8/2/2019 2946642

1/25

Globalization and the Inequality of NationsAuthor(s): Paul Krugman and Anthony J. VenablesSource: The Quarterly Journal of Economics, Vol. 110, No. 4 (Nov., 1995), pp. 857-880Published by: The MIT PressStable URL: http://www.jstor.org/stable/2946642

Accessed: 09/11/2010 12:35

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at

http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless

you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you

may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at

http://www.jstor.org/action/showPublisher?publisherCode=mitpress.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed

page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

of scholarship. For more information about JSTOR, please contact support@jstor.org.

The MIT Press is collaborating with JSTOR to digitize, preserve and extend access to The Quarterly Journal of

Economics.

http://www.jstor.org

http://www.jstor.org/action/showPublisher?publisherCode=mitpresshttp://www.jstor.org/stable/2946642?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/action/showPublisher?publisherCode=mitpresshttp://www.jstor.org/action/showPublisher?publisherCode=mitpresshttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/2946642?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=mitpress

8/2/2019 2946642

2/25

THEQUARTERLYOURNAL

O F ECONOMICSVol. CX November 1995 Issue 4

GLOBALIZATIONAND THE INEQUALITYOF NATIONS*PAUL KRUGMAN AND ANTHONY J. VENABLES

A monopolistically competitive manufacturing sector produces goods used forfinal consumption and as intermediates. Intermediate usage creates cost anddemand linkages between firms and a tendency for manufacturing agglomeration.How does globalization affect the location of manufacturing and gains from trade?At high transport costs all countries have some manufacturing, but when transportcosts fall below a critical value, a core-periphery spontaneously forms, and nationsthat find themselves in the periphery suffer a decline in real income. At still lowertransport costs there is convergence of real incomes, in which peripheral nationsgain and core nations may lose.

In recent years there has been growingconcernamong manyobservers in the advancednations over the impact of globalizationon their ability to sustain high living standards. As growth hassurgedin developingcountries such as China,these observersfearthat Third Worldgrowth-led by an expansion of manufacturesexports-will come at Westernexpense. The most extreme expres-sion of this fearwas Ross Perot's warningthat the North AmericanFree TradeAgreement would lead to a "great sucking sound" asAmericanjobs moved to Mexico.Yet more respectablevoices raisesimilar concerns. Indeed, the White Paper of the Commission ofthe EuropeanCommunities [1993], in effect asserted that the riseof Third World manufacturing nations has already had seriousadverseimpacts. It claimed that the single most importantreasonfor the secularupwardtrend in Europeanunemploymentrates wasthe rise of countries that "compete,even on our own markets, at

*We are grateful for the support of the Centre for Economic Performance(London School of Economics) financed by the U. K. Economic and Social ResearchCouncil.? 1995 by the President and Fellows of HarvardCollegeand the Massachusetts Institute ofTechnology.TheQuarterly ournalofEconomics,November1995

8/2/2019 2946642

3/25

8/2/2019 2946642

4/25

8/2/2019 2946642

5/25

860 QUARTERLY OURNAL OF ECONOMICSties are similar to those that arise from the interaction betweentransport costs and labor mobility in recent models of economicgeography (e.g., Krugman [1991]). However, our model differsfrom these in important ways. The mechanism creating theexternalities is linkages between firms (through the input-outputstructure), rather than linkages between firms and worker/customers (as in Krugman ([1991]). Since we do not assume labormobility, the model is applicable to international as well as tointerregional economics. Immobility of labor also changes resultsin important ways. Simplegeographymodels like Krugman [1991]respond in a monotone way to declining transport costs: whenthese costs fall below a critical level, industry concentrates in oneregion. Here, because labor is immobile (and thus wage differen-tials between regions emerge), continuing reductionsin transporta-tion costs eventually lead to a reindustrializationof the low-wageregion. We believe that this is not just an artifact of the model: itrepresents a real distinction between interregional and interna-tional economics because labor is in fact far less mobile betweenthan within nations.The remainder of this paper is in six parts. Section I offers aninformal exposition of the model's logic. Section II sets out theformal model, while Section III shows how equilibrium is deter-mined, and how this equilibrium changes as the world economybecomes increasingly integrated. Section IV then shows hownational welfare changes as globalization proceeds. Section Vexplores the effects of trade policy, and finally, Section VI offerssome conclusions and suggestions for further research.

I. THE BASIc STORYWe imagine a world consisting of two regions, North andSouth. Each regioncan producetwo kinds of goods: "agricultural"goods that are produced with constant returns to scale, and"manufactured"goods that are subject to increasingreturns. Themanufacturing sector producesboth final goods sold to consumersand intermediate goods used as inputs in production of othermanufactures. All countries are equally proficient in both sec-tors: neither region has any inherent comparativeadvantage inmanufacturing.Wesupposethat initiallytransportationcosts between the tworegions are very high. Clearly, in this case each region will be

8/2/2019 2946642

6/25

GLOBALIZATIONND INEQUALITY 861essentially self-sufficient,and each regionwill produceboth manu-facturedandagriculturalgoods.

Now imagine graduallyreducing transportation costs. Therewill now be the possibility of trade between the regions. If (as wewill assume to be the case) there are many differentiatedmanufac-tured products,some two-waytrade in manufactureswill arise. Solong as transport costs are high enough, however, there will be nospecializationat the aggregative evel.At some point, however,a circularprocessarises that leads toregional differentiation.Suppose that one region for some reasonhas a larger manufacturing sector than the other. This regionoffers a large market for intermediate goods, and thus makes theregion, other things equal, a more attractive place to locateproduction of such goods. (This effect corresponds to the tradi-tional development concept of "backward inkages.") But if oneregion producesa greater variety of intermediate goods than theother, better access to these goods will, again other things equal,mean lower costs of productionof finalgoods (aneffectcorrespond-ing to the conceptof "forward inkage"), leadingto a further shiftof manufacturingto that region, and so on. When transportationcosts fall below some critical point, then the world economy willspontaneously organize itself into an industrialized core and adeindustrializedperiphery.If the manufacturing sector is large enough, this differentia-tion of roles will be associated with a divergencein real wages aswell. The self-reinforcing advantage created by backward andforward inkageswill drive up demandfor laborin the industrializ-ing region, while the decline of industry in the other region willlead to falling labor demand. Thus, real wages will typicallyrise inthe region that becomes the core and fall in that which becomesthe periphery. Global economic integration leads to unevendevelopment.But now supposethat transportationcosts continue to fall. Asthey doso, the importanceofbeingclose to marketsandsuppliers-and thus the importanceof forwardand backward inkages-willdecline as well. Meanwhile,the peripheralregion will offerpoten-tial producersthe advantage of a lower wage rate. At some pointthe decline in transportationcosts will be sufficientthat the lowerwage rate in the peripherymore than offsets the disadvantageofbeing remote from markets and suppliers.At this point manufac-turing will have an incentive to move out from the core to theperipheryonce again, forcinga convergenceof wagerates.

8/2/2019 2946642

7/25

862 QUARTERLY OURNALOF ECONOMICSThis intuitive story suggests that a single cause-the long-term declinein transportationcosts, leadingto growingintegration

of world markets-can produce first a division of the world intorich and poor regions, and then a convergence in incomes andeconomicstructurebetween those regions.To study the insights of this intuitive story, however,we mustturn next to buildinga formal model.

II. A FORMALMODELWe assume the existence of two economies,North and South,which are identical in endowments, preferences, and technology.We describethe Northern economy, simply noting that analogousconditionshold in South.North is endowedwith L units of labor, with wage rate w. Itcontainstwo sectors,agricultureand manufacturing.Therepresen-tative consumer in each country receives only labor income, and

has Cobb-Douglaspreferencesbetween agriculture and manufac-turing. These preferences can be represented by an expenditurefunction Q('-z)QmVn which V is utility, QA is the price ofagriculture, QM s the price index for manufactures, and Mys theshare of manufactures in consumer's expenditure. The budgetconstrainttakes the form,( 1) wL = QA Y)QMV.

The manufacturing sector producesa number of varieties ofdifferentiatedproducts,which are aggregatedby a CES subutilityfunction into a compositegood.Thepriceindex of this manufactur-ing compositeis QM, nd takes the form,(2) QM = [npl'a + n*(p*t)l-a]l/(l-a),where n is the number of varieties producedin North. In equilib-rium these are all sold at the same price p. Similarly, n* is thenumberproduced n South and sold at pricep *. Southern productssold in North incur iceberg transport costs at a rate t; i.e., aproportion 1/t of the good arrives implying a consumer pricep*t.or> 1 is the elasticity of demandfor a single variety.Turning to the supply side, we assume that agriculture isperfectlycompetitive,and uses only labor with constant returns toscale. We let agriculture be the numeraire (QA= 1), and assumethat it can be costlessly traded. Choosingunits such that one unit

8/2/2019 2946642

8/25

GLOBALIZATION ND INEQUALITY 863of labor produces one unit of output gives the equilibrium condi-tion:(3) w > 1.The wage rate equals one if the economyproducesagriculture,andexceedsit only if agriculturalproduction s zero.Firmsin manufacturinguse laborand a compositemanufactur-ing intermediate good to produce output. We make the majorsimplifying assumption that the composite intermediate good isthe same as the compositeconsumption good.Thus, the priceindexof the intermediate is QM,as defined in (2) above. Labor and theintermediate are combined with a Cobb-Douglas echnology withintermediateshare pu.Each firmproducesoutput for domestic sale(y) and export (x), with productionusing a units of the input as afixed cost and P per unit output thereafter. Each firm's total costfunction is therefore(4) TC = w1-AQ [a + B(y + x)].

Given this description of preferences and technology, we cannow characterize equilibrium as follows. First, define the totalvalue of expenditure on manufactured goods in the Northerneconomyas E. Then we have(5) E =ywL + >(x + y)pn.The first term on the right-hand side is consumers'expenditureonmanufactures, and the second intermediate demand, where wehaveused the fact that proportionpu f costs (and since there are noprofits,of revenue) is spent on intermediates.Next, note that firms mark up price over marginal cost by afactor ur/(ur 1), so that pricesare set according o the condition,(6) p(l - 1/ur)= w1-1QMP.Now note that Northern and Southern demandfor a single varietytake the form,(7) y =paQa-1E, x = p-at1-a(Qm*)aE*.With free entry and exit of firms, there is a zero profit conditionthat, as usual in this type of model, establishes a unique size offirm,(8) y + x = (ur 1)a/P.We choose units of measurement such that the right-hand side of

8/2/2019 2946642

9/25

864 QUARTERLY OURNALOF ECONOMICSthis equationis equalto unity, and use (7) in (8) to expressthe zeroprofit conditionas(9) 1 = pLa[Qaf1E + tl-a(Qm*)a-lE*].Equilibrium s now characterizedby equations (2), (3), (5), (6), and(9) (and analogous equations for the other region) which can beused to find equilibriumvalues of variables QM,w,p, n, andE.Beforediscussing the solution of the model, it is importanttounderstandthe wayin whichn, the numberof firmsin manufactur-ing, affectsfirms' profitability.It does this throughthree channels.The first is the standard one. An increase in n reduces the priceindex QM, equation (2)), thus shifting the demandcurve for eachfirm down(equation(7)) andreducingfirms' profitability(equation(9)).The secondandthirdchannels operateonlyif pus positive;i.e.,manufacturinguses manufacturingas an input. The reduction inQM ssociatedwith an increasein n nowreducestotal and marginalcosts ((4) and (6)), thus raising firms' profits. This is a cost, orforward linkage between firms. An increase in n also increasestotal expenditureon manufacturedproducts,E (equation(5)), thusraising demand and profits of each firm (equations (7) and (9)).This is the demand, or backwardlinkage between firms. It is thepresenceof these linkages that generates the effects we describeinthis paper.

III. OUTPUT AND EMPLOYMENTIn order to see how the model works, we first see whatdetermines the allocation of manufacturing between the twocountries, and the allocation of labor in each country betweenactivities. Analytical study of the equilibrium is algebraicallycomplex, so our main tool for exposition of the properties of themodel is numericalsimulation.Analyticalresults are derived n theAppendix.This is a general equilibrium model, and as in any generalequilibriummodelof tradeeach industrymust in effect competeontwo fronts. On one side, it must compete for markets with foreignfirmsin the same industry.On the other side, it must competewiththe other domestic industry for inputs. It is possible to representthe determinationof equilibrium n terms of at least two diagrams,each of which focuses attention on one of these competitivefronts.One such diagram s illustratedin FigureI. On the axes of thisfigure are the number of manufacturingfirms n and n* in North

8/2/2019 2946642

10/25

8/2/2019 2946642

11/25

866 QUARTERLY OURNAL OF ECONOMICS

1.4

1.2-

S~~~~~~~~~-------- _ _--- -- -- - ---- -- -- -- -- -- ---LA LA0.8 -

LM0.6 -

I I I I0 20 40 60 80 100%Labor orcein manufacturingFIGUREI

Labor Demand: High Trade Costs

fromthe left- and right-handends, respectively.The vertical axis isthe wagew.The brokenline LALAis the demand function for agriculturallabor, it representsequation (3), and our simple structure ensuresthat it is horizontalat height unity. The solidline LMLMs demandfor labor in manufacturing. It gives the maximum wage thatNorthern firms can pay and break even as a function of Northernmanufacturingemployment,LM,given that Southern manufactur-ing is in equilibriumwith w* = 1. (Toput it another way, one maythink of derivingthis scheduleby slidingdown SS in Figure I, andcalculatingthe maximumwageconsistent with nonnegative profitsin Northern manufacturingat each point.) The schedule is com-puted as follows. Northern employment in manufacturing isrelatedto the value of outputby the equation,(10) WLM= (1 - pu)np(y+ x).That is, a proportion(1 - pu) f firms' revenue is devoted to thewage bill. We assume that agriculture is active in the othercountry,so w* = 1, and then use equations (2), (5), (6), and (9) (and

8/2/2019 2946642

12/25

GLOBALIZATION ND INEQUALITY 867

1.4 s

1.2 -LMA

1------------------ 1 ----LA LA0.8-

LM0.6-

0 20 40 60 80 100%Labor orce in manufacturingFIGUREIILaborDemand:LowTrade Costs

their foreign analogs) to trace out manufacturing equilibriumas afunction of w. Using this with equation (10) gives the illustratedrelationshipbetween w andLM.Equilibrium s at the intersection of the two curves, point S. Inthe case illustrated in Figures I and II, the equilibriumis symmet-ric with each economy having a wage equal to unity and producingboth agricultural and manufacturing output. The proportionsLM/L and LAIL are equal to y and 1 - y, respectively. There is nonet trade (althoughthere is intraindustrytrade in manufactures),so employmentshares are determinedby shares of the two sectorsin finalconsumption.Figures I and II are constructedwith a high level of trade costs(t = 3). FigureIII is the analogousdiagramwith a much lower levelof trade costs (t = 1.5). The striking point is that the slope of themanufacturing labor demand curve is now positive, so that theequilibriumat U is unstable, and there is another equilibriumatpoint S. At this point North specializesin manufacturingand has awage above the value marginalproductof laborin agriculture.Allagriculturaloutput is producedin South, which may also produce

8/2/2019 2946642

13/25

868 QUARTERLY OURNALOF ECONOMICSmanufactures.2The figure is producedwith the assumption thatSouth becomesthe agriculturalexporter,but countrylabels mayofcourse be reversed.Thereare thereforethree equilibria:U, S, andafurther stableequilibriumwith agriculture operatingin North andmanufacturing concentratedin South. In our discussion this casewill be ignored.The reason for the reversal of slopeof the manufacturing abordemand schedule is the presenceof linkagesbetween manufactur-ing firms. Imagine relocating a firm from South to North. Thisraises demand for Northern firms' output, via the demand linkage,since at positive trade costs firms' demand for intermediates fallsdisproportionatelyon firms at the same location. It also reducesNorthern firms' costs, via the cost linkage, as another variety ofintermediate does not have to bear trade costs. Both these linkagescreate forces for agglomeration of manufacturing in a singlelocation. At high trade costs (Figures I and II) these forces aredominatedby the need to be near finalconsumerdemand.At lowertrade costs they are powerful enough to make the symmetricequilibriumunstable and cause manufacturing agglomeration.If the share of manufactures in final consumption (y) is lessthan or equal to 1/2, then all manufacturing agglomerates in asingle country, and the equilibrium has w = w* = 1. In this caseworld manufacturing demand is small enough to be met from asingle location. But if y > 1/2 (as illustrated),then the equilibriummust involve w > w*. One country specializes in manufacturing,any further demand formanufacturing s met bythe other,and theinternational wage differential offsets the locational disadvan-tage suffered by Southern firms distant from their markets andsuppliers.Figure IV illustrates the structure of equilibria at an interme-diate level of trade costs (t = 2). Four equilibria are illustrated (afifth in which South has no agriculture is not shown). At thisintermediate level of trade barriers, linkages are not powerfulenough to destabilize the symmetric equilibrium.But if North hasall its labor employed in manufacturing, then linkages are suffi-cient to ensure that this is an equilibrium.Figures I-IV suggest that as the level of trade costs is reduced,there are two points at which the qualitativecharacter of the set ofequilibria changes.At high levels of trade costs, the unique, stable

2. North is specialized n manufactures;but does South specialize in agricul-ture, or does it producesome manufacturesas well? This is a somewhat difficultquestionto analyze;we discussit furtherin Appendix3.

8/2/2019 2946642

14/25

GLOBALIZATIONND INEQUALITY 869

1.4

1.2-LMS

1 -------- ____~~~~~U U LA0.8- Lm

0.6t , ~ ~~ ~~~~~~~~~~~~~~III0 20 40 60 80 100%Labor orcein manufacturing

FIGUREIVLaborDemand:IntermediateTrade Costs

equilibrium is one in which manufacturing is equally dividedbetween the countries. At some point additional, asymmetricequilibria emerge. Finally, when transport costs fall to a criticallevel, the symmetricequilibriumbecomesunstable. If we think of ahistorical sequence in which trade costs graduallyfall overtime, itis this latter level at which symmetry is broken and the core-periphery pattern emerges. In the Appendix we show that thecriticallevel of t is definedby

(11) t = (1~ tu(u(1 + p) 1Whatdeterminesthis criticallevel?Frominspectionof (11) we

first notice that asymmetryonly arises if there is a significantroleof manufactured goods as intermediates. If p.were close to zero,there would be few forwardandbackward inkages.Indeed,we cansee from (11) that p. = 0 would imply a critical t of unity; i.e., anyt > 1 (anypositivetradecosts) wouldimply symmetrybetween theeconomies.At the opposite extreme, if u(1 - p.) < 1, the expressionbecomes negative. The interpretation of this is that a core-

8/2/2019 2946642

15/25

8/2/2019 2946642

16/25

8/2/2019 2946642

17/25

872 QUARTERLY OURNALOF ECONOMICS1.2

1

Q)20.8 vvQ)

x -S -- - - - - - -0.6

0.4- l1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8

tFIGURE VIReal Wages: High y

more ambiguous. Relocation of firms reduces wages in terms ofagriculture. The consumer price index may, however, move ineither direction. On the one side, an increasing proportion ofmanufacturing s being importedand is thus subjectto trade costs.On the other side, these trade costs themselves are being reduced,directly tending to reduce the price index. Thus, real wages canmove either way. In the case illustrated in Figure V, real wages fallas trade costs arebrought down to very low levels.Figure V was constructed for the same parameter values asFigures I-IV. Figures VI and VII indicate the effects of changingtwo parameters, the share of manufactures in demand and theshare of intermediatesin manufacturing.In FigureVI the share ofmanufactures in demand is increased. This increases the amountof manufacturing activity in South, and thereby reduces real wagedifferences.In this case the "globalization"phase does not involvefallingrealwages in North.FigureVII illustrates the casewhen the share of intermediatesin manufacturing s raised.Agglomeration orces arenow stronger,creatinga wider realwage differential.Whereasin previousfiguresthere is manufacturingactivity in both North and South, there is

8/2/2019 2946642

18/25

GLOBALIZATION ND INEQUALITY 8731.2 -

V1

'4

0.8 Vva

0.6

0.4- I I1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8

tFIGURE IIRealWages:High p.

now a range of transportcosts-the intervalt E [1.28, 1.85]-overwhich all manufacturing s concentrated n North. Northernwagesare determinedby the condition that the value of manufacturingoutput equals the value of expenditure. This generates a largeNorth-Southwagegap,but not largeenoughforSouthernmanufac-turing to be profitable.Conditionsunder which this concentrationof manufacturing occurs are given in Appendix 3. Notice thatNorthern real wages are constant in this range, as trade costs areassumedto affectonlymanufactures.These results are, of course, based on numerical examples,that is, on particularparameter values. Nonetheless, the generalpicture-in particular,the sequence of phases with initial separa-tion into core and periphery followed by a return to factor priceequalization-is general given this model. As long as there aresome linkages (ix > 0) but these are not too strong (ou(l- pu) 1),there is always a critical level of t below which the equal-wageequilibriumis unstable. And it is always the case that as t -* 1 thewage differential between countries must also disappear. Thus,while the details depend on parameters, the general picture of aU-shapedresponseof relative wages to transportcosts does not.

8/2/2019 2946642

19/25

874 QUARTERLY OURNALOF ECONOMICS1.2

k IQ) 0.8Q \

------- i,0.6 V*

1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8t

FIGURE IIIRealWageswith a NorthernTariff

V. NOTES ON TRADE POLICYThe final phase of the processof globalizationdescribedby ourmodel, in which the spread of industry to the South reducesrelative and perhaps absolute Northern wages, obviously corre-sponds to the fears of many commentatorson the worldeconomy.Among these, some, such as billionaire-turned-punditSir JamesGoldsmith, whose recent book [1994] has been a European best-seller, advocateprotectionistpoliciesto prevent globalcompetitionfrom depressing wages.A Northern tariff affects location of industry in two ways.First, by worsening Southern producers' access to the largeNorthern market,it tends to drawfirms to the North. Against this,Northern firms now pay more for intermediate goods importedfrom South. The net effectis to attract firms to North from South,widening wage differentials.3This is illustrated in Figure VIII,

3. To see this, considera small Northerntariff dTat a point close to t = 1 (atwhichmanufacturings dividedequallybetweenNorth andSouth).The tariffraisesSouthernfirms' cost on half their sales by dr. It raises Northern firms costs on alltheir sales by pLdT/2sincehalf of intermediatesare imported).The former effectislargerthan the latter, so Southernfirmsare hit harderby the tariff than Northernones.

8/2/2019 2946642

20/25

GLOBALIZATIONND INEQUALITY 875which compares Northern and Southern real wages as t declinestoward unity under two different scenarios: free trade, and aNorthern tariff of 33 percent on manufactures imports. Parame-ters and free trade real wages (the lighter lines) are as in Figure V.The heavy lines give realwages with the Northern tariff.Northernwages are higher than under free trade. Thus, the claims of somefree tradersthat protectionismis necessarilya self-defeatingpolicyare not borne out. Additionally, North receives tariff revenue,which is not includedin the figure.Two crucial cautions should, however, be made about theseresults. First, in supposing that North as a whole imposes a tariffagainst South, we have in effect gone beyondregardingNorth andSouth as regions and treated them as political units or at leastcustoms unions. A general outbreak of protectionism, in whichhigh-wage nations restricted imports from each other as well aslow-wagenations, would clearlyproducea very differentoutcome.By raising the prices of intermediates traded intra-North, North-ern industry would suffer. To put it differently, the trade policyexperiment described by Figure VIII is one in which trade policyhas in effectbeen taken over by disciplesof Goldsmith,who wantsfreetradein manufacturesamonghigh-wagenationswhileprevent-ing imports fromlow-wagecompetitors.Furthermore, it is important to point out that the modeldoesnot at all bear out the claims of some modernprotectionists that aregime which allows trade only between countries with similarwagerates is somehowin the interests of laboreverywhere.On thecontrary, in the scenario described by Figure VIII Northernworkers are protected from wage decline only by suppressingincipient Southern industrialization, and thereby also keepingSouthern realwages low.

VI. SUMMARYANDSUGGESTIONSORFURTHER ESEARCHThe conventional wisdom of economic analysis is that while

greater global integration may hurt particular interest groups, itwill normally raise the overall real income of just about everynation. There are exceptionsto this rule even in the most conven-tional model: barriers to trade, natural as well as artificial, maysometimes act as de facto optimal tariffs, and their removal maytherefore leave some countries worse off. Nonetheless, standardtrade models do seem to suggest a presumption that integration isan all-around good thing.

8/2/2019 2946642

21/25

876 QUARTERLY OURNAL OF ECONOMICSCritics of this conventionalwisdom have long arguedthat, onthe contrary, greater integration usually produces national win-ners and losers. Traditionally, heterodox critics have argued thatintegration fosters inequality, that an integrated world economydividesinto a richcoreand a poor periphery,and that the wealth ofthe center comes at the periphery's expense. Only recentlyhas thecontrary argument,that globalizationbenefits the peripheryat thecore'sexpense, gained ground.What we have shown in this paper is that a simple model inwhich regionaldifferentiation s drivenby the interaction betweenscale economies and transport costs makes sense of both old andnew arguments. The world economy must achieve a certain criticallevel of integration before the forces that cause differentiation ntocore andperipherycan take hold. When that differentiationoccurs,the rise in core income is partly at peripheral expense. As integra-tion proceeds further, however, the advantages of the core areeroded,and the resulting rise in peripheral ncome may be partlyatthe core'sexpense.There are obviously many ways in which this analysis could beextended. We would, however, emphasize three directions inparticular.First, it would be desirable to get more geography into thismodel. As it stands, we postulate the existence of two exogenouslydefinedregions, which then take on endogenouslyderived roles. Inpractice, the core has gradually spread into what was the periph-

ery, with such areas as the southern United States, much ofsouthern Europe, Japan, and now some of East Asia effectivelymaking a transition from agricultural suppliersto manufacturingexporters.So we would like to extend the analysis to a multiregionsetting, perhapseven one with continuous space.Second, our model excludes capital mobility: indeed, it has nocapital.Yet much of the politicaldebate overintegration focuses onthe alleged impacts of capital movement rather than (or alongwith) trade flows. Thus, a natural step would be to add capitalmovements.Finally, it is obviously important to discipline this analysiswith some real numbers. We have offered a stark, one-factorexplanationof vast global trends. This is in itself worrying.Worseyet, it is an explanationthat will appealto the prejudicesof manypeople. Thus, it is crucial to do at least rough empiricalwork to seewhether the kind of story describedhere is at all likelyto be a large

8/2/2019 2946642

22/25

GLOBALIZATION ND INEQUALITY 877part of the explanationof real globaleconomic trends. We suspectthat as a practicalmatter growingintegration with the South is atbest a minor factor in the economic woes of the North, but theimportant point is that one should be careful about assuming thatsomethingthat is possiblein principleactuallyhappensin practice.In spite of these cautions, we regardit as a useful exercise toconstructa minimalist modelof the kind describedhere, and find itremarkable that so simple a structure can give rise to such asweeping picture of divergence and convergence in the globaleconomy.

APPENDIX 1: THE ALGEBRA OF SYMMETRY-BREAKINGDefine T_ t'- and the ratios of Northern to Southern valuesof endogenousvariables as follows:

- QM -p - E W(Al) QM-- Q*, P -* E -E*, W--*.Using equation (10), we can express the ratios of Northern andSouthern expressions (2), (5), (6), and (9), as(A2) O =+LX=TLmWf-r~+ L*M

(A3)~ ~ ~~~~- [(1 - Pi)L+ jiLM](A4)(A5) pa = 7QM 1.EliminatingQM ndE gives(A6) - L&~L- 1)(1 - uIp. = MZJU+T~?TLmWP-q'LMand

1)(u- ~ -y(l - j~i)L+ p~LL _____(A7) j3(U1-1 (A - [/ W [(1 - )L + p.LL = a1These equations express tb andf as a function of LMand L*. Byinspection, if LM= L;, there is a solution to these equations with

8/2/2019 2946642

23/25

878 QUARTERLY OURNAL OF ECONOMICSp = 1, and ib = 1. Consider a small change dLMwith associatedchange - dL* in the neighborhood of these values. Totally differ-entiating and applying Cremer'srule yield

dw (v - 1\ (j - 1)[U(,u- 1) + 1] - T(, + 1)[U(,u+ 1) - 1]dLm TOL 2f(a - 1)(1 - >L)+ (T - 1)[Uf(> + 1) - 1]

This derivative is the slope of the LMLMcurve at the symmetricequilibrium. Given that T E (0,1), a sufficient condition for thedenominatorof this expressionto be positive is a(1 - pu)> 1. Thenumerator is positive or negative accordingto whether t is greateror less than the value implied by equation (11) of the text.APPENDIX2: PARAMETERVALUES

The simulations of Figures I-V set L = L*, y = 0.6, tL= 0.5,a = 5. In Figure I, t = 3, and in Figures II-IV, t = 3, t = 1.5, t = 2,respectively. In Figure VI mys increased to 0.7. In Figure VII p.isincreased to 0.55, and mys returned to 0.6.

APPENDIX3: THE PATTERNOF SPECIALIZATIONAs long as the share of manufacturers n final demandexceedsone-half, any asymmetric equilibrium must involve specializationby one corecountryin manufactures. But doesthe other, peripherycountry also produce some manufactures? It is possible to shed

some light on this question algebraically.Suppose that all manufacturing is in North and thereforew= 1. We then have

wL wL(A9) E+E*= 1wL E*=yL*, E=ywL+ ( ).The first of these says that total expenditure equals the value ofoutput (which is the wage bill divided by the labor share). Thesecond and third give manufacturingexpenditurein each location,with Northern expenditure including intermediate demand. Set-ting L = L* = 1 and solving,(A10) w= - _y' E* = 'y, E = y

A necessary condition for this to be an equilibriumis that it isnot profitablefor any firmto start producing n South. This can be

8/2/2019 2946642

24/25

8/2/2019 2946642

25/25

880 QUARTERLY OURNAL OF ECONOMICSREFERENCES

Commissionof the EuropeanCommunities,"Growth,Competitiveness ndEmploy-ment: The Challengesand WaysForwardto the 21st Century,"WhitePaper,Bulletin of the European Communities, 6/93 (1993).Ethier, Wilfred, "National and International Returns to Scale in the ModernTheory of InternationalTrade," AmericanEconomicReview, LXXII(1982),389-405.Goldsmith,James, TheTrap(London,UK:Macmillan,1994).Krugman, Paul, "Scale Economies, Product Differentiationand the Pattern ofTrade," American Economic Review, LXX (1980), 950-59."IncreasingReturnsandEconomicGeography," ournalof PoliticalEconomy,XCIX 1991), 483-99., "Is the Third World a Threat?" Harvard Business Review, (July/August1994), 113-21.

Krugman, Paul, and Robert Lawrence, "Trade, Jobs, and Wages," ScientificAmerican, CCLXX1994), 22-27.