of mobilityanalysis The RC(II) model is appropriate formodeling the association of any two ordinalvariables but mobility tables are distinctivebecause of the correspondence between class oforigin and class of destination (Gerber and Hout2004691)

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443377

2322-ASR 703 filename70304-torche

10 For instance during the socialist period inHungary the status of the proletariat may have beenindistinguishable from the status of the routine non-manual class (Szelenyi 199862) In Scotland givento early industrialization the hierarchical barrierbetween the skilled and unskilled working classes thatcharacterizes other CASMIN countries may not exist(Erikson and Goldthorpe 1992a163)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I introduce two adjustments to account forthis correspondence transforming the formu-lation into a ldquohybrid modelrdquo First topologicalparameters will account for the large counts insome diagonal cells reflecting immobility(inheritance) effects These parameters distin-guish two inheritance levels ldquohigh inheritancerdquofor the service class (I+II) farmers (IVc) andfarm workers (VIIb) and ldquolow inheritancerdquo forall other classes with the exception of the rou-tine nonmanual workers

Model 4 in Table 2 reports the fit statistic forthis quasi-RC(II) model The BIC statisticrejects it and prefers the more parsimoniousChilean version of the core model ThereforeI introduce a second adjustment to account forthe correspondence between rows and columnsin mobility tables Origin and destination scoresare constrained to be the same assuming thatthe scaling of classes has not significantlychanged over time Model 5 in Table 2 reportsfit statistics for this homogeneous quasi-RC(II)model On the basis of the BIC statistic the fitof the model is as good as that of the Chileanversion of the core model (Wong [1994] showsthat BIC differences of less than 5 points shouldbe considered indeterminate) Thus under stan-dard statistical criteria the Chilean core modeland the homogeneous quasi-RC(II) model areindistinguishable Given that the Chilean versionof the core model was tailored to fit the Chileantable the fit of model 5 is impressive and it indi-cates that a unidimensional scale can adequatelycapture mobility distances between classes

The logical next question is what this scalerepresents and specifically whether it repre-sents the hierarchical ranking of classes in termsof SES If a close correspondence is foundbetween the mobility distances among classesand their distances in terms of SES this find-ing would suggest that the intragenerationaldistribution of resources and rewards expressedin the SES ranking of classes drives the processof intergenerational mobility In other words thisfinding would support the resource approachrsquosclaim that differences in resources associatedwith diverse class positions determine mobili-ty opportunities across generations

To examine this possibility I compare theclass scores obtained from the homogeneousquasi-RC(II) model with the empiricallyobtained SES ranking of classes based on edu-cation and earnings (Appendix) Figure 4 pres-

ents the comparison between the standardizedvalues of both sets of scores The correspon-dence is striking The mobility distancesbetween classes closely reproduce their dis-tances in terms of SES This close isomorphismsuggests that at least in Chile the intragener-ational inequality in the distribution of socialresources and rewards across classes largelydrives the intergenerational mobility process Totest the isomorphism further I use the SESclass scores as row and column rankings to esti-mate a quasindashlinear-by-linear association model(model 6 Table 2) The fit of model 6 is sig-nificantly worse than that of model 5 almost aspoor as that of the original core model This indi-cates that although the inter- and intragenera-tional dimensions of inequality are very similarin Chile they have a somewhat different struc-ture

In summary the central characteristic of theChilean mobility regime is the predominance oflong-range hierarchical barriers and the weak-ness of sector cleavages separating classes in themiddle and at the lower end of the hierarchicalstructure Findings from the ldquoChilean versionrdquoof the core model and the quasi-RC(II) hybridmodels are consistent depicting a social struc-ture characterized by signif icant barriersbetween the top echelon and the rest of the classstructure The good fit of the homogeneousquasi-RC(II) model provides preliminary sup-port for the resource approach linking inequal-ity and mobility Mobility dynamics seem to bedriven largely by the hierarchical distancesbetween classes and overall Chile seems not tobe less fluid than the CASMIN countries

CCOOMMPPAARRIISSOONN OOFF TTHHEE OORRIIGGIINNAALL AANNDD CCHHIILLEEAANN

VVEERRSSIIOONNSS OOFF TTHHEE CCOORREE MMOODDEELL

Although Chilean mobility dynamics seem to bedriven by inequality across classes in a veryunequal society the fitting of the core modelprovides no indication that Chile is significantlyless fluid than depicted by the core model Asa preliminary analysis of the Chilean level of flu-idity I compare the predicted propensities formobility and immobility between specific pairsof classes in Chile with those in the CASMINcountries Propensities are obtained from theoriginal and Chilean versions of the core modelrespectively (models 2 and 5 of Table 3)

443388mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I begin with the immobility of those at the topof the class structure namely the service class(I+II) The service classrsquo propensity for immo-bility in the original core model is captured bytwo parameters IN1 and IN2 This propensityis more than three times what it would be in theabsence of these effects (e43+84 = 356) InChile the propensity for immobility of class I+IIis captured by the same two parameters and itis lower than what the core model expresses286 times larger than it would be in the absenceof these effects (e34+71) What about the immo-bility of the two agricultural classes In theCASMIN countries immobility of the IVc andVIIb classes is respectively 785 and 154 timeswhat it would be in the absence of these param-eters (e43+84+101ndash22 and e43 respectively) InChile the comparable values are 422 and 286(e34+71+48ndash09 and e43+71 respectively) indi-cating the relatively low inheritance associatedwith land ownership in the Latin Americannation Thus if attention is focused on theimmobility in the upper and lower ends of thesocial hierarchy the comparison does not indi-cate less fluidity in the highly unequal Chileansociety

What about the mobility rates between thetwo most distant classes in the hierarchical rank-ing namely the service class (I+II) and farmworkers (VIIb) In the case of the CASMINcountries the mobility chances between theseclasses are captured by four parameters (HI1HI2 SE1 and AF1) and is 077 times smallerthan it would be in the absence of these effects(endash22 ndash45ndash106ndash83) In Chile the distance fromclass I to class VIIb is expressed by these samefour parameters in addition to the asymmetri-cal disaffinity AF1-C This yields mobilitychances 056 times what would be obtained inthe absence of these effects (endash09 ndash47ndash20ndash63ndash149)Thus the barrier to long-range downwardmobility from the elite is more significant inChile than in the CASMIN countries The sameoccurs with long-range downward mobility fromclass III

If we measure the reciprocal flow (ieupward mobility from class VIIb to class I+II)we obtain the same value for the CASMINcountries because all parameters are symmet-rical in the original core model For Chile how-ever this calculation removes the asymmetricalbarrier to elite downward mobility which rais-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443399

2322-ASR 703 filename70304-torche

Figure 4 Class Scores Obtained from the Homogeneous Quasi-RC(II) Model and from Socioeconomic Status(SES) Ranking

Note (1) = Class standardized socioeconomic status scores based on schooling and earnings (2) = Class stan-dardized scores based on the Homogeneous Quasi-RC(II) model (see text for details)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 17: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

I introduce two adjustments to account forthis correspondence transforming the formu-lation into a ldquohybrid modelrdquo First topologicalparameters will account for the large counts insome diagonal cells reflecting immobility(inheritance) effects These parameters distin-guish two inheritance levels ldquohigh inheritancerdquofor the service class (I+II) farmers (IVc) andfarm workers (VIIb) and ldquolow inheritancerdquo forall other classes with the exception of the rou-tine nonmanual workers

Model 4 in Table 2 reports the fit statistic forthis quasi-RC(II) model The BIC statisticrejects it and prefers the more parsimoniousChilean version of the core model ThereforeI introduce a second adjustment to account forthe correspondence between rows and columnsin mobility tables Origin and destination scoresare constrained to be the same assuming thatthe scaling of classes has not significantlychanged over time Model 5 in Table 2 reportsfit statistics for this homogeneous quasi-RC(II)model On the basis of the BIC statistic the fitof the model is as good as that of the Chileanversion of the core model (Wong [1994] showsthat BIC differences of less than 5 points shouldbe considered indeterminate) Thus under stan-dard statistical criteria the Chilean core modeland the homogeneous quasi-RC(II) model areindistinguishable Given that the Chilean versionof the core model was tailored to fit the Chileantable the fit of model 5 is impressive and it indi-cates that a unidimensional scale can adequatelycapture mobility distances between classes

The logical next question is what this scalerepresents and specifically whether it repre-sents the hierarchical ranking of classes in termsof SES If a close correspondence is foundbetween the mobility distances among classesand their distances in terms of SES this find-ing would suggest that the intragenerationaldistribution of resources and rewards expressedin the SES ranking of classes drives the processof intergenerational mobility In other words thisfinding would support the resource approachrsquosclaim that differences in resources associatedwith diverse class positions determine mobili-ty opportunities across generations

To examine this possibility I compare theclass scores obtained from the homogeneousquasi-RC(II) model with the empiricallyobtained SES ranking of classes based on edu-cation and earnings (Appendix) Figure 4 pres-

ents the comparison between the standardizedvalues of both sets of scores The correspon-dence is striking The mobility distancesbetween classes closely reproduce their dis-tances in terms of SES This close isomorphismsuggests that at least in Chile the intragener-ational inequality in the distribution of socialresources and rewards across classes largelydrives the intergenerational mobility process Totest the isomorphism further I use the SESclass scores as row and column rankings to esti-mate a quasindashlinear-by-linear association model(model 6 Table 2) The fit of model 6 is sig-nificantly worse than that of model 5 almost aspoor as that of the original core model This indi-cates that although the inter- and intragenera-tional dimensions of inequality are very similarin Chile they have a somewhat different struc-ture

In summary the central characteristic of theChilean mobility regime is the predominance oflong-range hierarchical barriers and the weak-ness of sector cleavages separating classes in themiddle and at the lower end of the hierarchicalstructure Findings from the ldquoChilean versionrdquoof the core model and the quasi-RC(II) hybridmodels are consistent depicting a social struc-ture characterized by signif icant barriersbetween the top echelon and the rest of the classstructure The good fit of the homogeneousquasi-RC(II) model provides preliminary sup-port for the resource approach linking inequal-ity and mobility Mobility dynamics seem to bedriven largely by the hierarchical distancesbetween classes and overall Chile seems not tobe less fluid than the CASMIN countries

CCOOMMPPAARRIISSOONN OOFF TTHHEE OORRIIGGIINNAALL AANNDD CCHHIILLEEAANN

VVEERRSSIIOONNSS OOFF TTHHEE CCOORREE MMOODDEELL

Although Chilean mobility dynamics seem to bedriven by inequality across classes in a veryunequal society the fitting of the core modelprovides no indication that Chile is significantlyless fluid than depicted by the core model Asa preliminary analysis of the Chilean level of flu-idity I compare the predicted propensities formobility and immobility between specific pairsof classes in Chile with those in the CASMINcountries Propensities are obtained from theoriginal and Chilean versions of the core modelrespectively (models 2 and 5 of Table 3)

443388mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

I begin with the immobility of those at the topof the class structure namely the service class(I+II) The service classrsquo propensity for immo-bility in the original core model is captured bytwo parameters IN1 and IN2 This propensityis more than three times what it would be in theabsence of these effects (e43+84 = 356) InChile the propensity for immobility of class I+IIis captured by the same two parameters and itis lower than what the core model expresses286 times larger than it would be in the absenceof these effects (e34+71) What about the immo-bility of the two agricultural classes In theCASMIN countries immobility of the IVc andVIIb classes is respectively 785 and 154 timeswhat it would be in the absence of these param-eters (e43+84+101ndash22 and e43 respectively) InChile the comparable values are 422 and 286(e34+71+48ndash09 and e43+71 respectively) indi-cating the relatively low inheritance associatedwith land ownership in the Latin Americannation Thus if attention is focused on theimmobility in the upper and lower ends of thesocial hierarchy the comparison does not indi-cate less fluidity in the highly unequal Chileansociety

What about the mobility rates between thetwo most distant classes in the hierarchical rank-ing namely the service class (I+II) and farmworkers (VIIb) In the case of the CASMINcountries the mobility chances between theseclasses are captured by four parameters (HI1HI2 SE1 and AF1) and is 077 times smallerthan it would be in the absence of these effects(endash22 ndash45ndash106ndash83) In Chile the distance fromclass I to class VIIb is expressed by these samefour parameters in addition to the asymmetri-cal disaffinity AF1-C This yields mobilitychances 056 times what would be obtained inthe absence of these effects (endash09 ndash47ndash20ndash63ndash149)Thus the barrier to long-range downwardmobility from the elite is more significant inChile than in the CASMIN countries The sameoccurs with long-range downward mobility fromclass III

If we measure the reciprocal flow (ieupward mobility from class VIIb to class I+II)we obtain the same value for the CASMINcountries because all parameters are symmet-rical in the original core model For Chile how-ever this calculation removes the asymmetricalbarrier to elite downward mobility which rais-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443399

2322-ASR 703 filename70304-torche

Figure 4 Class Scores Obtained from the Homogeneous Quasi-RC(II) Model and from Socioeconomic Status(SES) Ranking

Note (1) = Class standardized socioeconomic status scores based on schooling and earnings (2) = Class stan-dardized scores based on the Homogeneous Quasi-RC(II) model (see text for details)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 18: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

I begin with the immobility of those at the topof the class structure namely the service class(I+II) The service classrsquo propensity for immo-bility in the original core model is captured bytwo parameters IN1 and IN2 This propensityis more than three times what it would be in theabsence of these effects (e43+84 = 356) InChile the propensity for immobility of class I+IIis captured by the same two parameters and itis lower than what the core model expresses286 times larger than it would be in the absenceof these effects (e34+71) What about the immo-bility of the two agricultural classes In theCASMIN countries immobility of the IVc andVIIb classes is respectively 785 and 154 timeswhat it would be in the absence of these param-eters (e43+84+101ndash22 and e43 respectively) InChile the comparable values are 422 and 286(e34+71+48ndash09 and e43+71 respectively) indi-cating the relatively low inheritance associatedwith land ownership in the Latin Americannation Thus if attention is focused on theimmobility in the upper and lower ends of thesocial hierarchy the comparison does not indi-cate less fluidity in the highly unequal Chileansociety

What about the mobility rates between thetwo most distant classes in the hierarchical rank-ing namely the service class (I+II) and farmworkers (VIIb) In the case of the CASMINcountries the mobility chances between theseclasses are captured by four parameters (HI1HI2 SE1 and AF1) and is 077 times smallerthan it would be in the absence of these effects(endash22 ndash45ndash106ndash83) In Chile the distance fromclass I to class VIIb is expressed by these samefour parameters in addition to the asymmetri-cal disaffinity AF1-C This yields mobilitychances 056 times what would be obtained inthe absence of these effects (endash09 ndash47ndash20ndash63ndash149)Thus the barrier to long-range downwardmobility from the elite is more significant inChile than in the CASMIN countries The sameoccurs with long-range downward mobility fromclass III

If we measure the reciprocal flow (ieupward mobility from class VIIb to class I+II)we obtain the same value for the CASMINcountries because all parameters are symmet-rical in the original core model For Chile how-ever this calculation removes the asymmetricalbarrier to elite downward mobility which rais-

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash443399

2322-ASR 703 filename70304-torche

Figure 4 Class Scores Obtained from the Homogeneous Quasi-RC(II) Model and from Socioeconomic Status(SES) Ranking

Note (1) = Class standardized socioeconomic status scores based on schooling and earnings (2) = Class stan-dardized scores based on the Homogeneous Quasi-RC(II) model (see text for details)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 19: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

es the chances of mobility to 25 of what itwould be in the absence of these effects(endash09 ndash47ndash20ndash63) Surprisingly for a society withsuch a high level of inequality the barrier tolong-range upward mobility is significantly lessthan in the CASMIN countries

What about the relative chances of movingbetween classes that are closer in terms of sta-tus The Chilean version of the core model sug-gests that the pattern should be as fluid in Chileif not more so because of the weak sectoreffects For example in CASMIN countriesthe propensity to move from the farmer class(IVc) to the unskilled manual class (VIIa) is 44times what it would be in the absence of theseeffects (endash22ndash106+45) In contrast in Chile thereis higher fluidity with mobility 128 times whatneutral mobility would be (endash20+45) Againthere is no indication of more rigidity in theChilean fluidity pattern

Finally I compare the mobility flows betweenthe voluminous skilled and unskilled workingclasses Whereas in Chile there is no effectmodifying the fluidity between classes V-VIand VIIa in the CASMIN countries they areseparated by a hierarchical effect but connect-ed by an affinity effect This marginally increas-es fluidity to 108 of what it would be in theabsence of this effect (endash26+34) indicating nosignificant international differences

These case-by-case comparisons suggest thatalthough hierarchical effects especially thoseassociated with long-range downward mobili-ty from the elite are stronger in Chile sectorbarriers are weaker yielding a pattern that isas fluid if not more so than that in advancedindustrial nations

TTHHEE LLEEVVEELL OOFF FFLLUUIIDDIITTYY IINN CCHHIILLEE AAMMUULLTTIICCOOUUNNTTRRYY CCOOMMPPAARRIISSOONN

To explore further the level of fluidity in Chilein an international context I compare thestrength of the originndashdestination associationbetween Chile and seven industrialized coun-tries England France Sweden IrelandScotland the United States and Israel Therationale for including these countries is as fol-lows England and France are the central coun-tries from which the core model was derived(Erikson and Goldthorpe 1987a 1987b 1992a)Sweden and the United States are among themost fluid countries in the CASMIN pool

Scotland and Ireland are found to be amongthe most rigid nations within the CASMIN set(Erikson and Goldthorpe 1992a chapter 11)and Israel is the most fluid society in whichempirical studies have been conducted signif-icantly more fluid than any of the CASMINcountries (Goldthorpe et al 1997 Yaish 2000)11

In order to assess cross-national variation insocial fluidity I fit a set of models for the three-way table of class of origin class of destinationand country (Table 4) The model of condi-tional independence assuming no associationbetween origins and destinations across coun-tries given different national margins (Table 4column 1) is presented as a baseline againstwhich other models may be assessed As expect-ed the fit is very poor indicating significant ori-gin-destination association in the countriesanalyzed

The second model tested is that of ldquocommonfluidityrdquo which postulates that the strength ofthe origin-destination association is the sameacross countries12 As column 2 of Table 4shows the model significantly improves thefit when compared with conditional inde-pendence (L2 = 9872 df = 252 BIC = ndash17678)Although the model does not fit the data wellunder standard statistical criteria it accountsfor a large 96 percent of the association underindependence

I then turn to the question of internationalvariation To test the hypothesis that the strengthof the originndashdestination association variesacross countries I use a model independentlydeveloped by Xie (1992) and Erikson andGoldthorpe (1992a) and known as the multi-plicative layer effect or uniform difference(UNIDIFF) The multiplicative formulation ofthe model is the following

444400mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

11 Data on England France Sweden the UnitedStates Ireland and Scotland were obtained from theCASMIN dataset Data on Israel were obtained fromthe 1991 Israeli Social Mobility Survey Samplesare reduced to men ages 25 to 64 years in the CAS-MIN countries and to Jewish men ages 25 to 64years in Israel

12 Note that this is not the core model of fluidityestimated in the previous section but a full interac-tion model which uses one parameter for each cellof the table but constrains the parameters to behomogeneous across countries

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 20: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

Fijk = iO j

D kC ik

OC jkDC exp(ijk)

where i indexes class of origin j indexes classof destination k indexes country Fijk is theexpected frequency in the (ijk) cell the parameters are subject to the ANOVA-type nor-malization constraint that they multiply to 1along all appropriate dimensions n representsthe grand mean i

O pertains to the class of ori-gin marginal effect j

D pertains to the class ofdestination marginal effect k

C pertains to thecountry marginal effect ij describes the ori-ginndashdestination association over all countriesand the ks describe the country-specific devi-ation from the overall association The extent ofassociation in country k is now the product oftwo components the originndashdestination asso-ciation common to all countries and the nation-al deviation parameter k13

As indicated by column 3 of Table 4 the fitof the model is significantly improved whenthe strength of association is allowed to varyacross countries The ks indicate country-spe-cific departures from the overall level of asso-ciation14 The larger the country-specif ic

deviation k parameter the stronger the ori-ginndashdestination association in country k ie theless fluid the country is Comparison of theparameters yields a striking conclusion Chileis more fluid than any of the advanced Europeancountries and has a level of fluidity in betweenthe highly fluid United States and Israeli soci-eties Finding high fluidity in a developing coun-try is not completely novel In fact Park (2004)demonstrated that Korea is more fluid thanFrance England and even Sweden Howevereconomic inequality in Korea is comparablewith that in advanced industrial nations(Deininger and Squire 1996) What contradictsa resource approach-based expectation is tofind high fluidity in a country with one of thehighest levels of inequality in the world15

TTHHEE LLEEVVEELL OOFF SSOOCCIIAALL OOPPEENNNNEESSSS IINN CCHHIILLEETTEEMMPPOORRAALL CCOOMMPPAARRIISSOONN

To explore further the association betweenmobility and inequality I examine the changein mobility rates over time in Chile SpecificallyI test whether fluidity has changed across threeperiods the redistributive period (1964ndash1973)the market transformation period (1974ndash1988)and the growth and democratization period

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444411

2322-ASR 703 filename70304-torche

Table 4 Fit Statistics for Mobility Models in Eight Countries

1 Conditional 2 CommonModel Indendence Social Fluidity 3 UNIDIFF

L2 169717 9872 6397df 2880 2520 2450BIC 138230 ndash17678 ndash20388Association Explained 00 942 962

Israel 25 Chile 26 USA 30 Sweden 31 England 39 France 41 Ireland 42 Scotland 43

Note = country-specific deviation from overall origin-destination association (see text for details)

13 Alternatively I could have used the log-additivelayer model (Yamaguchi 1987) The problem with thismodel is its requirement that the origin and destina-tion categories be correctly ordered (in terms ofmobility distances) which produces a different resultfor each combination of origin and destination order-ing (Goodman and Hout 1998 Xie 1992)

14 Following Xiersquos (1992) formulation the scale ofthe k parameters is normalized so that Σk

2 equals 1

15 Note that I cannot say anything about the sourcesof fluidity in a particular country by using thismethod Instead of a summary test such as the oneused here this would require a local test in which spe-cific sets of cells are modeled (Wong 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 21: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

(1989ndash2000) Given the significant increase ininequality during the market reform (as Figure3 indicates the period-average Gini index grewfrom 49 to 55) an association between inequal-ity and fluidity should express itself as a signif-icant change in mobility rates during that period

The advantage of a trend analysis is that byfocusing on a single country it controls forunobserved factors producing international vari-ation However because the data come from asingle cross-sectional study I divide the sampleinto three successive birth cohorts and interpretintercohort change as period effects The firstcohort (born between 1937 and 1943) reachedoccupational maturitymdashdefined in this analysisas 30 years oldmdashbetween 1967 and 1973 dur-ing the redistributive period The second cohort(born between 1944 and 1958) reached occu-pational maturity during the market transfor-mation period (1974ndash1988) The third cohort(born between 1959 and 1970) reached occu-pational maturity during the period of sustainedeconomic growth and democratization(1989ndash2000) As is well known a limitation ofcohort analysis is the inability to distinguishbetween life cycle (age) period and cohort inter-pretations of change (Ryder 1965)16 The poten-tially confounding effects of life cycle differencesare minimized by including only individualswho have reached occupational maturity underthe assumption that there is little career mobil-ity after that point (Goldthorpe 1980 Heath andPayne 1999)17 However the analysis cannotdistinguish between cohort and period interpre-tations of change To assess changes over timeI use the UNIDIFF model introduced in the pre-vious section

Panel A in Table 5 presents the parameterestimates for the model of conditional inde-pendence (model 1) the ldquoconstant social fluid-ityrdquo model (model 2) and the UNIDIFF model(model 3) The conditional independence model

assumes no origin-destination association andas expected fits the data poorly The key modelcomparison is between the constant fluiditymodel assuming no variation across historicalperiods and the UNIDIFF model which postu-lates a significant change in the level of fluidi-ty over time A comparison of models 3 and 2indicates that allowing the originndashdestinationassociation to vary across cohorts does not leadto a significant improvement of fit over that ofthe constant fluidity model The model com-parison therefore indicates that mobility didnot decline as a result of the growth in eco-nomic inequality during the market transfor-mation nor is there any change associated withthe slight reduction of inequality and significanteconomic growth during the growth and rede-mocratization period

To provide a more robust assessment of trendsa replicate analysis is presented in panel Bwhich replaces current class position with classposition in first job as class destination Becausefirst job identifies the same life cycle stage forall respondents this formulation presents analternative way of controlling for the con-founding of age and period Findings are insen-sitive to the specification of class of destinationAs panel B shows the model allowing for changein fluidity across cohorts (model 3) fits the datasignificantly worse than the model assumingconstant fluidity over time (model 2)

In summary the cohort analysis shows nosignificant change in mobility rates over timedespite a growth in inequality during the mar-ket reform This finding suggests no associationbetween mobility and inequality An alternativeexplanation can be offered however I haveshown that economic inequality significantlyincreased during the market transformation(1974ndash1988) At the same time some compo-nents of the market reform may have weakenedthe traditional barriers across sectors of the econ-omy The agrarian counterreform and conse-quent marketization of the countryside likelyreduced the barriers separating agricultural class-es and the growth and diversification of theself-employed sector likely weakened the barri-ers separating the self-employed classes from therest of the Chilean class structure These process-es may have induced two types of mobilitychange going in opposite directions On the onehand an increase in economic inequality wouldhave heightened hierarchical barriers to mobil-

444422mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

16 It also is important to note that as a cohortgrows older it suffers attrition by some membersThus the groups analyzed here are not true cohortsbut their current survivors (Goldthorpe 1980)

17 Studies in industrialized countries usually defineoccupational maturity as the age of 35 years Forthis study 30 years of age was chosen for the Chileancase because of the earlier entry to work that char-acterizes developing nations

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 22: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

ity On the other hand marketization of the coun-tryside and transformation of the self-employedsector would have weakened sector barriers tomobility thereby inducing horizontal fluidityBecause these two effects counteract each otherthe aggregate result may have led to no changein the total level of fluidity Granted this inter-pretation is speculative at the moment but itsderivation from historical evidence and its con-sistency with observed trends make it a plausi-ble account of mobility trends in Chile18

SSUUMMMMAARRYY AANNDD CCOONNCCLLUUSSIIOONN

CCHHIILLEEAANN EEXXCCEEPPTTIIOONNAALLIISSMM AANNDD AA

RREEDDEEFFIINNIITTIIOONN OOFF TTHHEE MMOOBBIILLIITTYYndashndashIINNEEQQUUAALLIITTYY

RREELLAATTIIOONNSSHHIIPP

This article uses an international comparativestrategy to analyze the Chilean mobility regimeand exploits the particularities of the Chileancase to contribute to the international compar-

ative study of mobility The findings indicatethat the Chilean mobility dynamics are definedby strong hierarchical effects combined withweak horizontal barriers separating sectors ofthe economy In fact a single unidimensionalscale effectively captures the Chilean mobilityopportunities and this scale closely reflects thevertical distances between classes in terms ofearnings and education In other words theintergenerational mobility chances to a largeextent mirror the contemporaneous distributionof rewards and resources across Chilean class-es This finding is fully consistent with theresource approach which maintains that mobil-ity opportunities are largely driven by differen-tial access to resources across classesInternational and temporal comparative analy-ses seem to contradict however the associationbetween inequality and mobility posed by theresource perspective Comparison with sevenindustrialized countries shows that Chile is high-ly fluid as fluid as the most open nations in theworld despite its great inequality Furthermorethe analysis of Chilean mobility trends showsconstant mobility rates over time despite thesignificant increase in inequality during themarket transformation period

These findings depict Chile as an exceptionalcase in which high inequality does not seem todepress mobility opportunities However if weconsider the pattern of inequality and not only

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444433

2322-ASR 703 filename70304-torche

Table 5 Goodness of Fit Statistics for Chilean Mobility Models across Cohorts

1 Conditional 2 Common Model Independence Social Fluidity 3 UNIDIFF

Panel A Fatherrsquos Class Position Current Class PositionmdashL2 8221 1374 1300mdashdf 1080 720 700mdashBIC ndash313 ndash4319 ndash4244mdashAssociation explained 00 830 840mdash Redistributive period 57mdash Market reform period 51mdash Growth and redemocratization period 66

Panel B Fatherrsquos Class Position Class Position First JobmdashL2 12347 1160 1121mdashdf 1080 720 700mdashBIC 3809 ndash4532 ndash4413mdashAssociation explained 00 906 906mdash Redistributive period 55mdash Market reform period 54mdash Growth and redemocratization period 63

Note = Country-specific deviation from overall origin-destination association (see text for details)

18 An empirical strategy to test this hypothesiswould be to evaluate the change over time in mag-nitude of the different mobility effects of the Chileanversion of the core model Unfortunately given therelatively small number of cases in the cohort-spe-cific tables coefficients representing temporal changeare mostly statistically insignificant thus not allow-ing a conclusive test

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 23: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

its aggregate level the Chilean exceptionalismdisappears A focus on the pattern of inequali-ty shows that Chile is unequal because the eliteconcentrates a large proportion of the nationalincome High concentration at the top decile ishowever accompanied by much lower inequal-itymdashlower in fact than in the United Statesmdashacross the rest of the social structure The patternof mobility closely follows the type of inequal-ity that characterizes Chile High hierarchicalbarriers to mobility especially between the topstratum and the rest of the class structure arecombined with weak horizontal barriers betweenclasses that are close in terms of SES In otherwords the Chilean case can be seen as the com-bination of two distinct regimes of both inequal-ity and mobility Income concentration at the topleads to strong mobility barriers between the topechelon and the rest of the class structure anda more even income distribution betweennonelite classes leads to significant fluidityamong them Distinction of these two compo-nents of the Chilean structure suggests thatinequality and mobility are in fact related butthat their relationship is captured only whenthe specific features of these distributive phe-nomena are considered

The focus on the pattern of inequality andmobility also provides a plausible explanationfor the lack of change in Chilean mobility afterthe market-oriented transformation Fluiditydid not decline during the market reformbecause in concert with growing economicinequality the liberalization of the economyled to a decline in nonhierarchical sector barri-ers to mobility Thus the lack of temporal trendmay reflect the additive effect of two process-es growing hierarchical barriers between the topstratum and the rest of the class structure on theone hand and a decline in sector barriers tomobility across classes at the lower end of theclass structure on the other hand The lack oftemporal mobility trend challenges Friedmanrsquos(1962171) implication that ldquocompetitive freeenterprise capitalismrdquo leads to growing mobil-ity and suggests that even if some sector barri-ers may have declined the market reform did notyield growing opportunities for advancement forthe Chilean population

This analysis has multiple implications for thecomparative study of mobility The most imme-diate conclusion is that the link between inequal-ity and mobility should be explored in terms of

the pattern exhibited by these distributive phe-nomena and that it may be completely obscuredif as done in previous studies only their aggre-gate level is considered

The Chilean analysis also suggests that under-standing the sources of international variationin mobility would benefit from two develop-ments First the inclusion of countries beyondthe industrialized core would add significantvariation in terms of class structures and insti-tutional arrangements affecting national mobil-ity patterns Second understanding of the linkbetween mobility and other national attributeswould benefit from a combination of carefulcomparison of specific class barriers acrosscountries with a weighting of these barriersaccording to the hierarchical distance betweenthe classes they separate If we care about mobil-ity it is because it is not the same to be anunskilled manual worker as it is to belong to theservice class in terms of access to the scarceresources and rewards that determine lifechances If it was the same if there were nothierarchical but only sector differences betweenthese two classes then the issue of mobilitywould not be a question of equality of oppor-tunity but at most of diverse preferences orfunctional differentiation

Therefore it is important to distinguishbetween consequential and inconsequentialmobility barriers If a barrier is located betweentwo classes that have a similar position in theSES hierarchy then this barrier is less conse-quential in terms of equality of economic oppor-tunity in the sense that movers will not seetheir life chances signif icantly alteredAccordingly this barrier should be assigned alower weight in the analysis In contrast if thebarrier is located between two classes that aredistant in the social hierarchy mobility amongthem will be highly consequential because itwill imply a significant change in the lifechances of the movers Accordingly this barri-er should be assigned a much higher weightChile is a paradigmatic example of high inci-dence of inconsequential mobility but rareinstances of consequential mobility

At the theoretical level the Chilean analysisunderscores the limitations of theories that linkinequality and mobility by focusing only onmicro-level mechanisms such as individualresources or incentives The Chilean findingssuggest that these theories should be preceded

444444mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 24: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

by a macro-level understanding of the topogra-phy of inequality and mobility within the coun-try to determine the precise distance acrossclasses in terms of the resources enjoyed byeach and consequently the differential incen-tives involved in the competition for success

Shifting the focus from the level to the pat-tern of mobility and inequality also can illumi-nate other national cases For instance the caseof the US has puzzled researchers because ofits high level of fluidity despite its status as themost unequal country in the industrialized world(Erikson and Goldthorpe 1992a381)Attempting to explain this f inding someresearchers have argued for the incentiveapproach claiming that inequality promotesindividual attempts to ldquoget aheadrdquo thus leadingto enhanced fluidity An alternative interpreta-tion however would take into account the pat-tern of American inequality and mobilityDespite all the recent concern about ldquothe richgetting richerrdquo (Mishel et al 2005 Wolff 1995)the US pattern of inequality is exactly oppo-site that of Chile The United States is unequalbecause the poor receive an extremely smallportion of the national income (Alesina andGlaeser 200447 Atkinson 1996 Table 2Smeeding and Rainwater 2002) Consistentlyexamination of mobility patterns in the UnitedStates reflects significant barriers to upwardmobility for the lower class and more fluidityin the rest of the table which might result in anoverall high level of fluidity (see Gottschalkand Danziger 1998 Tables 2 and 3 for evi-dence based on long-term income mobility andFeatherman and Hauser 1978 Table 412 for anearly analysis of class mobility)

Combining a careful analysis of class rank-ingsmdashto evaluate the consequence of differentbarriers to mobilitymdashwith a detailed examina-tion of specific barriers between classes will

permit researchers to take a new step in com-parative analysis that addresses the systematicassociation between national economic andinstitutional characteristics of different countriesand the opportunities of their citizen to alter theirlife chances

Florencia Torche is an Assistant Professor ofSociology at Queens College CUNY and ResearchAssociate at the Center for the Study of Wealth andInequality at Columbia University Her researchfocuses on the processes of inequality reproductionin the occupational educational and wealth spheresin different national contexts She currently is com-pleting a book manuscript entitled InconsequentialMobility The Chilean Case in ComparativePerspective Other projects include a comparativeanalysis of educational stratification in LatinAmerica and a comparative analysis of access tohome ownership

AAPPPPEENNDDIIXX

CCLLUUSSTTEERR--AANNAALLYYTTIICC TTEECCHHNNIIQQUUEE FFOORR

CCOOLLLLAAPPSSIINNGG CCLLAASSSSEESS IINNTTOO HHIIEERRAARRCCHHIICCAALL

SSTTRRAATTAA

The variables used in the analysis are education(year of schooling) and earnings (Chilean pesosmonth) Figure A1 plots the mean standardizedvalue of schooling and earnings across classesHierarchical strata presented in Table A1 wereobtained from cluster analysis of mean school-ing and earnings across classes19

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444455

2322-ASR 703 filename70304-torche

19 In the cluster analysis the variables used to col-lapse classes are integrated into a single function of(Euclidian) distance A K-means clustering strategywhich allows the number of clusters produced to bespecified was used (Kaufman and Rousseeuw 1990)

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 25: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

RREEFFEERREENNCCEESS

Alesina Alberto and Edward Glaeser 2004 FightingPoverty in the US and Europe Oxford UK OxfordUniversity Press

Arum Richard and Walter Muumlller eds 2004 TheReemergence of Self-Employment A ComparativeStudy of Self-Employment Dynamics and SocialInequality Princeton NJ Princeton UniversityPress

Atkinson Anthony 1996 ldquoIncome Distribution inEurope and the United Statesrdquo Oxford Review ofEconomic Policy 1215ndash28

Behrman Jere 1999 ldquoSocial Mobility Conceptsand Measurementsrdquo Chapter 4 in New MarketsNew Opportunities Economic and SocialMobility in a Changing World edited by Nancy

Birdsall and Carol Graham Washington DCBrookings Institution

Bjorklund Anders and Markus Jantti 1997ldquoIntergenerational Income Mobility in SwedenCompared to the United Statesrdquo The AmericanEconomic Review 871009ndash18

Braun Juan Matias Braun Ignacio Briones andJose Diaz 2000 Economia Chilena 1810ndash1995Estadisticas Historicas (Chilean Economy1810ndash1995 Historical Statistics) Documento deTrabajo 187 Santiago Chile Instituto deEconomia P Universidad Catolica de Chile

Breen Richard ed 2004 Social Mobility in EuropeOxford UK Oxford University Press

Breen Richard and Ruud Luijkx 2004aldquoConclusionsrdquo Chapter 15 in Social Mobility in

444466mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Table A1mdashComparison of CASMIN and Chilean Empirically-Obtained Ranking of Classes

Stratum CASMIN Chilean-Empirical

1 Upper Service Class (I+II) Service Class (I+II)

2 Middle Routine Non-manual (III) Routine Non-manual (III)Self-Employed (IVab) Self-Employed (IVab)Skilled Manual (V+VI)Farmers (IVc)

3 Lower Unskilled Manual (VIIa) Skilled Manual (V+VI)Farm Workers (VIIb) Unskilled Manual (VIIa)

Farmers (IVc) Farm Workers (VIIb)

Figure A1mdash Standardized Values of Mean Schooling and Earnings across Classes Chile 2001The way

thisappearedin proof 2was as itexportedfrom thefile This

was amanual

fix

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 26: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

Europe edited by Richard Breen Oxford UKOxford University Press

mdashmdashmdash 2004b ldquoSocial Mobility in Europe between1970 and 2000rdquo Chapter 3 in Social Mobility inEurope edited by Richard Breen Oxford UKOxford University Press

Costa-Ribeiro Carlos 2003 The BrazilianOccupational Structure PhD dissertationDepartment of Sociology Columbia UniversityNew York NY

Deininger Klaus and Lyn Squire 1996 ldquoA NewData Set Measuring Income Inequalityrdquo The WorldBank Economic Review 10565ndash91

Edwards Sebastian and Alejandra Cox-Edwards1991 Monetarism and Liberalization The ChileanExperiment Cambridge MA Ballinger

Ellworth Paul T 1945 Chile An Economy inTransition New York Macmillan

Engerman Stanley and Kenneth Sokoloff 1997ldquoFactor Endowments Institutions and DifferentialPaths of Growth among New World EconomiesrdquoPp 260ndash304 in How Latin America Fell Behindedited by Stephen Haber Stanford CA StanfordUniversity Press

Erikson Robert and John Goldthorpe 1987aldquoCommonality and Variation in Social Fluidity inIndustrial Nations Part I A Model for Evaluatingthe FJH Hypothesisrdquo European SociologicalReview 354ndash77

mdashmdashmdash 1987b ldquoCommonality and Variation inSocial Fluidity in Industrial Nations Part II TheModel of Core Social Fluidity Appliedrdquo EuropeanSociological Review 3145ndash66

mdashmdashmdash 1992a The Constant Flux Oxford UKClarendon Press

mdashmdashmdash 1992b ldquoThe CASMIN Project and theAmerican Dreamrdquo European Sociological Review8283ndash305

Erikson Robert John Goldthorpe and LuciennePortocarero 1979 ldquoIntergenerational ClassMobility in Three Western European SocietiesrdquoBritish Journal of Sociology 30415ndash41

mdashmdashmdash 1982 ldquoSocial Fluidity in IndustrialNations England France and Swedenrdquo BritishJournal of Sociology 331ndash34

Featherman David and Robert Hauser 1978Opportunity and Change New York AcademicPress

Featherman David Lancaster Jones and RobertHauser 1975 ldquoAssumptions of Social MobilityResearch in the US The Case of OccupationalStatusrdquo Social Science Research 4329ndash60

Friedman Milton 1962 Capitalism and FreedomChicago University of Chicago Press

Galbraith James 2002 ldquoA Perfect Crime Inequalityin the Age of Globalizationrdquo Daedalus 13111ndash24

Ganzeboom Harry and Paul De Graaf 1984ldquoIntergenerational Occupational Mobility in theNetherlands in 1954 and 1977 A Log-Linear

Analysisrdquo Pp 71ndash90 in Social Stratification andMobility in the Netherlands edited by Bart BakkerJaap Dronkers and H Ganzeboom AmsterdamNetherlands SISWO

Ganzeboom Harry Ruud Luijkx and DonaldTreiman 1989 ldquoIntergenerational Class Mobilityin Comparative Perspectiverdquo Pp 3ndash84 in Researchin Social Stratification and Mobility vol 8 edit-ed by Arne Kalleberg Greenwich CT JAI Press

Ganzeboom Harry and Donald Treiman 2003ldquoThree Internationally Standardised Measures forComparative Research on Occupational StatusrdquoPp 159ndash93 in Advances in Cross-NationalComparison A European Working Book forDemographic and Socio-Economic Variables edit-ed by Jurgen Hoffmeyer-Zlotnik and Christof WolfNew York Kluwer Academic Press

Ganzeboom Harry and Ruud Luijkx 2004 ldquoRecentTrends in Intergenerational Occupational ClassReproduction in the Netherlands 1970ndash99rdquoChapter 13 in Social Mobility in Europe edited byRichard Breen Oxford UK Oxford UniversityPress

Gazmuri Cristian 2000 Eduardo Frei Montalva ysu Epoca (Eduardo Frei Montalva and his Epoch)Vol II Santiago Chile Aguilar

Gerber Theodore and Michael Hout 2004ldquoTightening Up Declining Class Mobility duringRussiarsquos Market Transitionrdquo AmericanSociological Review 26677ndash700

Goldthorpe John 1980 Social Mobility and ClassStructure in Modern Britain Oxford UKClarendon Press

mdashmdashmdash 2000 ldquoOutline of a Theory of SocialMobilityrdquo Chapter 11 in On Sociology NumbersNarratives and the Integration of Research andTheory Oxford UK Oxford University Press

Goldthorpe John and Anthony Heath 1992 RevisedClass Schema 1992 JUSST Working Paper 13Nuffield College and SCPR Oxford UK

Goldthorpe John Meir Yaish and Vered Kraus1997 ldquoClass Mobility in Israeli Society AComparative Perspectiverdquo Research in SocialStratification and Mobility 153ndash28

Gomez Sergio and Jorge Echenique 1988 LaAgricultura Chilena Las dos Caras de laModernizacion (Chilean Agriculture The TwoFaces of Modernization) Santiago ChileFlacsoAgraria

Goodman Leo 1979 ldquoSimple Models for theAnalysis of Association in Cross-Classificationshaving Ordered Categoriesrdquo Journal of theAmerican Statistical Association 74537ndash52

Goodman Leo and Michael Hout 1998 ldquoStatisticalMethods and Graphical Displays for AnalyzingHow the Association between Two QualitativeVariables Differs among Countries among Groupsor over Time A Modif ied Regression-Type

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444477

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 27: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

Approachrdquo Sociological Methodology28175ndash230

Gottschalk Peter and Sheldon Danzinger 1998ldquoFamily Income Mobility How Much is Thereand Has it Changedrdquo Pp 92ndash111 in The InequalityParadox edited by James Auerbach and RichardBelous Washington DC National PolicyAssociation

Grusky David and Robert Hauser 1984ldquoComparative Social Mobility Revisited Modelsof Convergence and Divergence in 16 CountriesrdquoAmerican Sociological Review 4919ndash38

Gwynne Robert 1996 ldquoDirect Foreign Investmentand Nontraditional Export Growth in Chile TheCase of the Forestry Sectorrdquo Bulletin of LatinAmerican Research 15341ndash57

Hauser Robert 1978 ldquoA Structural Model of theMobility Tablerdquo Social Forces 56919ndash53

mdashmdashmdash 1984a ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica2787ndash110

mdashmdashmdash 1984b ldquoVertical Class Mobility inEngland France and Swedenrdquo Acta Sociologica27387ndash90

Hauser Robert John Koffel Harry Travis and PeterDickinson 1975a ldquoTemporal Change inOccupational Mobility Evidence for Men in theUnited Statesrdquo American Sociological Review40279ndash97

Hauser Robert Peter Dickinson Harry Travis andJohn Koffel 1975b ldquoStructural Changes inOccupational Mobility among Men in the UnitedStatesrdquo American Sociological Review 40585ndash98

Heath Anthony and Clive Payne 1999 TwentiethCentury Trends in Social Mobility in BritainWorking Paper 70 CREST National Centre forSocial Research and Department of SociologyUniversity of Oxford Oxford UK

Hofman Andre 2000 The Economic Developmentof Latin America in the Twentieth CenturyCheltenham UK Edward Elgar

Hopkins Terence and Immanuel Wallerstein 1982ldquoPatterns of Development in the Modern World-Systemrdquo Pp 41ndash82 in World Systems AnalysisTheory and Methodology edited by T Hopkins andI Wallerstein Beverly Hills CA Sage

Hout Michael 1983 Mobility Tables Beverly HillsCA Sage

mdashmdashmdash 1988 ldquoMore Universalism Less StructuralMobilityrdquo The American Occupational Structurein the 1980srdquo American Journal of Sociology931358ndash1400

mdashmdashmdash 2004 ldquoHow Inequality Might AffectIntergenerational Mobilityrdquo Pp 969ndash87 in SocialInequality edited by Kathryn Neckerman NewYork Russell Sage Foundation

Hout Michael and Robert Hauser 1992 ldquoSymmetryand Hierarchy in Social Mobility AMethodological Analysis of the CASMIN Model

of Class Mobilityrdquo European Sociological Review8239ndash65

Hout Michael and John Jackson 1986 ldquoDimensionsof Occupational Mobility in the Republic ofIrelandrdquo European Sociological Review 2114ndash37

ILO 1990 International Standard Classification ofOccupations ISCO-88 Geneva Switzerland ILO

INE 2002 Compendio Estadistico 2002 (StatisticalYearbook 2002) Santiago Chile INE

Infante Ricardo and Emilio Klein 1995 ldquoThe LatinAmerican Labor Market 1950ndash1990rdquo Pp 315ndash32in Latin Americarsquos Economic DevelopmentConfronting Crisis edited by James DietzBoulder CO Lynne Rienner

Inter-American Development Bank 1999 FacingUp to Inequality in Latin America 1998ndash99Report Washington DC IADB and The JohnsHopkins University Press

Ishida Hiroshi 1993 Social Mobility inContemporary Japan Stanford CA StanfordUniversity Press

Ishida Hiroshi John Goldthorpe and Robert Erikson1991 ldquoIntergenerational Class Mobility in PostwarJapanrdquo American Journal of Sociology 94952ndash92

Jonsson Jan and Colin Mills 1993 ldquoSocial Mobilityin the 1970s and 1980s A Study of Men andWomen in England and Swedenrdquo EuropeanSociological Review 9229ndash48

Kaufman Leonard and Peter Rousseeuw 1990Finding Groups in Data An Introduction to ClusterAnalysis New York Wiley and Sons

Kay Cristobal 2002 ldquoChilersquos Neoliberal AgrarianTransformation and the Peasantryrdquo Journal ofAgrarian Change 2464ndash501

Kerckhoff Alan Richard Campbell and Idee Laird1985 ldquoSocial Mobility in Great Britain and theUnited Statesrdquo American Journal of Sociology91281ndash308

Klein Emilio and Victor Tokman 2000 ldquoSocialStratification under Tension in a Globalized ErardquoCEPAL Review 727ndash29

Larranaga Osvaldo 1999 ldquoDistribucion de Ingresosy Crecimiento en Chilerdquo (Income Distribution andGrowth in Chile) Santiago Chile Serie RefEconomicas 35 MIDEPLAN

Layte Richard and Christopher Whelan 2004 ldquoClassTransformation and Trends in Social Fluidity in theRepublic of Ireland 1973ndash1994rdquo Chapter 7 inSocial Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Leon Arturo and Javier Martinez 2000 ldquoSocialStratification in Chile at the Close of the 20thCenturyrdquo Pp 283ndash308 in Chile in the Ninetiesedited by Cristian Toloza and Eugenio LaheraSantiago Chile Dolmen

Luxembourg Rosa 1951 The Accumulation ofCapital London UK Routledge and Kegan Paul

Mamalakis Markos 1976 The Growth and Structure

444488mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 28: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

of the Chilean Economy New Haven CT YaleUniversity Press

Marcel Mario and Andres Solimano 1994 ldquoTheDistribution of Income and EconomicAdjustmentrdquo Pp 217ndash55 in The Chilean EconomyPolicy Lessons and Challenges edited by BarryBosworth Rudiger Dornbusch and Raul LabanWashington DC The Brookings Institution

Marshall Gordon and Adam Swift 1999 ldquoOn theMeaning and Measurement of Inequalityrdquo ActaSociologica 42241ndash50

Marshall Gordon Adam Swift and Stephen Roberts1997 Against the Odds Social Class and SocialJustice in Industrial Societies Oxford ClarendonPress

Martinez Javier and Alvaro Diaz 1999 Chile TheGreat Transformation Washington DC TheBrookings Institution and Geneva SwitzerlandUN Research Institute for Social Development

McRoberts Hugh and Kevin Selbec 1981 ldquoTrendsin Occupational Mobility in Canada and the UnitedStates A Comparisonrdquo American SociologicalReview 46406ndash21

Meller Patricio 1991 ldquoAdjustment and Social Costsin Chile During the 1980srdquo World Development191545ndash51

mdashmdashmdash 1996 Un Siglo de Economia PoliticaChilena (One Century of Political Economy inChile) Santiago Chile Andres Bello

MIDEPLAN 2000 Pobreza Indigencia e Impactodel Gasto Social en la Calidad de Vida (PovertyIndigence and Impact of Social Spending on LifeQuality) Informe Ejecutivo N1 Encuesta Casen2000 Santiago Chile MIDEPLAN

mdashmdashmdash 2001 Socioeconomic CharacterizationSurvey (CASEN) Year 2000 Codebook andTechnical Report Santiago Chile Ministry ofPlanning Chile MIDEPLAN Chile [producer anddistributor]

Mishel Lawrence Jared Bernstein and SylviaAllegreto 2005 The State of Working America2004ndash05 Ithaca NY ILR Press

Muumlller Walter 1986 ldquoSoziale Mobilitaumlt DieBundesrepublik im internationalen Vergleichrdquo(Social Mobility The Federal Republic inComparative Perspective) Pp 339ndash54 in PolitischeWissenschaft und politische Ordnung edited by MKaase Opladen Germany Westdeutscher Verlag

Munoz Oscar 1968 Crecimiento Industrial en Chile1914ndash1965 (Industrial Growth in Chile1914ndash1965) Santiago Chile Universidad deChile

Noorderhaven Niels Roy Thurik Sander Wennekersand Andre Van Stel 2003 Self-Employment across15 European Nations The Role of DissatisfactionRotterdam Netherlands Mimeo Faculty ofEconomics Erasmus University

Park Hyunjoon 2004 ldquoIntergenerational SocialMobility among Korean Men in Comparative

Perspectiverdquo Research in Social Stratification andMobility 20227ndash53

Pisati Maurizio and Antonio Schizzerotto 2004ldquoThe Italian Mobility Regime 1985ndash97rdquo Chapter6 in Social Mobility in Europe edited by RichardBreen Oxford UK Oxford University Press

Portes Alejandro Manuel Castells and LaurenBenton 1989 ldquoThe Policy Implications ofInformalityrdquo Pp 298ndash311 in The InformalEconomy Studies in Advanced and LessDeveloped Countries Baltimore MD The JohnsHopkins University Press

Portes Alejandro and Kelly Hoffman 2003 ldquoLatinAmerican Class Structures Their Compositionand Change During the Neoliberal Erardquo Latin-American Research Review 3841ndash82

Prebisch Raul 1950 The Economic Development ofLatin America and Its Principal Problems NewYork United Nations

Raczynski Dagmar 2000 ldquoOvercoming Poverty inChilerdquo Pp 119ndash48 in Social Development in LatinAmerica The Politics of Reform edited by JosephTulchin and Allison Garland Boulder CO LynneReinner

Raftery Adrian 1995 ldquoBayesian Model Selection inSocial Researchrdquo Sociological Methodology edit-ed by Peter Marsden Washington DC TheAmerican Sociological Association

Rivera Rigoberto 1988 Los Campesinos Chilenos(The Chilean Farmers) Santiago Chile GIA

Ryder Norman B 1965 ldquoThe Cohort as a Concepton the Study of Social Changerdquo AmericanSociological Review 30843ndash61

Schkolnik Mariana 2000 Impacto de laGlobalizacion en la Estratificacion Social Chilena(Impact of Globalization on Chilean SocialStratification) Mimeo Santiago Chile ILO

Scott Christopher 1996 ldquoThe Distributive Impact ofthe New Economic Model in Chilerdquo Pp 147ndash84in The New Economic Model in Latin Americaand Its Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Smeeding Timothy and Lee Rainwater 2002Comparing Living Standards across Nations RealIncomes at the Top the Bottom and the MiddleSPRC Discussion Paper No 120 Social PolicyResearch Center University of South WalesSydney Australia

Solon Gary 2002 ldquoCross-Country Differences inIntergenerational Earnings Mobilityrdquo Journal ofEconomic Perspectives 1669ndash6

Sorensen Jesper 1992 ldquoLocating Class Cleavagesin Intergenerational Mobility Cross-NationalCommonalities and Variations in MobilityPatternsrdquo European Journal of Sociology8267ndash81

Stallings Barbara 1978 Class Conflict and

SSOOCCIIAALL MMOOBBIILLIITTYY IINN CCHHIILLEEmdashmdashndashndash444499

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411

Page 29: inequality.stanford.edu · in this distribution over time (Marshall, Swift, and Roberts 1997). As expressed by standard statistical concepts, inequality refers to the vari-ance of

Economic Development in Chile 1958ndash1973Stanford CA Stanford University Press

Stephens John 1979 The Transition from Capitalismto Socialism Urbana University of Chicago Press

Szekely Miguel and Marianne Hilgert 1999 Whatis Behind the Inequality We Measure AnInvestigation Using Latin American Data for the1990s Mimeo Washington DC IADB

Szelenyi Sonja 1998 Equality by Design The GrandExperiment in Destratification in SocialistHungary Stanford CA Stanford University Press

Tahlin Michael 2004 ldquoDo Opposites Attract HowInequality Affects Mobility in the Labor MarketrdquoResearch in Social Stratification and Mobility20255ndash82

Tawney Richard H 1965 Equality New YorkBarnes amp Noble

Thomas Jim 1996 ldquoThe New Economic Model andLabor Markets in Latin Americardquo Pp 79ndash102 inThe New Economic Model in Latin America andIts Impact on Income Distribution and Povertyedited by Victor Bulmer-Thomas New York StMartinrsquos Press

Tyree Andrea Moshe Semyonov and Robert Hodge1979 ldquoGaps and Glissandos Inequality EconomicDevelopment and Social Mobility in 24 CountriesrdquoAmerican Sociological Review 44410ndash24

Velasco Andres 1994 ldquoThe State and EconomicPolicy Chile 1952ndash92rdquo Pp 379ndash429 in TheChilean Economy Policy Lessons and Challengesedited by Barry Bosworth Rudiger Dornbuschand Raul Laban Washington DC The BrookingsInstitution

Velasquez Mario 1990 ldquoEvolucion del EmpleoPublico en Chile 1974ndash1985rdquo (Trends of PublicEmployment in Chile 1974ndash1985) Pp 113ndash33 inEmpleo Publico Frente a la Crisis Estudios sobreAmerica Latina edited by Adriana MarshallGeneva Switzerland IIEL

Williamson John 1990 ldquoWhat Washington Meansby Policy Reformrdquo Pp 7ndash20 in Latin AmericanAdjustment How Much Has Happened edited by

John Williamson Washington DC Institute forInternational Economics

Wolff Edward 1995 Top Heavy New YorkTwentieth Century Fund

Wong Raymond Sin-Kwok 1990 ldquoUnderstandingCross-National Variation in OccupationalMobilityrdquo American Sociological Review55560ndash73

mdashmdashmdash 1992 ldquoVertical and Nonvertical Effects inClass Mobility Cross-National VariationsrdquoAmerican Sociological Review 53396ndash410

mdashmdashmdash 1994 ldquoModel Selection Strategies and theUse of Association Models to Detect GroupDifferencesrdquo Sociological Methods and Research22460ndash91

World Bank 1993 The East Asian MiracleEconomic Growth and Public Policy A WorldBank Policy Research Report New York OxfordUniversity Press

mdashmdashmdash 2001 World Development Indicators 2001Washington DC The World Bank

mdashmdashmdash 2003a World Development Indicators2003 Washington DC The World Bank

mdashmdashmdash 2003b Inequality in Latin America andthe Caribbean Breaking with HistoryWashington DC The World Bank

Xie Yu 1992 ldquoThe Log-Multiplicative Layer EffectModel for Comparing Mobility Tablesrdquo AmericanSociological Review 57380ndash95

Yaish Meir 2000 lsquoOld Debates New EvidenceClass Mobility Trends in Israeli Society1973ndash1991rdquo European Sociological Review16159ndash183

mdashmdashmdash 2004 ldquoOpportunity Little Change ClassMobility in Israeli Society 1974ndash1991rdquo Chapter13 in Class Mobility in Europe edited by RichardBreen Oxford Oxford University Press

Yamaguchi Kazuo 1987 ldquoModels for ComparingMobility Tables Toward Parsimony andSubstancerdquo American Sociological Review52482ndash94

445500mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW

2322-ASR 703 filename70304-torche

Delivered by Ingenta toColumbia University (cid 66000309)

IP 127001Mon 12 Sep 2005 120411