economic growth and income equality: implications … · economic growth and income equality s87...

Economic Growth and Income Equality S87

CANADIAN PUBLIC POLICY – ANALYSE DE POLITIQUES, VOL. XXIX, SUPPLEMENT/NUMÉRO SPÉCIAL 2003

Economic Growth and IncomeEquality: Implications of aBehavioural Model of EconomicGrowth for Public PolicyMORRIS ALTMAN

Department of EconomicsUniversity of SaskatchewanSaskatoon, Saskatchewan

Nous présentons un modèle comportemental d’entreprise et de croissance économique dans lequel le niveaud’efficacité économique, le choix de la technologie et le rythme des changements techniques sont affectéspar l’organisation de l’entreprise et les variables institutionnelles. Dans ce modèle, les entreprises à hauts etbas salaires peuvent offrir des prix compétitifs, même dans les régimes de produits de marché les pluscompétitifs, grâce à l’efficacité générée par l’effet que les taux relativement élevés de rétribution destravailleurs peuvent avoir sur l’organisation de l’entreprise. Dans cette perspective, il n’est pas nécessaireque l’amélioration du bien-être matériel de ceux qui, dans la société, sont relativement moins aisés, se fasseaux dépens des mieux nantis, ni qu’il y ait une relation de dépendance entre plus d’égalité des revenus etcroissance. Nos analyses de séries de données internationales viennent étayer cette théorie.

A behavioural model of the firm and economic growth is presented whereby the level of economic efficiency,the choice of technology, and the rate of technical change, are all affected by firm organization and institutionalvariables. In this model, high- and low-wage firms can be cost competitive even in the most competitive ofproduct market regimes because of the efficiency effect that relatively high rates of labour compensationcan have through their effect on firm organization. From this perspective, improving the material well-being of the relatively less well-off in society need not be at the expense of those who are better off andthere need not be a trade-off between more income equality and growth. International datasets are analyzed,lending support to the view.

INTRODUCTION

In this paper, I elaborate upon a behavioural theoryof the firm and induced economic growth (Altman

1996, 1998, 2001a) that complements both ortho-dox and non-orthodox growth theories, such asendogenous growth theory, and present here as well

a dynamic theory of economic welfare (Altman2000) that overlaps and extends the global theoreti-cal framework contained in Pareto Optimality. Bothof these alternative theoretical frameworks informour understanding of the complex relationship be-tween economic growth and income distribution, thefocus of this paper, with significant potential public

S88 Morris Altman


policy implications. Unlike conventional economicwisdom, the analytical predictions of the modelspresented here are consistent with the stylized factsof economic life. For, unlike the conventional model,the behavioural model does not predict that conver-gence in real per capita output should be taking placeamongst nations through market forces or that thereexists a fundamental trade-off between incomeequality and levels of per capita income and percapita income growth. Needless to say, these con-ventional analytical predictions have beenchallenged in the empirical literature. Moreover,given that significant doubt has been cast on thetechnology-cum-education explanation of increas-ing wage inequality, the model presented hereattempts to address the question of changes in wageinequality and its relationship to economic growthas a function of behavioural and institutional vari-ables, which many scholars now agree are largelyresponsible for contemporary changes in wage in-equality. Income or wage inequality and changestherein need not be driven simply by technical vari-ables. In the behavioural model, changes in wageand income inequality can take place independentof technological change and changes in human capi-tal formation. Moreover, increasing inequality isshown to be consistent with, and a potential causeof, lower levels of per capita income and higher lev-els of economic inefficiency.1

To elaborate, the conventional theoretical eco-nomic wisdom is grossly inconsistent with threebasic stylized facts.

1. Convergence has not occurred (Bairoch 1978;Baumol 1986; Baumol and Wolff 1988; Boyer1996; DeLong 1988; Pritchett 1997). The tradi-tional perspective predicts that the per capita realincome of nations should converge over time ascapital flows from the richer to the poorer coun-tries and labour, from the poorer to the richercountries. Even without such mobility, conver-gence is expected to take place through theeffects of international trade, as the poorer, low-wage economies are expected to make gains

relative to the wealthier, higher wage and, there-fore, higher cost, economies. The forcesfavouring convergence are accentuated when oneintroduces the possibility of technology andknowledge transfers. However, contrary to whatis expected by the conventional economic wis-dom, market forces have not generatedconvergence in real per capita income, exceptamongst a few select, developed economies.Market forces have not achieved what many haveexpected of them, ending the significant dispar-ity of income per person amongst nations (Olson1996; Parente and Prescott 2000). Indeed, suchdisparities have only increased over historical time(Altman 1999a; Bairoch 1978; Pritchett 1997).2

2. There appears to be no so-called equity-incomeor equity growth trade-off (Aghion, Caroli andGarcía-Peñalosa 1999; Osberg 1995; Todaro1989; United Nations Conference on Trade andDevelopment 1998). Following upon the workof Simon Kuznets (1955), the conventional wis-dom maintains that income inequality withinnations can be expected to increase over thecourse of the initial stages of economic devel-opment, as per capita income enters into a periodof sustained economic growth, and to diminishas economies enter into more advanced stagesof economic development. This relationship be-tween the growth of per capita income andincome equality is referred to as the inverse-Uhypothesis. It is, therefore, stipulated that an in-verse relationship exists between equity and percapita income growth as countries develop.Moreover, economists have tended to support theview that income redistribution yielding moreincome equality damages the growth process byreducing profits and increasing production costs.This position is reinforced in a theoretical veinby the conventional welfare literature, domi-nated by the concept of Pareto Optimality, whichstipulates that any form of income redistribu-tion reduces the economic well-being of society.But the empirical literature does not support thenotion of an equity-growth or equity-income



trade-off. In a most recent comprehensive sur-vey of the empirical literature, Aghion, Caroliand García-Peñalosa write: “the picture theydraw is impressively unambiguous, since theyall suggest that greater inequality reduces therate of growth” (1999, p. 1617). Moreover, “thetraditional view in economic theory, then, is thatthere is a fundamental trade-off between pro-ductive efficiency (and/or growth) and socialjustice … Overall, the view that inequality isnecessary for accumulation and that redistribu-tion harms growth is at odds with the empiricalevidence” (ibid., p. 1620). Todaro (1989,pp. 165-66) makes a similar point: “few devel-opment economists would argue that the Kuznetsprocess of increasing then declining inequalityis inevitable. There are now enough case stud-ies and specific examples of countries likeTaiwan, South Korea, China, Costa Rica, SriLanka, Hong Kong, and others to demonstratethat higher income levels can be accompaniedby falling and not rising inequality. It all de-pends … on the nature and character of thedevelopment process.” These findings, whichcontravene the conventional economic wisdom,are elaborated upon below in an analysis of in-come and income inequality datasets.3

3. In addition, the conventional model predicts thatdecreasing wage inequality, which is a productof institutionally based (minimum wages and un-ions) income increases of the relatively lowerpercentiles of the wage distribution should re-duce growth and employment. But there is noapparent relationship between such wage com-pression and increasing production costs orincreasing unemployment (Card and Krueger1995; Freeman and Medoff 1984). These em-pirical findings are of particular importancegiven the evidence that institutional factors suchas minimum wages, unions, and government-induced low levels of unemployment serve toreduce wage and income inequality as opposedto, or in addition to, relative changes in the sup-ply and demand for a specific type of labour

(Atkinson 1997; Blau and Kahn 1996; Fortin andLemieux 1997, 2000; Freeman 1996; Galbraith1998; Howell 1999).

The behavioural model of growth discussed hereis based upon a behavioural modelling of the eco-nomic agent and the firm discussed in detailelsewhere (Altman 1996, 1998, 2001a). Moreover,the argument presented in this paper builds uponsome significant insights into the growth processmade by Adam Smith and Arthur Cecil Pigou whoargue that income redistribution to the relatively lesswell-off can be expected to induce improved eco-nomic efficiency (Altman 2000). I also borrow fromthe work of economic historian John Habbakkuk(1962) who argues that relatively higher wages have,historically, induced increased efficiency and higherrates of technical change. In addition, the basic be-havioural framework builds upon HarveyLeibenstein’s x-efficiency theory and the relatedefficiency wage literature (Akerlof and Yellen 1986;Stiglitz 1987). Using the more realistic behaviouralassumptions embedded in efficiency wage and x-efficiency theory, I argue that there are at least twosustainable paths to economic growth. At one ex-treme is the low-wage path and at the other is ahigh-wage path. The latter is associated with a higherlevel of equilibrium per capita real gross domesticproduct (GDP) and the former with a lower level.Convergence in real per capita output need not takeplace through the workings of the free market.Moreover, competitive pressures (or market forcesin general) need not result in the convergence ofwage rates. In a word, what I show here is that botha low-wage economy and a high-wage economy canbe consistent with a competitive economic regimein the long run. But a high-wage economy poten-tially yields a higher level of material well-beingfor all the members of a given economy. I furthermaintain that in the short run and, under certainconditions in the longer run, the high-wage economymay also be characterized by a higher rate of growth.

High wages serve to pressure firms to become moreefficient and innovative, whereas low wages allow

S90 Morris Altman


firms to engage in less efficient economic behaviour,all the while remaining competitive and profitable. Forthis reason, high wages are viewed as a potential boonto the process of economic growth and development,whereas low wages are viewed as a potential drag. Thisis contrary to the view typically held by both neo-clas-sical and non-conventional economists wherein highwages are considered to be an anathema to the growthprocess. In the model presented here, rational economicagents can choose between different available growthpaths. There is no one road to growth. The choice made,however, affects the long-run equilibrium level of realper capita GDP and the rate of economic growth. Inmy model, choosing the low-wage growth path wouldhave predictable negative consequences for the levelof material well-being for the majority of a region’spopulation. And to the extent that the purpose of growthis to allow for the maximization of the consumptionbasket of the typical consumer, the “preferred” path togrowth of the conventional wisdom, the low-wage path,generates inferior or suboptimal welfare outcomes.Only if the low-wage path is deemed the only path togrowth, can the low-wage path to growth and all thatit entails be considered in any way optimal or welfaremaximizing.

What I argue, therefore, is that there is no theo-retical reason to expect convergence to take place,even in the competitive long run. There is also nogood theoretical reason to predict, ex ante, that moreequity or, closely related to this, higher rates of la-bour compensation, need result in less growth and,therefore, invariably, to lower levels of per capitaincome. The market economy is, from the perspec-tive presented here, consistent with wide andpersistent divergences in per capita income as wellas with higher wage-more equitable economies thatare both high growth and competitive. In this sce-nario, there need not be any equity-income trade-off.More equity, that is a product of relatively higherreal wages, can even yield a higher and sustainablelevel of real per capita income. This does not implythat more income or wage equality causes more percapita growth. Rather, what is argued from the per-spective of the behavioural model presented here is

that more income equality that is a product of in-creasing real wages and improved workingconditions can yield higher rates of per capitagrowth. Increasing income inequality per se is nopanacea for laggard economic growth performance.

WHAT IS INEQUALITY?

The empirical and theoretical literature speaks todifferent types of inequality. The theoretical coreof this paper focuses upon wage changes and howthis might impact upon wage and income inequal-ity. The most general measure of inequality isfamily-income inequality. This particular measureis the focus of much of the empirical literature. It isimportant, therefore, to appreciate what drives fam-ily-income inequality and it is equally important tounderstand that change in wages plays a fundamen-tal role in driving changes in income inequality(Atkinson 1997; Gottschalk and Smeeding 1997).Measures of family-income inequality are based onthe construction of family-income estimates. Ofcourse, a critical component of family income is thewage that active family labour market participantsreceive as employees. However, the number of fam-ily members who are employed, hours worked bythese same family members, and investment or capi-tal income, also determines family income. Familyincome can also be estimated, pre-tax and pre-trans-fer or post-tax and post-transfer. Holding wage ratesconstant, family income can vary considerably givenvariations in employment, hours worked, investmentincome, tax rates, and transfers. For this reason,measures of income inequality can vary indepen-dently of wage rates and changes in relative wagerates, where the latter two variables are the focus ofthe theoretical thrust of this paper.

A simple measure of income inequality between twogroups, a and b, where a is the top income decile andb is the bottom income decile, can be expressed as:

INEw L H K NT

w L H K NTa / ba a a a a

b b b b b( ) =

∑ + +∑ + + , (1)



where INE is family-income inequality, w is thewage rate, L is the number of employed householdmembers, H is total annual hours worked per em-ployed household member, K is capital income, NTis transfer income net of taxes. Income inequality isgenerated, for example, when the divorce rate in-creases, reducing the number of employees (L) inthe household, thereby reducing the income in theaffected households relative to the households notsubject to divorce. In the United States, divorce hasplayed an important role in increasing income in-equality (Burtless 1999, 2001). Also, changes inhours worked can affect the extent of income in-equality (Bell and Freeman 2000; Osberg 2001).Ceteris paribus, reducing hours worked in group ayields less income inequality while increasing hoursworked increases income inequality. An important fac-tor preventing American income inequality from risingany more than it did was increasing hours worked atthe bottom end of the income distribution, whereas inGermany an important factor contributing to increas-ing income inequality was the fall in hours worked atthe bottom end of the distribution. Of course, incomeinequality can increase, holding all other variablesconstant, simply by increasing capital income at thetop end of the income distribution. Therefore, differ-ences in income inequality across countries andcross-country changes in income inequality can bedriven by variables, other than wages, which is thefocus of this paper and a large segment of the theoreti-cal literature on income inequality. Nevertheless, it iswage inequality and changes therein that tend to driveincome inequality and changes in income inequality.

SOME STYLIZED FACTS

In an effort to supplement the existing inventory ofevidence on the relationship between income andincome inequality, I present income-inequality es-timates from World Bank sources and Gottschalkand Smeeding (1997) and estimates for real grossGDP per person and per worker from the Hestonand Summers databank. As well, I construct growthrates from the output estimates. The income inequal-

ity and the output and growth estimates are cross-ranked with each other to determine the relationshipbetween income inequality and per capita and perworker output and income inequality and economicgrowth. The estimates for the Gini coefficients andthe income held by the highest and lowest quintiles ofthe population are measures of family-income inequal-ity, not wage inequality, albeit wage inequality is animportant component of this. The inequality estimatesare far from perfect and are not easily comparableacross nations and across points in time (Atkinson1997). Nevertheless, although the results they yieldmust be taken with more than a few grains of salt, theyare drawn from data that are commonly used and theypaint a rough picture of the relationship between GDP,labour productivity, and income inequality.

Table 1 presents data and estimates on GDP percapita and income distribution for the 1960–90 pe-riod, whereas Table 2 does the same for GDP perworker. One question addressed by these tables isthe cross-sectional relationship between GDP percapita and GDP per worker and income inequality.The per capita and per worker estimates are rankedin descending order while the measures of incomeinequality are in ascending order so that the mostegalitarian economy is ranked number 1. These dataare for 83 countries, the only ones for which bothoutput and income equality estimates are available.The Spearman rank correlation coefficients, whichcorrelate the rankings of the above variables, are0.46 for the Ginis and per capita GDP and 0.49 forthe Ginis and GDP per worker. In other words, in-come equality is positively related, in terms ofranking to GDP per capita and GDP per worker.Moreover, the coefficients suggest a strong relation-ship. The correlation coefficients relating GDP percapita and per worker to Ginis are both negative,–0.51 and –0.42 respectively, suggesting that thereis a strong and negative relationship between bothoutput variables and income inequality. Increasesin the output variables are negatively related to in-creases in income inequality. In other words, thesesimple statistics suggest that higher levels of percapita GDP and labour productivity tend to coincide

S92 Morris Altman


TABLE 1Income Distribution and Gross Domestic Product per Person

GDP/pop 1985 Income Distribution (1990s)International Prices

GDP/pop Rank GDP/pop Rank GDP/pop GDP/pop Rank Gini Rank Highest/ RankGrowth Growth Growth Index Lowest

1960–73 1973–90 1973–90/ 20%(%) (%) 1960–73 $1990 1990s

United States 2.92 58 1.09 55 0.37 18,054 1 40.8 46 8.9 51Canada 3.85 35 1.78 33 0.46 17,173 2 31.5 16 5.2 17Switzerland 3.05 54 0.82 64 0.27 16,505 3 33.1 25 5.8 27Luxembourg 2.72 64 1.79 32 0.66 16,280 4 27.1 9 7.3 40Norway 3.76 37 2.39 18 0.64 14,902 5 25.8 8 3.7 8Hong Kong 7.51 3 4.61 3 0.61 14,849 6 45.0 56 10.4 58Sweden 3.15 52 1.25 50 0.40 14,762 7 25.0 5 3.6 5Australia 3.06 53 1.09 56 0.35 14,445 8 35.2 29 7.0 37Germany, West 3.53 42 1.58 38 0.45 14,341 9 30.0 13 4.7 14Japan 8.51 1 2.50 14 0.29 14,331 10 24.9 3 3.4 2Finland 4.43 24 1.99 28 0.45 14,059 11 25.6 7 3.6 4Denmark 3.53 43 1.30 47 0.37 13,909 12 24.7 2 3.6 6France 4.50 22 1.43 45 0.32 13,904 13 32.7 22 5.6 24Iceland 3.98 31 2.33 19 0.59 13,362 14 n/a n/a n/a n/aBelgium 4.29 26 1.60 37 0.37 13,232 15 25.0 4 3.6 7United Kingdom 2.67 65 1.53 41 0.57 13,217 16 36.1 34 6.5 35Netherlands 3.96 32 1.24 51 0.31 13,029 17 32.6 20 5.5 23Austria 4.10 30 1.83 31 0.45 12,695 18 23.1 1 3.2 1Italy 4.68 18 1.98 29 0.42 12,488 19 27.3 10 4.2 11Singapore 8.45 2 4.38 4 0.52 11,710 20 41.0 47 8.2 48New Zealand 2.25 70 0.38 70 0.17 11,513 21 43.9 54 17.4 74Spain 6.34 10 1.55 39 0.24 9,583 22 32.5 19 5.4 22Israel 5.71 11 1.25 49 0.22 9,298 23 35.5 31 6.2 31Ireland 4.12 29 2.44 17 0.59 9,274 24 35.9 33 6.4 34Cyprus 6.63 8 2.79 13 0.42 8,368 25 n/a n/a n/a n/aTaiwan 6.69 7 4.97 2 0.74 8,063 26 31.6 17 7.9 46Trinidad and Tobago 2.41 68 0.06 76 0.02 7,764 27 n/a n/a n/a n/aPortugal 6.92 5 2.49 15 0.36 7,478 28 35.6 32 5.9 29Greece 7.26 4 1.26 48 0.17 6,768 29 32.7 23 5.4 21Korea, Rep. 6.56 9 5.74 1 0.87 6,673 30 31.6 18 10.0 56Venezuela 1.17 85 –0.94 91 – 6,055 31 48.8 66 14.4 68Mauritius –0.78 99 3.96 6 + 5,838 32 n/a n/a n/a n/aMexico 3.81 36 1.12 53 0.30 5,827 33 53.7 73 16.2 71Hungary n/a 107 1.51 42 n/a 5,357 34 30.8 14 4.5 13Malaysia 4.62 21 3.37 9 0.73 5,124 35 48.5 63 12.0 60Argentina 2.21 71 –1.09 94 – 4,706 36 47.6 61 n/a n/aUruguay 0.25 93 0.55 68 2.23 4,602 37 42.3 52 8.9 52Yugoslavia 5.10 15 1.03 59 0.20 4,548 38 n/a n/a n/a n/aChile 1.75 79 0.87 63 0.50 4,338 39 56.5 75 17.4 75Czechoslovakia 4.68 19 1.65 35 0.35 4,095 40 25.4 6 3.5 3Brazil 4.67 20 1.08 58 0.23 4,042 41 60.0 81 25.5 78Fiji 3.22 51 1.10 54 0.34 4,007 42 n/a n/a n/a n/aSeychelles 2.89 61 3.78 7 1.31 3,973 43 n/a n/a n/a n/aGabon 6.74 6 –0.26 82 – 3,958 44 n/a n/a n/a n/aSyria 3.74 38 2.06 25 0.55 3,897 45 n/a n/a n/a n/aPoland n/a 108 –0.05 78 n/a 3,820 46 32.9 24 5.3 19Turkey 3.29 48 1.99 27 0.61 3,741 47 41.5 50 8.2 49Thailand 4.39 25 3.76 8 0.86 3,580 48 41.4 49 7.6 44Costa Rica 3.39 45 0.38 71 0.11 3,499 49 47.0 60 13.0 65Iran 5.32 14 –2.51 104 – 3,392 50 n/a n/a n/a n/aSouth Africa 3.46 44 –0.23 81 – 3,248 52 59.3 79 22.3 77Jordan 4.48 23 1.69 34 0.38 2,919 53 36.4 35 5.8 28Tunisia 3.88 33 2.30 21 0.59 2,910 54 40.2 43 7.8 45Panama 4.82 17 –0.02 77 – 2,888 55 48.5 64 14.7 69Namibia 5.51 13 –1.09 95 – 2,854 56 n/a n/a n/a n/aAlgeria 1.20 84 1.55 40 1.29 2,777 57 35.3 30 6.1 30

... continued



TABLE 1(Continued)

GDP/pop 1985 Income Distribution (1990s)International Prices

GDP/pop Rank GDP/pop Rank GDP/pop GDP/pop Rank Gini Rank Highest/ RankGrowth Growth Growth Index Lowest

1960–73 1973–90 1973–90/ 20%(%) (%) 1960–73 $1990 1990s

Ecuador 3.35 47 0.99 61 0.30 2,755 58 43.7 53 9.2 53Jamaica 3.68 40 –0.52 89 – 2,545 59 36.4 36 6.3 32Congo 3.29 49 1.23 52 0.37 2,211 60 n/a n/a n/a n/aPeru 2.90 59 –1.38 100 – 2,188 61 46.2 59 11.6 59Dominican Rep. 3.70 39 0.59 67 0.16 2,166 62 48.7 65 12.5 62Morocco 4.16 28 2.12 23 0.51 2,151 63 39.5 41 7.2 39Paraguay 2.00 76 1.61 36 0.81 2,128 64 59.1 78 27.1 79Guatemala 2.17 72 –0.15 80 – 2,127 65 59.6 80 30.0 80Sri Lanka –0.04 97 2.48 16 + 2,096 66 34.4 28 5.4 20Indonesia 2.05 74 4.21 5 + 1,974 67 36.5 37 5.6 25Egypt 2.96 56 2.32 20 0.78 1,912 68 28.9 12 4.0 9El Salvador 2.46 66 –0.33 83 – 1,824 69 52.3 72 16.6 72Philippines 2.35 69 0.67 65 0.29 1,763 70 46.2 58 9.7 54Bolivia 3.53 41 –0.40 87 – 1,658 71 42.0 51 8.6 50Papua New Guinea 3.38 46 –1.37 99 – 1,425 72 50.9 71 12.6 63Pakistan 2.74 63 2.07 24 0.75 1,394 73 31.2 15 4.3 12Bangladesh –1.76 102 2.94 11 – 1,390 74 33.6 27 4.9 15Honduras 1.86 77 0.20 75 0.11 1,377 75 53.7 74 17.1 73China 2.07 73 2.81 12 1.36 1,324 76 40.3 44 7.9 47Nicaragua 3.00 55 –2.82 106 – 1,294 77 50.3 68 13.1 67India 0.20 94 2.29 22 11.53 1,264 78 37.8 38 5.7 26Cameroon 2.01 75 1.87 30 0.93 1,226 79 n/a n/a n/a n/aIvory Coast 3.23 50 –1.58 101 – 1,213 80 n/a n/a n/a n/aZimbabwe 1.52 82 –0.08 79 – 1,182 81 56.8 76 15.6 70Senegal 0.11 96 0.36 72 3.28 1,145 82 41.3 48 7.5 43Guyana 1.74 80 –2.83 107 – 1,094 83 n/a n/a n/a n/aCape Verde Island 0.96 87 3.34 10 3.48 1,058 84 n/a n/a n/a n/aNigeria 4.99 16 –0.34 84 – 995 85 50.6 70 12.7 64Lesotho 5.66 12 2.01 26 0.36 972 86 n/a n/a n/a n/aBenin 0.34 91 –1.05 93 – 920 87 n/a n/a n/a n/aKenya 1.54 81 0.60 66 0.39 911 88 44.5 55 6.3 33Ghana 0.65 89 –0.36 86 – 902 89 32.7 21 5.0 16Sierra Leone n/a 106 –1.29 97 n/a 901 90 62.9 83 57.6 82Gambia 2.86 62 –0.40 88 – 799 91 n/a n/a n/a n/aMauritania 0.66 88 –0.34 85 – 791 92 38.9 39 7.4 41Guinea 0.19 95 1.40 46 7.34 767 93 40.3 45 7.4 42Mozambique 2.94 57 –3.71 108 – 760 94 39.6 42 7.2 38Sudan n/a 105 0.34 73 n/a 757 95 n/a n/a n/a n/aRwanda 1.12 86 0.94 62 0.84 756 96 28.9 11 4.0 10Zambia 1.45 83 –2.46 103 – 689 97 49.8 67 13.0 66Guinea-Bissau 4.17 27 –1.03 92 – 689 98 n/a n/a n/a n/aMadagascar –0.83 100 –2.17 102 2.62 675 99 46 57 10.2 57Togo 3.87 34 0.31 74 0.08 641 100 n/a n/a n/a n/aCentral African Republic 0.29 92 –1.10 96 – 579 101 61.3 82 32.5 81Comoros 2.44 67 –1.30 98 – 564 102 n/a n/a n/a n/aUganda 0.45 90 –0.64 90 – 554 103 39.2 40 7.0 36Burundi –2.72 104 0.99 60 + 550 104 33.3 26 5.3 18Mali –1.77 103 1.08 57 + 531 105 50.5 69 12.2 61Malawi 1.80 78 0.38 69 0.21 519 106 n/a n/a n/a n/aBurkina Faso –1.47 101 1.47 43 + 511 107 48.2 62 10.0 55Chad –0.35 98 –2.78 105 – 399 108 n/a n/a n/a n/a

Notes: All per capita GDP growth rates are derived from real GDP estimates expressed in 1985 international prices based upon purchasingpower parity conversions of output values. The income distribution estimates are based on estimates from the World Bank (2001). Theseestimates refer to total household income for the high-income economies and largely to per capita income and consumption for the lowerincome economies. For the latter economies, these estimates are adjusted for household size.Sources: Heston and Summers (2001); World Bank (2001, ch. 12).

S94 Morris Altman


TABLE 2Income Distribution and Gross Domestic Product per Worker

GDP/Workers 1985 Income Distribution (1990s)International Prices

GDP/Workers Rank GDP/Workers Rank GDP/pop GDP/ Rank Gini Rank Highest/ RankGrowth Growth Growth Workers Index Lowest

1960–73 1973–90 1973–90/ 20%(%) (%) 1960–73 $1990 1990s

Luxembourg 3.24 51 1.38 39 0.43 37,903 1 27.1 9 7.3 40United States of

America 2.22 75 0.59 64 0.27 36,771 2 40.8 46 8.9 50Canada 2.66 63 1.08 49 0.41 34,380 3 31.5 16 5.2 17Switzerland 2.86 58 0.58 65 0.20 32,812 4 33.1 25 5.8 27Belgium 4.32 29 1.18 46 0.27 31,730 5 25.0 4 3.6 7Netherlands 3.64 43 0.65 59 0.18 31,242 6 32.6 20 5.5 23Italy 5.16 17 1.78 28 0.35 30,797 7 27.3 10 4.2 11France 4.54 24 1.12 47 0.25 30,357 8 32.7 22 5.6 24Australia 2.45 69 0.66 58 0.27 30,312 9 35.2 29 7.0 37Germany, West 4.27 31 1.00 51 0.23 29,509 10 30.0 13 4.7 14Norway 2.97 56 1.61 31 0.54 29,248 11 25.8 8 3.7 8Sweden 2.50 66 0.82 54 0.33 28,389 12 25.0 5 3.6 5Finland 3.98 38 1.69 30 0.43 27,350 13 25.6 7 3.58 4United Kingdom 2.61 64 1.25 45 0.48 26,755 14 36.1 34 6.5 35Austria 5.04 18 1.31 41 0.26 26,700 15 23.1 1 3.2 1Spain 7.02 6 1.38 38 0.20 26,364 16 32.5 19 5.4 22New Zealand 1.74 83 –0.23 78 – 25,413 17 43.9 53 17.4 73Iceland 2.79 60 1.57 32 0.56 24,978 18 n/a n/a n/a n/aDenmark 2.86 57 0.74 56 0.26 24,971 19 24.7 2 3.6 6Singapore 7.22 3 3.27 9 0.45 24,369 20 n/a n/a n/a n/aIreland 4.48 25 2.33 17 0.52 24,058 21 35.9 33 6.4 34Israel 5.48 14 0.98 52 0.18 23,780 22 35.5 31 6.2 31Hong Kong 7.17 4 3.88 5 0.54 22,827 23 45.0 55 10.4 57Japan 7.98 1 2.46 13 0.31 22,624 24 24.9 3 3.4 2Trinidad and Tobago 2.22 74 –0.58 85 – 19,880 25 n/a n/a n/a n/aTaiwan 6.27 8 4.41 2 0.70 18,409 26 31.6 17 7.9 46Cyprus 6.20 11 2.45 14 0.40 18,053 27 n/a n/a n/a n/aGreece 7.69 2 1.30 42 0.17 17,717 28 32.7 23 5.4 21Yugoslavia 5.19 16 3.68 6 0.71 17,426 30 n/a n/a n/a n/aVenezuela 1.38 86 –1.60 99 – 17,426 29 48.8 65 14.4 67Mexico 3.91 41 0.39 67 0.10 17,012 31 53.7 73 16.2 70Portugal 6.47 7 2.01 25 0.31 16,637 32 35.6 32 5.9 29Korea, Rep. 5.82 12 5.10 1 0.87 16,022 33 31.6 18 10.0 55Syria 4.58 21 2.14 23 0.47 15,871 34 n/a n/a n/a n/aArgentina 2.44 70 –0.69 90 – 13,406 35 47.6 60 n/a 82Jordan 4.96 19 1.95 27 0.39 12,634 36 36.4 35 5.8 28Malaysia 4.42 27 2.67 12 0.60 12,527 37 48.5 62 12.0 59Algeria 2.82 59 1.26 44 0.45 12,176 38 35.3 30 6.1 30Chile 1.97 79 0.23 69 0.12 11,854 39 56.5 74 17.4 74Uruguay 0.47 94 0.61 62 1.29 11,828 40 42.3 51 8.9 51Fiji 2.46 68 0.60 63 0.24 11,817 41 n/a n/a n/a n/aIran 5.55 13 –2.69 105 – 11,400 42 n/a n/a n/a n/aBrazil 4.22 32 0.72 57 0.17 11,041 43 60.0 80 25.5 77Hungary n/a 106 1.73 29 n/a 10,822 44 30.8 14 4.5 13Mauritius –1.16 102 3.32 8 + 10,198 45 n/a n/a n/a n/aColombia 3.00 55 1.08 48 0.36 10,108 46 57.1 76 20.3 75Costa Rica 3.13 54 –0.07 76 – 10,040 47 47.0 59 13.0 64South Africa 3.28 49 0.00 75 0.00 9,595 48 59.3 78 22.3 76Namibia 6.26 9 –0.59 86 – 9,528 49 n/a n/a n/a n/aSeychelles 3.21 53 4.26 3 1.33 9,137 50 n/a n/a n/a n/aEcuador 3.86 42 1.02 50 0.26 9,032 51 43.7 52 9.2 52Tunisia 4.15 35 1.36 40 0.33 8,861 52 40.2 43 7.8 45Turkey 4.30 30 2.15 22 0.50 8,632 53 41.5 49 8.2 48Gabon 7.14 5 –0.34 80 – 8,064 54 n/a n/a n/a n/aPanama 4.72 20 –0.36 81 – 7,999 55 48.5 63 14.7 68

... continued



TABLE 2(Continued)

GDP/Workers 1985 Income Distribution (1990s)International Prices

GDP/Workers Rank GDP/Workers Rank GDP/pop GDP/ Rank Gini Rank Highest/ RankGrowth Growth Growth Workers Index Lowest

1960–73 1973–90 1973–90/ 20%(%) (%) 1960–73 $1990 1990s

Czechoslovakia 4.11 36 1.53 33 0.37 7,728 56 25.4 6 3.5 3Poland n/a 105 0.10 70 n/a 7,467 57 32.9 24 5.3 19Guatemala 2.59 65 0.03 71 0.01 7,435 58 59.6 79 30.0 79Dominican Rep. 4.15 34 –0.08 77 – 6,898 59 48.7 64 12.5 61Egypt 3.34 46 2.28 20 0.68 6,889 60 28.9 12 4.0 9Peru 3.46 45 –1.70 101 – 6,847 61 46.2 58 11.6 58Morocco 4.57 23 1.39 37 0.30 6,770 62 39.5 41 7.2 39Thailand 4.47 26 3.43 7 0.77 6,754 63 41.4 48 7.6 44Paraguay 2.07 78 1.50 34 0.72 6,383 64 59.1 77 27.1 78Sri Lanka 0.08 97 2.32 18 28.00 5,742 65 34.4 28 5.4 20El Salvador 2.37 72 –0.37 82 – 5,485 66 52.3 71 16.6 71Bolivia 4.09 37 –0.24 79 – 5,315 67 42.0 50 8.6 49Jamaica 3.92 39 –1.56 98 – 5,146 68 36.4 36 6.3 32Indonesia 2.30 73 4.01 4 1.74 5,024 69 36.5 37 5.6 25Bangladesh –0.75 100 3.13 10 + 4,790 70 33.6 27 4.9 15Philippines 2.69 62 0.63 61 0.23 4,784 71 46.2 57 9.7 53Pakistan 3.27 50 1.97 26 0.60 4,639 72 31.2 15 4.3 12Congo 3.58 44 0.63 60 0.18 4,497 73 n/a n/a n/a n/aHonduras 2.39 71 0.02 74 0.01 4,464 74 53.7 72 17.1 72Nicaragua 3.33 47 –2.98 107 – 4,159 75 50.3 67 13.1 66India 0.89 90 2.38 16 2.67 3,235 76 37.8 38 5.7 26Ivory Coast 4.40 28 –0.68 89 – 3,075 77 n/a n/a n/a n/aPapua New Guinea 3.92 40 –1.01 95 – 3,020 78 50.9 70 12.6 62Guyana 1.38 87 –3.80 108 – 2,970 79 n/a n/a n/a n/aCape Verde Island 0.87 91 2.77 11 3.17 2,755 80 n/a n/a n/a n/aCameroon 2.48 67 1.41 36 0.57 2,489 81 n/a n/a n/a n/aSierra Leone n/a 108 –0.78 92 n/a 2,487 82 62.9 82 57.6 81Zimbabwe 1.95 80 –0.79 93 –0.41 2,437 83 56.8 75 15.6 69Senegal 0.28 96 0.32 68 1.15 2,398 84 41.3 47 7.5 43Sudan n/a 107 0.48 66 n/a 2,333 85 n/a n/a n/a n/aChina 2.10 77 2.12 24 1.01 2,189 86 40.3 44 7.9 47Nigeria 5.39 15 –1.07 96 – 2,082 87 50.6 69 12.7 63Lesotho 6.23 10 2.40 15 0.39 2,080 88 n/a n/a n/a n/aZambia 1.81 82 –2.30 103 – 2,061 89 49.8 66 13.0 65Benin 0.78 92 –0.64 88 – 1,903 90 n/a n/a n/a n/aGhana 1.50 85 –1.33 97 – 1,873 91 32.7 21 5.0 16Kenya 1.90 81 0.03 73 0.01 1,863 92 44.5 54 6.3 33Gambia 3.21 52 0.03 72 0.01 1,735 93 n/a n/a n/a n/aMauritania 1.30 88 –1.99 102 – 1,648 94 38.9 39 7.4 41Guinea 0.46 95 2.32 19 5.05 1,594 95 40.3 45 7.4 42Togo 4.19 33 0.76 55 0.18 1,583 96 n/a n/a n/a n/aMadagascar –0.46 99 –1.62 100 – 1,561 97 46.0 56 10.2 56Mozambique 3.30 48 –2.92 106 – 1,560 98 39.6 42 7.2 38Rwanda 1.59 84 1.28 43 0.80 1,539 99 28.9 11 4.0 10Guinea-Bissau 4.57 22 –0.53 84 – 1,484 100 n/a n/a n/a n/aComoros 2.70 61 –0.91 94 – 1,264 101 n/a n/a n/a n/aCentral African Republic 0.98 89 –0.40 83 – 1,217 102 61.3 81 32.5 80Malawi 2.20 76 0.87 53 0.39 1,217 103 n/a n/a n/a n/aChad 0.02 98 –2.45 104 – 1,148 104 n/a n/a n/a n/aUganda 0.71 93 –0.69 91 – 1,142 105 39.2 40 7.0 36Mali –1.35 103 –0.62 87 0.46 1,107 106 50.5 68 12.2 60Burundi –2.33 104 1.47 35 + 1,062 107 33.3 26 5.3 18Burkina Faso –1.16 101 2.22 21 + 1,058 108 48.2 61 10.0 54

Notes: See Table 1.Sources: Heston and Summers (2001); World Bank (2001, ch. 12).

S96 Morris Altman


with relatively lower levels of income inequality.Therefore, the cross-sectional data suggest that in-creases in these output variables do not go hand inhand with increases in income inequality.4 Some ofthese facts are illustrated in Figures 1 and 2, whichare drived from Tables 1 and 2 respectively. What isclear from these figures is that the lowest GDP percapita and per worker economies are characterizedby the most income inequality, while the highestGDP per capita and per worker economies coincidewith the relatively lowest levels of income inequal-ity. It is also clear that low levels of per capita outputare achieved over a wide range of income inequali-ties. Low income inequality is obviously not asufficient condition for high levels of per capitaoutput and labour productivity. On the other hand,a range of low Ginis coincides with a wide range ofper capita output levels. Although low income in-equality correlates well with high GDP per capitaand per workers, low income inequality is not suffi-cient to realize high measures of output. However,

low income inequality does not seem to impair theachievement of high per capita GDP and labour pro-ductivity. Although a relatively high level of incomeinequality characterizes the United States, outputmeasures not so far removed from the Americaneconomy are realized in economies that are muchmore egalitarian.

The available data are used to generate growthestimates, and these are related to the available es-timates for income inequality. Growth estimates areconstructed for two period, 1960–73 and 1973–90.The former period is often associated with proactivegovernment measures to achieve low levels of un-employment and also with relatively low levels ofincome inequality. The latter period, in contrast, isassociated with higher levels of income inequalityand with government concern over inflation(Maddison 1991). With few exceptions, growth de-clined from 1960–73 to 1973–90. This is especiallytrue for the growth in labour productivity. What is

70

60

50

0

40

5,000 10,000

Per Capita 1990 GDP

Gini

30

20

15,000

USA

20,000

Hong KongSingapore

Canada

Norway

Taiwan

Sierra Leone

FIGURE 1Per Capita GDP and Inequality (1985 international prices)



interesting about these estimates is that althoughthere has been much discussion about technicalchange driving increasing income inequality overtime, to the extent that labour productivity growthis a proxy for technical change, the rate of technicalchange appears to have diminished substantiallyover the 1960–90 time frame. The growth estimatesfor the 1973–90 period are ranked in descendingorder and Spearman rank correlation coefficients arecalculated for the ranking of the growth in GDP percapita and labour productivity and the ranking ofGinis. This is to provide some rough informationon the relationship between growth and income in-equality. The correlation coefficients are 0.51 and0.49 respectively, suggesting that there is a strongpositive relationship between growth and incomeequality. These results, for labour productivitygrowth, are illustrated in Figure 3, which is derivedfrom Table 2. The worst performing economies arethose with the highest Ginis. The strongest growth

is generated in those economies with the lowestGinis, with some important exceptions, such as In-donesia, Thailand, Taiwan, and Korea. But even inthese cases, income inequality is far less than in awide range of slow-growth economies. Moreover,the highest growth performers of these outliers, Tai-wan and Korea, are more egalitarian than Indonesiaor Thailand.

One last set of estimates is provided in Table 3and Figure 4, which is derived from Table 3. Forcountries for which Ginis are available, I calculatethe change in Ginis over the 1970s and 1980s andrelate these changes to labour productivity growthin the 1973–90 period. The Gini estimates are farfrom perfect or comparable, but be this as it may,they reveal that income inequality increased in a bigway in the United States and the United Kingdom,although 12 of the 18 economies considered expe-rienced some increase in inequality. The Spearman

FIGURE 2Labour Productivity and Inequality (1985 international prices)

70

60

50

0

40

10,000 20,000

Per Worker 1990 GDP

Gini

30

20

30,000

USA

40,000

Hong KongNew Zealand

Luxembourg

Austria

S98 Morris Altman


FIGURE 3Labour Productivity Growth and Inequality

70

60

50

–2%

40

0% 2%

Labour Productivity Growth (1973–90)

Gini

s

30

20

4% 6%

Korea

–4%

Thailand

Indonesia

Taiwan

JapanAustria

FIGURE 4Productivity Growth and Changes in Inequality

1.3

1.2

1.1

0.5%

1

1.0% 1.5%

Productivity Growth (1973–90)

Chan

ges

in G

inis

0.9

0.8

2.0% 2.5%

UK

–0.5% 0.0%

Japan

Ireland

US

Italy

Norway

Finland

Belgium

Spain

France

New Zealand

Canada

Australia Denmark

SwedenGermany

Israel

Netherlands



rank correlation coefficient for the rankings ofgrowth and changes in income inequality, 0.58, sug-gests a strong positive relationship between labourproductivity growth and either the greatest fall inincome inequality or the relatively smallest increase.A clear exception to this is Japan, which was numberone in growth but number 13 in terms of changes inincome inequality. The UK is also somewhat of anoutlier here as well. The US, which is typically thefocal point of discussion on income inequality, sitssquarely, not only as one of the poorest productiv-ity performers in our sample, but also one with thesecond greatest increase in income inequality.5 Theestimates presented here speak against the hypoth-esis that increasing technical change, proxied by

increasing labour productivity growth, serves to in-crease income inequality.6

The estimates presented here are consistent withthe view that high levels of income inequality andincreases in income inequality are not necessaryconditions for high levels of per capita output andlabour productivity. Moreover, these estimates areconsistent with the view that less income inequalityneed not yield low levels of per capita GDP and la-bour productivity or low growth rates of thesevariables. These estimates also cast doubt on theview that increasing rates of technical change havebeen responsible for increasing levels of income in-equality. The behavioural model discussed in this

TABLE 3Labour Productivity Growth and Changes in Inequality

Growth (1973–90) Rank Changes in Ginis Rank(Percentage)

Japan 2.46 1 1.09 13Ireland 2.33 2 0.87 2Italy 1.78 3 0.86 1Finland 1.69 4 0.97 4Norway 1.61 5 1.04 8Spain 1.38 6 1.00 5United Kingdom 1.25 7 1.30 18Belgium 1.18 8 1.05 10France 1.12 9 0.94 3Canada 1.08 10 1.02 7Germany, West 1.00 11 1.04 9Israel 0.98 12 1.01 6Sweden 0.82 13 1.06 12Denmark 0.74 14 1.10 14Australia 0.66 15 1.11 16Netherlands 0.65 16 1.05 11United States of America 0.59 17 1.17 17New Zealand –0.23 18 1.10 15

Notes: Labour productivity growth is estimated from output per worker estimates (Table 2). The changes in the Ginisare expressed in absolute terms.Sources: Table 2; Gottschalk and Smeeding (1997, Appendix); Deininger and Squire (2000).

S100 Morris Altman


paper provides one possible explanation for howmore equity, driven by increases in wages, is con-sistent with high levels of per capita GDP and labourproductivity and with higher rates of economicgrowth.

THE CONVENTIONAL WISDOM

The conventional wisdom assumes that the economyis x-efficient in terms of effort inputs and also interms of the choice of technology. What is meantby x-efficiency is that the economy is assumed tobe always operating along the production possibil-ity frontier. Effort per unit of input is thereforemaximized and cost-minimizing technology isadopted in this scenario, thereby maximizing out-put per person, irrespective of the system ofindustrial relations in place within the firm and re-lated rates of labour compensation (Leibenstein1966, 1978, 1979).7 It is therefore assumed that ef-fort discretion does not exist in terms of economicagents having some choice over how hard or howwell they work. On the other hand, effort discretionis a necessary condition for the existence of x-inef-ficiency. It is facilitated by the existence ofincomplete contracts (based on their transactioncosts) and different behavioural functions, especiallybetween employees or agents of the firm, and mem-bers of the firm hierarchy or its principles (Akerlofand Yellen 1986; Altman 1996; Miller 1992; Stiglitz1987). The conventional wisdom also assumes thatthe rate of technological change is determinedexogenously, outside of the causal boundaries of themodel. All told, it is assumed that there are no largebills lying around just waiting to be picked up. So-ciety is doing the best it can, given the constraintsfaced by economic agents, such as imperfect infor-mation and transaction costs. The conventionalmodel is therefore consistent with a fundamentalprinciple of contemporary economic theory whichMcCloskey has referred to as the American Ques-tion or the Axiom of Modest Greed: “The Axiom ofModest Greed involves no close calculation of ad-vantage or large willingness to take risks. The

average person sees a quarter and slides over it …he sees a $500 bill and jumps for it. The Axiom isnot controversial. All economists subscribe to it,whether or not they believe in the market … and soshould you” (1990, p. 112). This is what conven-tional economic theory predicts rational economicagents must do. If they do not, then market forceswill make them do it. The big bill hunters and gath-erers will out-compete their relatively lazycompetitors and drive them into economic purga-tory or to the production-possibility frontier (Altman1999b; Reder 1982). In other words, for firms toremain competitive, they must be x-efficient.

The conventional assumption with regard to x-efficiency in production yields important analyticalpredictions with regard to the potential economicimpact and the welfare implications of income re-distribution. These analytical predictions are, in turn,embedded in the concept of Pareto Optimality, whichserves as a basis for public policy discussions re-lated to income redistribution. Within this modellingframework, absence the existence of relative pricedistortions and, therefore, the existence of allocativeefficiency and given the assumption of x-efficiency,it follows that income redistribution is a zero-sumgame: that which goes to labour, for example, re-duces the income to employers. Pareto Optimality,the best of all possible economic worlds, is givenby the realization of the joint conditions of allocativeand x-efficiency. Given the assumption that incomeredistribution has no effect on efficiency, any exist-ing income distribution at hand is not considered tobe of economic importance since it is assumed thatone cannot and should not make interpersonal com-parison of the utility attached to a particulardistribution. It is assumed without any empiricalbacking, for example, that the marginal utility ofreal income is the same for both the poor and therich. So, income redistribution cannot, by assump-tion, yield increases in aggregate social utility orwelfare.8 On the other hand, if it is assumed thatincome redistribution negatively affects efficiency,income redistribution has the dynamic effect of re-ducing savings and therefore growth and can, thus,



be critiqued from a positive analytical perspective(Altman 2000).

In contrast to this conventional perspective onincome redistribution, the argument flowing fromthe behavioural model of economic growth is thatthere need not be an equity-efficiency trade-off in acompetitive market economy to the extent that wagespositively affect productivity and do not increaseproduction costs. Therefore, shifting from a low- toa high-wage economy can be welfare improving.This behavioural model of economic welfare, whichis interconnected with the behavioural model ofgrowth, paints a dynamic picture of economic wel-fare in contradistinction to the static frameworkprovided by Pareto Optimality. In this dynamic sce-nario, the conditions of Pareto Optimality need notbe violated, in the sense that it becomes possible toredistribute income to the relatively less well-offwithout, in the long run, reducing the economicwelfare of the relatively well-to-do. Indeed, suchredistribution might serve to induce increasing effi-ciency and technological change, thereby increasingthe economic welfare of all members of society.

From the perspective of the Edgeworth Box dia-gram, increasing the real income or non-pecuniarybenefits to workers in period t0 serves to increasethe size of the economic pie in period t1, through itsdynamic affect on economic efficiency. In Dia-gram 1, if x-efficiency is assumed and CXIHXI is theproduction-possibility frontier for cars and housing,any redistribution of income yields a reduction ineconomic welfare to at least one individual — a clearviolation of Pareto Optimality. This would be thecase, for example, if income is reallocated from in-dividual a to b yielding a change in the equilibriumdistribution of income (zero allocative inefficien-cies) from e0 to e1. However, if CXIHXI is anx-inefficient production possibility frontier, a redis-tribution of income from a to b might result in anincrease in the level of x-efficiency, shifting the fron-tier for cars and housing outward to CXEHXE. In thisscenario, income redistribution need not result inthere being any losers to the extent that the economic

An important footnote to this discussion isPigou’s analysis of market economies which arguesfor income distribution away from high-income in-dividuals to the extent that a dollar shifted in thisdirection has a higher marginal product for lower-income individuals in terms of productivity effectsfrom higher food intake, improved health, and im-proved or higher levels of human capital. Pigouargues that increasing the income of the less well-to-do serves to improve their productivity, byenhancing their capacities as workers by improvingthe nutritional levels and health of workers. In ad-dition, investing in the education and skill upgradingof labour also has the effect of increasing labourproductivity (Pigou 1952, ch. 10). But Pigou arguesthat workers do not have the means to optimally in-vest in their education. Firms do not have theincentive to optimally invest in general on-the-jobtraining of their workers. Such investment wouldyield a rate of return, in terms of increased labourproductivity that would greatly exceed the return of

DIAGRAM 1Pareto Optimality, Income Distribution and X-Efficiency

Housing

Car

s

CXE

CXI

HXEHXI

Ub2

Ua1

Ua2

Ua0

S

Ub0

Ub1

e2

e0e1

t

pie is sufficiently expanded. Moreover, income dis-tribution is here consistent in a dynamic sense withthe conditions of Pareto Optimality.

S102 Morris Altman


investing in plant and equipment. Investing in thehealth and nurture of the sick will also yield highreturns by improving the capacity of workers (ibid.,pp. 746-49). He writes: “Here there is immensescope for profitable investment. It is just when theirchildren are young, and, therefore, in many waysafford the most fruitful soil for investment, that poorfamilies find themselves in the greatest straits, and,therefore, least able to provide adequately for them”(ibid., p. 750). Education and the general trainingof workers is connected to investments in the nour-ishment, housing, and medical care of the children,if education is to build up their productive capaci-ties. A poorly fed, housed, and sickly child couldnot effectively learn (ibid., pp. 751-54). In effect,Pigou speaks to market failures with regard to theinvestment in human capital formation.

Such market failures are now an important topicof analysis in recent efforts to explain why lessincome inequality is not necessarily connected witha poorer growth record (Aghion 1998; Aghion,Caroli and García-Peñalosa 1999; Osberg 1995).This Pigouvian-type analysis in effect argues thatincome redistribution increases the size of the eco-nomic pie as government intervenes to correct forsystematic market failures. Hence, redistributingincome need not negatively impact on economicgrowth or on the level of real per capita income re-alized in the affected economies. This being said,the focus of this paper is not on the economic sig-nificance of Pigouvian-type income redistributionsin a world of market imperfections. Rather, the fo-cus here is upon changes in income inequality thatare independent and unrelated to these types ofmarket imperfections. What is of concern here ishow labour-market-related institutions can affect thedistribution of income and how this might be effi-ciency and, thereby, growth enhancing.

THE BEHAVIOURAL ALTERNATIVE

In the behavioural model of the firm upon whichthe behavioural theory of growth is based, as in the

conventional one, economic agents are assumed tobe rational, in the sense that they are calculating andforward looking. Unlike in the conventional model,however, economic agents are assumed to beheterogeneous in preferences with respect to theworking conditions and the underlying system ofindustrial relations prevailing in the firm. For sim-plicity, assume that all firms produce an identicalhomogeneous output. Also assume two groups ofeconomic agents, employees and employers, wherepreferences are homogeneous within each group, butheterogeneous between groups. Assume the objec-tive function of workers or employees is concernedwith adjusting the quality and quantity of effort in-puts per unit of time to the prevailing system ofindustrial relations. It is, therefore, assumed that ef-fort discretion exists. Effort inputs are increased aswages and/or working conditions improve. Onlyunder an ideal system of industrial relations are ef-fort inputs maximized and, therefore, x-efficiencyachieved. However, the preferences of many employ-ers, who are the pre-eminent decisionmakers withinthe firm, are for short-term economic gain, no orlittle power-sharing between workers and owners,and low-wage systems of industrial relations, all ofwhich can yield x-inefficiency in production. In theabsence of material gains to the decisionmakers, forexample, in terms of lower unit costs or increasedprofits, the realization of x-efficiency can be ob-tained only if the preferences of the decisionmakersshift toward an x-efficient consistent system of in-dustrial relation (Altman 1996, 2000, 2001b;Appelbaum and Batt 1994; Gordon 1998;Ichniowski, Kochan, Levine, Olson and Strauss1996; Levine and Tyson 1990).9 I also assume thattechnological change (whether or not a firm adoptsan available technology) is a function of, or is in-duced by, relative production costs. This relates tothe theory of induced technical change (Hayami andRuttan 1971; Hicks 1932; Ruttan 1997) and to therole of effort discretion in determining technologi-cal change (Leibenstein 1973). Therefore,technological change is not exogenously determinedas it is in the conventional wisdom, nor is it simplya function of investments in human capital or



research and development as per the new growththeories (Aghion and Howitt 1998; Fortin andHelpman 1995; Helpman 1992; Romer 1990,1994).10 This said, in a behavioural model of thefirm output is not only a function of the traditionalvariables capital, labour, and technology, but alsoof effort inputs as in the following equation:

Y f(K,L, )(t)A= ε , (2)

where Y is real output or income, A is technology,K is capital, L is labour, and ε is effort.

Given the existence of effort discretion and theassumed causal linkage between wages, workingconditions, and the overall system of industrial re-lations and effort inputs, in the behavioural model,unlike in the conventional model, unit costs neednot be affected by changes in factor prices, inclusiveof the price of labour which serves here as a proxyfor all labour-related costs. X-inefficient firms cansurvive even with perfect product markets if wagerates differ between firms and if the different wagerates and the work cultures or systems of industrialrelations, for which they may be a proxy, affect la-bour productivity by affecting the effort inputted intothe process of production. Given technology, at leastfor an array of wage rates, there may be a uniqueunit cost of production. On the flip side, there neednot be any unique wage rate or labour compensa-tion package that minimizes costs or maximizes thelevel of profits. Induced technological change onlyreinforces this argument.

The following two equations and Diagram 2 serveto illustrate this argument:

ACOC L wL

YL=

+’ (3)

where AC is average costs; OCL is the cost of or-ganizational capital per unit of labour input; w israte of labour compensation; L is labour input; andY is output. For simplicity, I assume that labour is

the only factor input and that the cost of labour iscomprised of wages and organizational capital, thoseinvestments in the work culture of the firm that posi-tively affect effort inputs (Tomer 1987). Equation 3can be rewritten as:

ACOC

Y

L

wY

L

L= +(4)

Given the assumptions of the model, a low-wageregime goes hand in hand with a low productivityregime and a high-wage regime goes hand in handwith a high productivity regime and the two wageregimes need not necessarily yield any differencesin unit costs or even in rates of return. In this case,high-wage firms can compete with low-wage firmsby becoming more productive, where output is afunction of effort per unit of labour input as well asof labour, capital, and “technical change.” In Dia-gram 2, there is a unique average cost C* that isassociated with an array of labour costs or wage andother labour costs up to W* where diminishing re-turns set in with respect to the relationship betweeneffort inputs and these labour costs. Up to W* ef-fort increases sufficiently to compensate for anyincrease in labour costs. On the other hand, effort

DIAGRAM 2Average Costs, Wages and Effort Discretion

Labour Costs

Ave

rage

Cos

t

C*

W*0

B

D

Unit Costs

Technical Change

S104 Morris Altman


diminishes sufficiently as one moves from W* to 0,sufficiently to offset any potential gains to the firmthat might accrue from lower wages, working con-ditions, and other related labour costs. At W*x-efficiency in production is achieved. Further, in-creases in labour costs can be compensated forthrough induced technological change, discussedbelow (Altman 2001a). Technological change servesto shift the unit cost curve to the right to C*BD.

This behavioural approach to the firm, it shouldbe noted, differs in important ways fromLeibenstein’s x-efficiency theory, where x-ineffi-ciency in production results in higher unitproduction costs and can be sustained only in ahighly imperfect or protected product market envi-ronment. In effect, in Leibenstein’s world,suboptimal or x-inefficient effort inputs occurlargely independently of wage rate and the workculture of the firm. In this scenario, changes in ef-fort inputs yield changes in productivity yieldchanges in unit costs as per equations 3 and 4(Leibenstein 1966, 1979). The approach presentedhere also differs from the traditional efficiency wagemodelling of the firm where there is a unique wage(the efficiency wage) that yields a unique minimumunit cost of production through its impact on effortinputs. All rational firm decisionmakers wouldchoose, in this scenario, this unique wage and there-fore all firms would be paying the efficiency wage(Akerlof and Yellen 1986; Altman 2001a, ch. 1;Stiglitz 1987). This assumes a non-linear and me-chanical relationship between wages and effortinputs that need not be obtained in the real world(Akerlof and Yellen 1990; Stiglitz 1987).

In the behavioural model, employers have nomaterial incentive to become more x-efficient if thisaccrues to them no advantage in terms of unit costs,profits, or own-income. The industrial relations lit-erature clearly finds that making workers moreefficient is a costly process. This same literaturefinds that most firms have not adopted the availableand known more x-efficient systems of industrialrelations. Why? In the behavioural model, workers

eat up the productivity gains from their increasedefficiency. If this were not the case, rational firmdecisionmakers would choose to design a more x-efficient firm, especially if subject to competitiveproduct market forces. There is also a disincentiveto become more x-efficient if this involves a loss ofpower and prestige within the firm to employers aswell as the reduction in the percentage of the firm’slabour force that consists of management. In thisscenario, an economy would be x-inefficient, evengiven the availability of relatively x-efficient alter-natives, resulting in a loss of income to society atlarge (Altman 1996, 2000, 2001b; Appelbaum andBatt 1994; Gordon 1998; Ichniowski, Kochan,Levine, Olson and Strauss 1996; Levine and Tyson1990). X-inefficiency is here a product of the ra-tional private choices made by the firm’sdecisionmakers.

In this behavioural modelling of the firm, the rela-tively x-efficient system of industrial relations, withtheir higher rates of labour compensation and over-all higher levels of working conditions, would bepreferred by workers. However, in this model, thefirm’s decisionmakers prefer the less costly, morehierarchical x-inefficient system of industrial rela-tions, as long as it is cost competitive. The firm canbe induced into adopting more efficient methods offirm organization as labour costs rise or as prefer-ences of firm decisionmakers shift toward morex-efficiency in production. Also, the legal environ-ment of a country might have the same effect(Buchele and Christainsen 1995; Gordon 1996,1998). Ceteris paribus, the supply of jobs in x-effi-cient firms is, therefore, a product of the availabilityof the low-wage x-inefficient alternatives and thepreferences of the firms’ decisionmakers. The supplyof such jobs is also a function of the macroeconomicenvironment that affects the costs of introducing themore x-efficient systems of industrial relations(Levine and Tyson 1990). Given the preferences offirm decisionmakers, there is no reason to expectthat the demand for jobs in a higher paying and lesshierarchical work environment will be met by thesupply. This argument is illustrated in Diagram 3.



AB and CD represent the supply of x-efficient jobs,the supply of which is negatively related to themacro-institutional costs of producing such jobs. ABand CD are given by the preferences of the firms’decisionmakers by pressures on the firm, such ashigher wages, to produce more x-efficient jobs.Workers demand 0D of x-efficient jobs. But givensupply curve AB, a maximum of 0B such jobs canbe generated. This is when the macro-institutionalcosts of so doing are zero. However, more x-effi-cient jobs are supplied as the supply curve shifts tothe right due to changes in the decisionmakers’ pref-erences, resulting from labour cost pressures, or tolegal constraints forcing firms to adopt a more co-operative industr ial relations organizationalframework. In this case, the maximum supply of x-efficient jobs is 0D. Moreover, given a particularmacro-institutional cost of becoming more x-effi-cient, such as 0E, the supply of x-efficient jobs canvary, from 0F to 0G, for example, depending uponthe position of the supply curve. In other words, themacro cost of providing x-efficient jobs does notcompletely constrain the supply of such jobs. In thereal world, despite the existence of the x-efficientalternative, the supply of x-efficient jobs, in the

world economy, remains in the direction of AB. Thisis to be expected, given the preferences of firmdecisionmakers and the economic viability of thex-inefficient option.

A classic example of the positive effects of a su-perior system of industrial relations on productivityis the NUMMI (New United Motor ManufacturingInc.) experiment in the United States, where Toyotaentered into a joint venture with General Motors,taking over the management of GM’s plant inFreemont, California in 1983 which GM had closedin 1982, eventually transforming what was GM’sleast productive plant into its most productive. Priorto the takeover, the Freemont plant’s labour produc-tivity was half of Toyota’s, with one of the highestdefect rates in the industry. And this was after GM’s$150 million investment prior to closing the plant,making it a state-of-the-art manufacturing facilityin terms of plant and equipment. Toyota introducedinto the Freemont plant its particular system of pro-duction, inclusive of team production, wageflattening, job security, worker input into plant man-agement, and a form of gainsharing. But little elsechanged. Toyota even hired back the militant tradeunion leaders and most of the other workers formerlyemployed by GM in the Freemont plant. Toyota en-gaged in this experiment, in spite of the fact thatAmerica’s macro-institutional environment is con-sidered to be a deterrent to adopting more efficientand cooperative systems of industrial relations. Inquick order, productivity doubled and the defect ratecollapsed in the Freemont plant causing it to becameGM’s most productive American plant, even moreso than its much more automated, technologicallyadvanced plants (Levine 1995; Levine and Tyson1990; Ichniowski, Thomas, Levine, Olson andStrauss 1996). Adopting a more x-efficient systemof work organization was obviously not contingentupon the existence of a more ideal macro-institu-tional environment. In spite of this success, theNUMMI example was not replicated in other GMplants with the exception of the Saturn venture. Thetypical GM plant remains largely within the domainof the traditional, relatively non-cooperative, and

DIAGRAM 3Institutional Costs, Preferences and Work Culture

Supply and Demand for X-efficient Jobs

Inst

itutio

nal C

osts

E

G0 B DF

A

C

S106 Morris Altman


often conflict-ridden system of industrial relations.The NUMMI alternative was an option that couldbe forgone by firm decisionmakers even in the rela-tively competitive automobile manufacturing marketgiven the economic viability of the x-inefficient al-ternatives. The cost of foregoing the NUMMIalternative is the resulting x-inefficiency.11

GROWTH MODELLING

In the traditional Solow growth model per capitagrowth cannot be sustained over time unless oneintroduces exogenous technical change, since percapita growth is otherwise a function of capital perworker, where the productivity effect of the latter issubject to diminishing returns. The production func-tion is subject to constant returns. Increasing thepropensity to save can increase growth, but this pro-pensity is subject to the binding constraint of theconsumption wants of the economic agent. Increas-ing savings can have a permanent effect on the levelof per capita income, but this effect is limited by howmuch savings can be increased by. The Solow growthnarrative is contained in the following equation:

m ns

(K

Y)

+ =, (5)

where m is the exogenously given rate of technicalchange, n is the rate of growth in the labour force, s

is the marginal propensity to save, and K

Y

is the

capital to labour ratio. The technical change coeffi-cient m becomes a permanent feature of the growthequation insofar as technical change is a permanentand persistent feature of the economy. Applied tothe domain of the determination of per capita out-put and its rate of growth, given the assumption ofdiminishing returns to factor inputs, sustainedgrowth is a product of sustained (exogenously de-termined) technical change since it is technicalchange, independent of changes in the capital to la-

bour ratio that drives sustained increases in labourproductivity. Moreover, given the assumption of di-minishing returns, differences in per capita outputmust be largely a product of different rates of tech-nical change between economies (Parente andPrescott 2000).12 More generally, the growth in realGDP is a function of the growth in factor inputs plusthe rate of technical change, and the growth in la-bour productivity becomes a function of the growthin the capital to labour ratio (which is subject todiminishing returns) and the rate of technical change.In the Solow growth model, this argument takes thefollowing form, where A(t) is the “technical change”shift parameter, K is total capital stock, and L is em-ployment. Constant returns to scale are assumed:

Y f(K,L)(t)A= (6)

It is important to note that for Solow, technicalchange refers to a bundle of variables that are re-quired to explain the growth in the economy thatcannot be explained by simple increases in tradi-tional factor inputs. The new growth literatureattempts to explain differences in growth rates andper capita output by endogenizing technologicalchange by causally relating research and develop-ment and human capital formation to technical changeor Solow’s shift parameter and by assuming imperfectproduct markets. The behavioural model of growthoffers an alternative and perhaps complementary per-spective on the causes of technical change, writ large,which is more consistent with the Solow model andrelatively competitive product markets.

In terms of the behavioural model of growth, Iassume three effects, the first two being most im-portant (Altman 1998; Altman 2001a, ch. 6): first,the x-efficiency effect whereby increasing labour costsinduce increases in the level of x-efficiency. This shiftsthe Solow production function outwards.13

Second, the technological progress effectwhereby high-wage firms are forced to adopt alreadyavailable technologies that are new to these firms.



These firms may also find it worthwhile to developnew technologies (to innovate), involving investmentin research and development. Technologicalprogress shifts the production function outwards.14

Low-wage firms need not adopt the new technol-ogy or innovate if they can effectively compete onthe basis of low wages. Moreover, the low-wage re-gime may result in labour being too x-inefficientfor the new technology to be cost-effective. For thenew technology to be adopted by all firms it mustbe cost effective in the sense of lowering unit coststo all firms, even to the low-wage firms, such thatby not adopting the new technology they will losetheir competitive edge. When new technology yieldssuch a threshold effect, the production function isshifted outwards for all firms (Rosenberg 1982). Inthe absence of a threshold effect, only the high-wagefirms adopt the new technology.15

Third, the savings effect refers to a situationwhere high-wage firms are forced to increase thepropensity to save so as to increase the rate of in-vestment under pressure to adopt new technologyor new forms of organization by higher rates of la-bour compensation. This may result in an increasein the economy’s propensity to save. To the extentthat the savings rate actually increases, the produc-tion function shifts upwards. It is important to keepin mind that to the extent that increasing wages (withworkers being characterized by a relatively low pro-pensity to save) serves to pull down theeconomy-wide or average propensity to save, theproduction function can be shifted downward. How-ever, there is no clear and evident relationshipbetween savings rates and economic growth or lev-els of per capita output. High rates of per capitaoutput can be achieved by low-savings economies.Whereas, high-savings economies are often foundamongst the world’s least-developed economies(Parente and Prescott 2000).

In the behavioural model, per capita growth isaffected by the choices economic agents make interms of effort, technology, and savings. The rela-tive cost of labour, embedded in a particular system

of industrial relations, is a key determinant of thesechoices. Where labour costs can be kept low, percapita income and growth rates can be expected tobe relatively low. Moreover, in this model conver-gence in terms of per capita output and labourproductivity is not the expected outcome by dint ofmarket forces alone. There first must be a convergenceof labour costs and work cultures. In this case, con-vergence can be either toward the low- or high-wageeconomy since both are consistent with competitive-ness in the market economy. Neither the low- norhigh-wage economy need have a unit cost advantageover the other. What the behavioural model suggestsis that Solow’s shift parameter is driven by what hap-pens inside the firm. Higher wages and improvedworking conditions serve to shift the production func-tion outward as well as inducing more capital intensityin production (Altman 2001a, ch. 6). This point isroughly contained in the following equation:

Y ( ) f(K,L)t

i

jA W

W= . (7)

Here the level of output is driven by relative ratesof labour compensation between economies (Wi/Wj), where in a simple model the output of the two(or more) economies are identical. Output increasescan also be driven by increases in wages or labourcosts in the sense that such increases would other-wise squeeze the share of income going to theemployers. Labour productivity is increased largelythrough the x-efficiency effect and induced techni-cal change. In the short run, the rate of per capitagrowth increases in this model as the economy ad-justs to the new higher level of per capita output. Inthe long run, the growth rate in output will convergetoward the growth rate of the labour force. But thethree wage effects can be expected, at a minimum,to increase long-run real per capita output and theshort-run per capita growth rate.

In this model, the focus is upon particular be-havioural variables as driving the growth processand differentials in per capita productivity and

S108 Morris Altman


income. But this in no way excludes other variablesexogenous to this model from being causally im-portant. Research and development expendituresinside and outside (public sphere) the firm andSchumpeterian process and product innovations, thatare so important to the new growth theory, can alsoplay a prominent role. However, the behaviouralmodel suggests that the wage effects, discussedabove, can play an important causal role in affect-ing these potential contributions to the growthprocess, with labour costs being the independentvariable. Moreover, the behavioural model suggeststhat labour costs and the level of x-inefficiency inthe firm affect the profitability of adopting any newinnovation.16 However, innovations that yieldthreshold levels of unit costs or profits make it costcompetitive for all firms, irrespective of factor pricesand the work cultures of their firms, to invest in suchinnovations. In time, such globally profitable tech-nologies should yield increasing productivity to alleconomies. But if this were the only game in town,there should not be the type of persistent inequali-ties in income that characterize the contemporaryworld economy (Thorbecke and Charumilind 2001).The behavioural model offers one potential expla-nation for why persistent inequalities exist and whythese inequalities can persist even under competi-tive economic regimes.17

INCOME DISTRIBUTION

The behavioural model of the firm and growth pre-sented here has implications for an understandingof the evolution of income distribution through his-torical t ime and the distribution of incomedistributions across economies at a particular pointin time. Referring back to our discussion of ParetoOptimality, the behavioural model suggests that in-creasing workers’ wages can positively affectproductivity growth and thereby the level of percapita output, this without negatively affecting theincome of the other group income claimants withinthe firm. Wherever this type of modelling scenarioapplies, first-period income redistribution that is a

product of increasing the income or in-firm benefitsto employees is not only consistent with economicgrowth, it also contributes to economic growth. It issimply a product of the x-efficiency, technologicalprogress, and savings effects, discussed above. Inthis scenario, it is possible to increase the incomeof one individual or group of individuals withoutreducing the income of another. Therefore, incomeredistribution need not be a zero-sum game and itcan be consistent with Pareto Optimality seen froma dynamic perspective.

The behavioural model predicts that because apositive causal relationship exists between labourcosts and labour and total factor productivity, it ishighly unlikely that increasing wage inequality is anecessary condition for per capita economic growth.At a minimum, maintaining low real-wage growthof the least paid workers is a highly unlikely neces-sary condition for per capita economic growth.Moreover, under reasonable assumptions, one wouldexpect that such growth is consistent with lower lev-els of wage inequality. Given that wage inequalityis a key determinant of income inequality one wouldalso expect that when wage inequality diminishesso should income inequality. But this need not al-ways hold true. For example, as already discussed,increasing divorce rates and, related to this, increas-ing atomization of families can contribute towardincreasing income inequality, holding wage inequal-ity constant. So would an increasing correlationbetween the income of husbands and wives, hold-ing wage inequality constant (Burtless 1999, 2001).Moreover, income distribution becomes more un-equal independent of changes in wage inequalitywhen employment rates and hours worked changesdifferentially in favour of the higher income groups.Thus, even if there are no changes in wage inequal-ity, income inequality increases when employmentrates and hours worked change at either the bottomor the top-end of the income distribution (Osberg2001). This being said, the behavioural model predictsthat high levels of income inequality or increasinglevels of income inequality are not necessary condi-tions to economic prosperity and growth.



In a simple behavioural model, where x-ineffi-ciency exists and technical change is induced bylabour costs, and assuming one homogenous outputand two income groups, workers and employers,wage inequality is reduced within the firm as thewage rate increases. This is illustrated in Table 4, inthe Sector One dynamics, where wage inequalitydrops from 5 to 3.4 as wages increase. If one as-sumes that minimum wage legislation and unionsdrive the income of workers, then in this scenariowage inequality is reduced through such interven-tions in the economy and increased as suchinterventions are reduced or eliminated. Bear inmind that in the behavioural model, introducing orincreasing existing minimum wages and successfulunion efforts to increase labour compensation andimprove working conditions need not increase unitcosts or reduce profits. This narrative is consistentwith evidence for the US and the UK, for example,that experienced the developed world’s most sub-stantive increases in wage inequality over the1970–80 to 1990–2000 period (Fortin and Lemieux1997; Freeman 1996; Atkinson 1997). Other vari-ables might also have affected the extent of wageinequality, but the behavioural model suggests theimportance of legislative or institutional interven-tions to this process. One would also expect that bothminimum wages and unions to positively affect thegrowth process, whereas their removal or weaken-ing should contribute to the slowing of economicgrowth. This type of scenario can also be applied todiscuss the possible effects of low unemploymentrate policy upon income inequality, a point elabo-rated upon empirically by Galbraith (1998). In thiscase, lower unemployment rates affect most signifi-cantly the wages of the lowest paid, leastwell-organized workers, thereby increasing theirwage relative to those of the higher income workersand other claimants on income. This increase inwages not only serves to reduce wage inequality,but in the behavioural model it is paid for throughthe increased productivity that is generated. Theseresults easily translate to a multi-sector model whereone sector is high wage and low inequality and theother is low wage and high inequality. In Table 4,

Panel 6, simply by increasing the wage in SectorTwo (the low-wage sector), wage inequality is re-duced and productivity is enhanced. Reducing thewage in Sector Two, on the other hand, serves toincrease inequality and diminish labour productiv-ity. Note that in these scenarios, it is not incomeredistribution that is driving changes in income dis-tribution. Rather, it is the relative improvements inworkers’ income joined with improvements in pro-ductivity that yield reductions in income inequality.

In another scenario, with more than one sector, itis clear that the evolution of inequality criticallydepends upon what occurs within sectors andchanges, if any, in the relative importance of thesector within the economy, a point made decadesago by Kuznets (1955) in his classic work on in-come inequality. Although the behavioural modelpredicts less wage and income inequality whenwages rise, this need not take place in a multi-sec-tor economy where wages are increasing in onesector while diminishing in another. Here, the re-sults are ambiguous, much depending on the extentof wage changes in the sectors and changes in therelative importance of the sectors (Table 4, Panels 1and 2). Even in a scenario where wages are increas-ing in one sector, with consequent reductions ininequality (say as a result of unionization and tightlabour markets), while inequality remains stable inthe high inequality sectors, inequality increases ifthe high inequality sectors become relatively moreimportant (Table 4, Panel 3). In this case, meanwages might be increasing in the economy as awhole, while inequality is increasing as well, as aresult of a change in the relative importance of thedifferent sectors to the economy. On the other hand,inequality diminishes as a consequence of reducingthe relative importance of the high inequality sec-tors (Table 4, Panels 4, 5, and 8). Inequality can alsoincrease as the high-wage sector leads the growthprocess. In this scenario, the wage gap between thehighly paid workers in the high-wage sector and thelow-wage workers in the low-wage sector increases.In this case, to the extent that the low-wage work-ers are x-inefficient, the wage gap could be reduced

S110 Morris Altman


TABLE 4Inequality in Artificial Economies

Income

Workers Employers Output Wage inequality Growth

Sector one

Income per unit of timet0 100 500 600 5.00t1 110 500 610 4.55 1.02t2 121 500 621 4.13 1.02t3 133 500 633 3.76 1.02t4 146 500 646 3.42 1.02

Sector two


Sector three (export price-shocked economy)


Average wage inequality (alternative employment weights)

Weights Inequality Mean Output Growth

Panel 1 Sector 1(t0) Sector 2 (t0) S1 (50) S2 (50) 5.0 600Sector 1(t4) Sector 2 (t4) S1 (50) S2 (50) 5.5 606 1.01










by introducing minimum wages, for example, whichserves to increase the wages for the lowest paidworkers, thereby keeping wage inequality from in-creasing to the extent that it would otherwise.

The behavioural model is also consistent with ascenario that typifies economies where importantoutputs are exported and the price of which can besubject to price shocks, however determined. In suchan economy real wages might be increasing andwage inequality decreasing in the domestic sector.However, in the export sector, if positive priceshocks result in the income of the employers increas-ing, this can cause an economy-wide increase inwage inequality (Table 4, Panel 7). In this case, in-creasing mean wage inequality is consistent withboth increasing wages in at least one sector and in-creasing measured income in the economy. Ofcourse, such increasing inequality need not occur ifemployees are able to capture some of the rentsachieved in the export sector as a result of positiveprice shocks.

At a more specific level, the behavioural modelalso speaks to the potential role of human capitalformation in terms of both formal education and skillformation, age structure, and technical change aspossible determinants of income inequality. If oneassumes that all economic agents are homogeneousand producing an identical product using an identi-cal technology, in a world of x-efficiency, alleconomic agents should be paid the same wage (rep-resenting all forms of labour compensation) inlong-run competitive product market equilibrium.In other words, in this scenario, the marginal pro-duct would be identical for all workers with wagesequalized across all workers. But this scenario sim-ply establishes an ideal type. In a world wherex-inefficiency prevails differentially across firms asa product of different institutional parameters facedby firms and of different work cultures or systemsof industrial relations adopted by firms, wages neednot be identical in long-run equilibrium. Higherwage firms are relatively more x-efficient and, there-fore, cost competitive, with the relatively low-wage

and x-inefficient firms. So, for example, wages canbe higher for unionized workers in what would bethe relatively more x-efficient firms than in the rela-tively x-inefficient non-unionized shops. In otherwords, otherwise identical workers would be receiv-ing different wage rates.18 This would have theeffect of causing a certain level of wage inequalitythat is unnecessary from the perspective of techni-cal constraints faced by the firm.

But even if human capital factors, age distribu-tion, and technology dictate a particular inequalityin the distribution of income in an x-efficient long-run equilibrium, the distribution of income predictedin such an x-efficient scenario need not be obtainedin a world where x-inefficiency exists. The key pointhere is that even if the technical constraints withinthe firm dictate a Gini coefficient of 0.3, a Gini of0.4 might be obtained if different firms are charac-terized by different work cultures as in where oneset of firms pays higher wages and is more x-effi-cient than another. The relatively low-wage firmsare simply paying workers less than need be giventhe existence of organizationally induced x-effi-ciency. Technical factors such as human capital andtechnology only set the parameters within whichwage or income distribution levels must fall. Butaccording to the behavioural model, whether theminimum inequality allowed for by these technicalparameters is realized depends on how the firm isorganized. And this, in turn, is affected by institu-tional parameters such as unions, minimum wages,social goods like unemployment insurance, andemployment policy.

More specifically, for example, human capitalformation need not generate increasing real incometo those with more human capital vested in themfrom the perspective of the behavioural model. Norneed the convergence in levels of human capitalyield a convergence in wage rates (Herzenberg, Alicand Wial 1998). On the other hand, a divergence inlevels of human capital need not yield a divergencein wage rates. In point of fact, convergence in for-mal education has not resulted in the convergence

S112 Morris Altman


of incomes (O’Neill 1995; Parente and Prescott2000). For one, higher levels of human capital yieldhigher wages and a higher level of productivity de-pending on the production function in which it iscontained. PhDs employed in fast food restaurants,for example, will not earn high wages. In the be-havioural model, human capital endowments, vestedin the overall production function of the firm, setthe outerbound parameters to the productivity aworker can achieve, ceteris paribus. Some workerscan produce below potential and earn below poten-tial, no matter their human capital endowment,depending upon the organizational parameters of thefirm. Such x-inefficiency can yield more incomeinequality than need be. For example, workers witha high school education or less, might earn less thanis necessary, given their potential productivity, inthe absence of minimum wage legislation (Fortinand Lemieux 1997; Freeman 1996). In a behaviouralmodel, ceteris paribus, introducing or increasingminimum wages, within limits, can serve to increasethe level of x-efficiency and induce technicalchange, allowing for the higher wages to be costcompetitive whilst also serving to reduce incomeinequality.

CONCLUSION

Building upon the insights of the x-efficiency, effi-ciency wage, and industrial relations literature, thebehavioural model of the firm and economic growthdiscussed in this paper suggests that labour marketconditions and how the firm is organized play animportant causal role in determining productivity,per capita output differentials, and in determiningthe extent of income inequality. Where labour costsand firm organization affect the efficiency of thefirm, high-wage firms can be cost competitive withlow-wage firms, union firms with non-union firmsand, more generally, high-wage economies can becosts competitive with low-wage economies. More-over, in this behavioural analytical framework,increasing income equality need not be a zero-sumgame when this involves improving the level of

material well-being of workers. For this can posi-tively affect the size of the economic pie by affectingthe firm’s overall level of efficiency. In this case, amore equal distribution of income is consistent withand a contributing factor to higher rates of growthand higher levels of per capita output. In this sce-nario, more wage or income equality need not be atsomeone else’s expense since more income equal-ity is not achieved through the process of incomeredistribution, at least in a multi-period framework.Related to this, institutional factors such as mini-mum wages, unions, and tight labour market policycan serve not only to reduce wage income inequal-ity, but also toward making an economy relativelymore efficient. Finally, from a behavioural perspec-tive, the extent of wage and income inequality isnot simply a product of technical parameters em-bodied in the traditional production function, it isalso affected by institutional variables. The produc-tion function only sets the outer boundary withinwhich an array of income distributions, all of whichare cost competitive, can be realized. This type ofmodelling of the firm and economy suggests agreater degree of freedom in the determination ofthe degree of income inequality, from the perspec-tive of public policy and economic analysis than issuggested by more traditional perspectives whereincome redistribution and, more generally, moreincome equality, is at best a zero-sum game and, atworst, a cause of economic inefficiency.

There are some direct public policy implicationsof the behavioural model for issues related to in-come distr ibution. First and foremost, thebehavioural model suggests that to the extent thatthere is an efficiency or technological change effectthat flows from improvements in the material well-being of workers, policy should be designed toencourage such improvements and, moreover, tofacilitate the realization of such an efficiency andtechnological change effect. For example, minimumwages and unions can serve to improve both thewell-being of workers and to the productivity per-formance of the firm. Additionally, from a moremacroeconomic perspective, policy that serves to



maintain low unemployment rates by improving thebargaining power of labour can have an efficiencyeffect and technology effect through the impact thatresulting improvements in wages and workingconditions might have on the firm’s level of x-efficiency and on its technology. A similar case canbe made with regard to maintaining reasonableunemployment insurance as well as social insuranceprograms, both of which affect the bargaining powerof labour. However, the behavioural model does notinfer that efficiency and technology effects automati-cally or costlessly flow from improvements inmaterial level of well-being of workers. Governmentpolicy that facilitates firms adopting and investingin improvements in industrial relations, the skillenhancement of workers, and the development andadoption of new technology or adopting existingrelatively more productive technologies, for exam-ple, facilitate firms realizing higher levels ofproductivity to, at a minimum, offset the higher costsaccruing from improvements to labour compensa-tion. None of these policies are redistributive innature. Rather, they are geared toward building amore productive economy in a manner where work-ers, inclusive of those at the lower deciles of theincome distribution, capture a significant share ofan increasing economic pie. As a consequence ofthe dynamic relationship, specified by the behav-ioural model, between labour compensation andeconomic efficiency and growth, facilitated by gov-ernment policy, the distribution of income canbecome more equal over time. But it is not moreincome equality per se that generates more effi-ciency and growth. In the behavioural modeldiscussed here, causality runs from efficiency andgrowth to more income equality, where increasedincome equality is a likely outcome of a more pro-ductive economy built upon improvements in thematerial well-being of workers.

NOTES

This paper was written for the IRPP-CSLS Project onEconomic Growth and Inequality Conference, Ottawa,

26–27 January 2001, while the author was Visiting Scholarwith the Department of Economics at Cornell Universityand with the Centre of North American Studies at DukeUniversity. It has benefited greatly from comments andsuggestions from Louise Lamontagne, Pierre Fortin,Andrew Sharpe, my discussant Lorne Carmichael, as wellas from other participants in the Ottawa conference andthe project’s preconference, held in Montreal, 2–4 No-vember 2000. Many thanks as well to my researchassistant, Haul Kamsari.

1See Aghion and Howitt (1998) and Foley and Michl(1999) for comprehensive surveys of the growth litera-ture. See Temple (1999) for a survey of recent empiricalcontributions to the literature. For a survey of the litera-ture and an extension of the argument presented in thispaper, see Altman (2003).

2Parente and Prescott (2000), however, have arguedthat convergence has failed to take place because institu-tional parameters have blocked market forces fromsuccessfully generating convergence amongst nations. Onthis point, see also Olson (1996).

3For a detailed discussion on the possible negative re-lationships between income inequality and growth andvice versa, see Thorbecke and Charumilind (2001).

4Using the less comprehensive measure of income in-equality, the quintile to quintile measure, yields similarresults.

5Burtless (2001) argues, using more recent data, thatincreasing US inequality joined with its superior recenteconomic performance provides strong evidence support-ing the hypothesis that increasing income inequality is anecessary condition for improvement in economic per-formance. However, Burtless recognizes that much of theUS’s growth success has little to do with a strong pro-ductivity performance and much to do with increasingrates of employment. Moreover, he pays little attentionto the different macroeconomic policies between the USand its developed economy competitors, which appearsto have played an important role in generating differen-tial employment rates between the United States andrelated economies. Moreover, Burtless assumes that in-creasing divorce rates, an important determinant of risingAmerican income inequality is a necessary by-product ofgrowth. This ignores the social and institutional determi-nants of divorce, independent of any growth-relateddeterminants.

S114 Morris Altman


6See Galbraith (1998) for a critical discussion of thishypothesis, which has been used to explain the contem-porary increase in American wage inequality.

7For example, Leibenstein argues that in a world ofeffort discretion x-efficiency in production requires firmmembers making effort choices involving, “cooperationwith peers, superiors, and subordinates, in such a way asto maximize their contribution to output” (1978, p. 206).The work environment within the firm, in this scenario,plays a fundamental causal role in affecting the level ofx-efficiency achieved by the firm.

8This modern perspective is in stark contrast to theconventional wisdom prior to World War II, which is wellarticulated by Pigou: “it is evident that any transferenceof income from a relatively rich man to a relatively poorman of similar temperament, since this enables more in-tense wants to be satisfied at the expense of less intensewants, must increase the aggregate sum of satisfaction.The old law of ‘diminishing utility’ thus leads securelyto the proposition: Any cause which increases the abso-lute share of real income in the hands of the poor, providedthat it does not lead to a contraction in the size of thenational dividend from any point of view, will, in gen-eral, increase economic welfare” (1952, p. 89). See alsoCooter and Rappoport (1984).

9This literature finds that more and higher quality ef-fort inputs are associated with a bundle of characteristicsthat include more active employee participation in thefunctioning of the firm, more labour-management coop-eration associated with a less hierarchical system ofmanagement, labour compensation related to labour pro-ductivity, and a relatively long-term employmentrelationship with the firm. On the latter point, see alsoCarmichael and MacLeod (2000).

10The conventional wisdom assumes that changes inrelative factor prices affect the relative efficient combi-nation of factor inputs, such as capital and labour. Thus,increasing the relative price of labour is expected to in-crease the capital to labour ratio, modelled as movementsalong the production isoquant, thereby making labour pro-ductive on the margin. This most efficient use of factorinputs, however, yields higher unit production costsceteris paribus. Changes in factor prices are not expectedto affect the position of the production isoquant, how-ever. In contrast, I argue that changes in relative factorscan also be expected to shift the isoquant. Increasing the

relative price of labour, for example, can be expected toalso shift inwards to the production isoquant, thus eitherkeeping unit costs from rising or reducing unit costs. Thisinduced technical change serves to increase total factorproductivity as well as labour productivity (Altman2001a, chs. 2 and 6).

11See Gordon (1996) and Smith (1995), for furtherexamples of relatively x-efficient forms of organizationthat remain outside the mainstream of industrial organi-zation.

12Per capita output (Y/P) is given by output per worker(Y/L) and the ratio of employment to the population

(L/P): Y

P(Y

L)* (

L

P)= . Assuming a constant employ-

ment rate, in the Solow model, per capita output isdetermined by the level of labour productivity which, inturn, is largely a product of technical change.

13This is consistent with Adam Smith’s world view.Smith argues:

The liberal reward for labour, as it encourages thepropagation, so it increases the industry of the com-mon people. The wages of labour are theencouragement of industry, which, like every otherhuman quality, improves in proportion to the encour-agement it receives. A plentiful subsistence increasesthe bodily strength of the labourer, and comfortablehope of bettering his condition, and of ending his daysperhaps in ease and plenty, animates him to exert thatstrength to the utmost. Where wages are high, accord-ingly, we shall always find the workmen more active,diligent, and expeditious, than where they are low(1937, p. 81).

14It important to note that where output is assumed tobe a function of the quantity of factor inputs, the shift inthe production function is here assumed to be greater thanwhat one would expect from an increase in the capital tolabour ratio which, of course, is subject to diminishingreturns. The shift parameter here is also a function of in-duced technological change which augments total factorproductivity. It is the latter type of shift, be it induced orexogenously determined, which is of critical analyticalimportance to Solow.

15See Altman (2001a, ch. 6), for a detailed discussionof a behavioural model of technical change. As would beexpected, wage-driven technical change induces increases



in the capital-labour ratio as well as inward shifts of thefirm’s production isoquant.

16Nelson (1998) argues that the new growth theory,however important it is, provides little insight into thosefactors underlying the immediate sources of growth. Heargues that one must investigate the inner workings ofthe firm, institutional factors in general, as well as lawand custom, to arrive at a determination of the underlyingcauses of economic growth. The behavioural model pre-sented here is advanced as one possible step in this direction.

17One alternative explanation for persistent inequali-ties revolves around the assumed protection afforded tothe inefficient firms (Parente and Prescott 2000). Fromthis perspective, firms can avoid using the most produc-tive technologies because such an inefficient decisionneed not meet the test of the competitive market. Thispoint of view actually is one that Leibenstein (1966, 1973,1978, 1978; Frantz 1997) articulated in his formulationof x-efficiency theory. For Leibenstein in a perfectly com-petitive product market regime all firms would bex-efficient and all firms would be using the most effi-cient technologies. The behavioural model presented heresuggests that, although protection can hinder the realiza-tion of efficiencies, assuming that the decisionmakers arewilling to knowingly forsake big bills lying on thesidewalk, other variables can also be of considerable im-portance. And this would be true even in the mostcompetitive of product market regimes.

18This point is elaborated upon in Altman (2001a);Krueger and Summers (1988); and Thaler (1992).

REFERENCES

Aghion, P. 1998. “Inequality and Economic Growth,” inGrowth, Inequality and Globalization, ed. P. Aghionand J.C. Williamson. Cambridge: Cambridge Univer-sity Press, pp. 7-102.

Aghion, P. and P. Howitt. 1998. Endogeneous GrowthTheory. Cambridge, MA: MIT Press.

Aghion, P., E. Caroli and C. García-Peñalosa. 1999. “In-equality and Economic Growth: The Perspective of theNew Growth Theories,” Journal of Economic Litera-ture 37(4):1615-60.

Akerlof, G.A. and J.L. Yellen., eds. 1986. Efficiency WageModels of the Labour Market. Cambridge and NewYork: Cambridge University Press.

______ 1990. “The Fair Wage Hypothesis and Unemploy-ment,” Quarterly Journal of Economics 105(2):255-83.

Altman, M. 1996. Human Agency and Material Welfare:Revisions in Microeconomics and their Implicationsfor Public Policy. Boston, Dordrecht, and London:Kluwer Academic Publishers.

______ 1998. “A High Wage Path to Economic Growthand Development,” Challenge: The Magazine of Eco-nomic Affairs 41(1):91-104.

______ 1999a. “Social Capacity and Convergence,” inEncyclopedia of Political Economy, ed. P. O’Hara.London: Routledge, pp. 1033-35.

______ 1999b. “The Methodology of Economics and theSurvivor Principle Revisited and Revised: Some Welfareand Public Policy Implications of Modeling the EconomicAgent,” Review of Social Economics 57(4):427-29.

______ 2000. “A Behavioural Theory of Economic Wel-fare and Economic Justice: A Smithian Alternative toPareto Optimality,” International Journal of SocialEconomics 27(11/12):1098-1131.

______ 2001a. Worker Satisfaction and Economic Per-formance: The Microfoundations of Economic Successand Failure. Armonk, NY: M.E. Sharpe.

______ 2001b. “Economic Theory, Public Policy and theChallenge of Innovative Work Practices,” Economicand Industrial Democracy: An International Journal23(2):271-90.

______ 2003. Economic Growth and the High WageEconomy: Choices, Constraints and Opportunities inthe Market Economy . London and New York:Routledge (forthcoming).

Appelbaum, E. and R. Batt. 1994. The New AmericanWorkplace: Transforming Work Systems in the UnitedStates. Ithaca, NY: Cornell University Press.

Atkinson, A.B. 1997. “Bringing Income Distribution infrom the Cold,” Economic Journal 107(441):297-321.

Bairoch, P. 1978. “The Main Trends in National EconomicDisparities since the Industrial Revolution,” in Dis-parities in Economic Development Since the IndustrialRevolution, ed. P. Bairoch and M. Lévy-Leboyer. NewYork: St. Martin’s Press, pp. 3-17.

Baumol, W.J. 1986. “Productivity Growth, Convergence,and Welfare: What the Long-Run Data Show,” Ameri-can Economic Review 76(5):1072-85.

Baumol, W.J. and E.N. Wolff. 1988. “ProductivityGrowth, Convergence, and Welfare: Reply.” AmericanEconomic Review 78(5):1155-59.

S116 Morris Altman


Bell, L. and R. Freeman. 2000. “The Incentive for Work-ing Hard: Explaining Hours Worked Differences in theU.S. and Germany,” NBER Working Paper No. 8051.Cambridge, MA: National Bureau of EconomicResearch.

Blau, F.D. and L.M. Kahn. 1996. “International differ-ences in Male Wage Inequality: Institutions versusMarket Forces,” Journal of Political Economy104(4):791-837.

Boyer, R. 1996. “The Convergence Hypothesis Revisited:Globalization but Still the Century of Nations?” inNational Diversity and Global Capitalism, ed. S.Berger and R. Dore. Ithaca, NY and London: CornellUniversity Press.

Buchele, R. and J. Christainsen. 1995. “Worker RightsPromote Productivity Growth,” Challenge: The Maga-zine of Economic Affairs 38(5):32-37.

Burtless, G. 1999. “Effects of Growing Wage Disparitiesand Changing Family Composition on the U.S. IncomeDistribution,” European Economic Review 43(4/6):853-65.

______ 2001. “Has Widening Inequality Promoted orRetarded Growth?” Paper presented at the LinkagesBetween Economic Growth and Inequality Confer-ence, Institute for Research on Public Policy/Centrefor the Study of Living Standards, Ottawa, 26–27January.

Card, D. and A.B. Krueger. 1995. Myth and Measurement:The New Economics of the Minimum Wage. Princeton,NJ: Princeton University Press.

Carmichael, L.H. and W.B. MacLeod. 2000. “WorkerCooperation and the Ratchet Effect,” Journal of La-bour Economics 18(1):1-19.

Cooter, R. and P. Rappoport. 1984. “Were the OrdinalistsWrong about Welfare Economics?” Journal of Eco-nomic Literature 22(2):507-30.

DeLong, B.J. 1988. “Productivity Growth, Convergence,and Welfare: Comment,” American Economic Review78(5):1138-54.

Deininger, K. and L. Squire. 2000. “Measuring IncomeInequality: a New Database,” Discussion Paper, WorldBank Group. At website <http://www.worldbank.org/research/growth/dddeisqu.htm>.

Foley, D. and T.R. Michl. 1999. Growth and Distribu-tion. Cambridge, MA and London: Harvard UniversityPress.

Fortin, N.M. and T. Lemieux. 1997. “InstitutionalChanges and Rising Wage Inequality: Is There a Link-age?” Journal of Economic Perspectives 11(2):75-96.

______ 2000. “Income Redistribution in Canada: Mini-mum Wages versus Other Policy Instruments,” inAdapting Public Policy to a Labour Market in Transi-tion, ed. W.C. Riddell and F. St. Hilaire. Montreal:Institute for Research on Public Policy, pp. 211-41.

Fortin, P. and E. Helpman 1995. “Endogenous Innova-tion and Growth: Implications for Canada,” OccasionalPaper No. 10. Ottawa: Industry Canada.

Frantz, R. 1997. X-Efficiency: Theory, Evidence and Ap-plications , 2d ed. Boston: Kluwer AcademicPublishers.

Freeman, R.B. 1996. “Labour Market Institutions andEarnings Inequality,” New England Economic Review(June/July):156-67.

Freeman, R.B. and J.L. Medoff. 1984. What Do UnionsDo? New York: Basic Books.

Galbraith, J.K. 1998. Created Unequal: The Crisis inAmerican Pay. Chicago: University of Chicago Press.

Gordon, D.M. 1996. Fat and Mean: The CorporateSqueeze of Working Americans and the Myth of Mana-gerial ‘Downsizing.’ New York: Free Press.

______ 1998. “Conflict and Cooperation: An EmpiricalGlimpse of the Imperatives of Efficiency and Redis-tribution,” in Recasting Egalitarianism: New Rules forCommunities, States and Markets, ed. S. Bowles andH. Gintis. London: Verso, pp. 181-207.

Gottschalk, P. and T.M. Smeeding. 1997. “Cross-NationalComparisons of Earnings and Income Inequality,”Journal of Economic Literature 35(2):633-87.

Habbakkuk, H.J. 1962. American and British Technologyin the Nineteenth Century: The Search for Labour-Saving Inventions. Cambridge: Cambridge UniversityPress.

Hayami, Y. and V.W. Ruttan. 1971. Agricultural Devel-opment: An International Perspective. Baltimore, MD:Johns Hopkins University Press.

Helpman, E. 1992. “Endogenous Macroeconomic GrowthTheory,” European Economic Review 36(2/3):237-67.

Herzenberg, S.A., J.A. Alic and H. Wial. 1998. New Rulesfor a New Economy: Employment and Opportunity inPostindustrial America. Ithaca, NY: Cornell Univer-sity Press.

Heston, A. and R. Summers. 2001. “The Penn World Ta-bles (Mark 5.6).” At <http://datacentre.chass.utoronto.ca:5680/pwt/index.html>.

Hicks, J. 1932. The Theory of Wages. London: Macmillan.Howell, D.R. 1999. “Theory-Driven Facts and the Growth

in Earnings Inequality,” Review of Radical PoliticalEconomics 31(1):54-86.



Ichniowski, C., T.A. Kochan, D. Levine, C. Olson and G.Strauss. 1996. “What Works at Work: Overview andAssessment,” Industrial Relations 35(3):299-333.

Krueger, A.B. and L.H. Summers. 1988. “EfficiencyWages and the Inter-Industry Wage Structure,”Econometrica 56(2):259-93.

Kuznets, S. 1955. “Economic Growth and Income In-equality,” American Economic Review 45(1):1-28.

Leibenstein, H. 1966. “Allocative Efficiency vs. X-Effi-ciency,” American Economic Review 56(3):392-415.

______ 1973. “Notes on X-Efficiency and TechnicalChange,” in Micro-Aspects of Development, ed. E.B.Ayal. New York: Praeger, pp. 18-38.

______ 1978. “X-Inefficiency Xists—Reply to anXorcist,” American Economic Review 68(1):203-11.

______ 1979. “A Branch of Economics is Missing: Mi-cro-Micro Theory,” Journal of Economic Literature17(2):477-502.

Levine, D.I. 1995. Reinventing the Workplace: How Busi-ness and Employees Can Both Win. Washington, DC:The Brookings Institute.

Levine, D.I. and L. D’Andrea Tyson. 1990. “Participa-tion, Productivity, and the Firm’s Environment,” inPaying for Productivity: A Look at the Evidence, ed.A.S. Blinder. Washington, DC: The Brookings Insti-tute, pp. 183-237.

Maddison, A. 1991. Dynamic Forces in Capitalist Devel-opment: A Long-Run Comparative View. Oxford andNew York: Oxford University Press.

McCloskey, D.N. 1990. If You’re So Smart: The Narra-tive of Economic Expertise . Chicago: ChicagoUniversity Press.

Miller, G.J. 1992. Managerial Dilemmas: The PoliticalEconomy of Hierarchy. New York: Cambridge Univer-sity Press.

Nelson, R.R. 1998. “The Agenda for Growth Theory: ADifferent Point of View (Shortcomings of Neoclassi-cal Economics in Explaining),” Cambridge Journal ofEconomics 22(4):497-520.

Olson, M. 1996. “Big Bills Left on the Sidewalk: WhySome Nations are Rich, and Others Poor,” Journal ofEconomic Perspectives 10(2):3-24.

O’Neill, D. 1995. “Education and Income Growth: Im-plications for Cross-Country Inequality,” Journal ofPolitical Economy 103(6):1289-301.

Osberg, L. 1995. “The Equity/Efficiency Trade-off in Ret-rospect,” Canadian Business Economics 3(3): 5-19.

______ 2001. “Labour Supply and Inequality Trends inthe United States and Elsewhere.” Paper presented at

the Linkages Between Economic Growth and Inequal-ity Conference, Institute for Research on Public Policy/Centre for the Study of Living Standards, Ottawa, 26–27 January.

Parente, L.and E.C. Prescott. 2000. Barriers to Riches.Cambridge, MA and London: MIT Press.

Pigou, A.C. 1952. The Economics of Welfare. London:Macmillan.

Pritchett, L. 1997. “Divergence, Big Time,” Journal ofEconomic Perspectives 11(3):3-17.

Reder, M.W. 1982. “Chicago Economics: Permanence andChange,” Journal of Economic Literature 20(1):1-38.

Romer P.M. 1990. “Endogenous Technological Change,”Journal of Political Economy 98(5):S71-S102.

______ 1994. “The Origins of Endogenous Growth,” Jour-nal of Economic Perspectives 8(1):3-22.

Rosenberg, N. 1982. “The Historiography of TechnicalProgress,” in Inside the Black Box: Technology andEconomics, ed. N. Rosenberg. Cambridge: CambridgeUniversity Press, pp. 3-33.

Ruttan, V.W. 1997. “Induced Innovation, EvolutionaryTheory and Path Dependence: Sources of TechnicalChange,” Economic Journal 107(444):1520-29.

Schumpeter, J.A. 1974. The Theory of Economic Devel-opment: An Inquiry into Profits, Capital, Credit,Interest, and the Business Cycle, 1st published, 1934.London: Oxford University Press.

Smith, A. 1937. An Inquiry into the Nature and Causesof the Wealth of Nations. New York: Modern Library.

Smith, H. 1995. Rethinking America. New York: RandomHouse.

Solow, R.M. 1957. “Technical Change and the AggregateProduction Function,” Review of Economics and Sta-tistics 39(3):312-20.

______ 1994. “Perspectives on Growth Theory,” Journalof Economic Perspectives 8(1):45-54.

Stiglitz, J.E. 1987. “The Causes and Consequences of theDependence of Quality on Price,” Journal of EconomicLiterature 25(1):1-48.

Temple J. 1999. “The New Growth Evidence,” Journal ofEconomic Literature 37(1):112-56.

Thaler, R.H. 1992. “Interindustry Wage Differentials,” in TheWinner’s Curse: Paradox’s and Anomalies of EconomicLife, ed. R.H. Thaler. New York: Free Press, pp. 36-49.

Thorbecke, E. and C. Charumilind. 2001. “Economic In-equality and its Interrelationship with Growth, PoliticalConflict, Health, Education, and Crime.” Paper pre-sented to the Cornell University Social ScienceSeminar, Ithaca, NY.

S118 Morris Altman


Todaro, M.P. 1989. Economic Development in the ThirdWorld, 4th ed. New York and London: Longman.

Tomer, J. 1987. Organizational Capital: The Path to HigherProductivity and Well-Being. New York: Praeger Publishers.

United Nations Conference on Trade and Development.1998. “Income Distribution, Capital Accumulation,and Growth,” Challenge: The Magazine of EconomicAffairs 4 (2):61-80.

World Bank. 2001. World Development Report 2000/2001.New York: Oxford University Press.

economic growth and income equality: implications … · economic growth and income equality s87...

Documents