Does European Unemployment Prop Up American Wages?National Labor Markets and Global Trade: Comment

By JURGEN MECKL*

In his much discussed “two sides of the samecoin” hypothesis, Paul R. Krugman (1995) at-tributes the contrasting labor-market experiencein America and Europe to institutional differ-ences. When faced with common exogenousshocks that raise the relative labor demand infavor of skilled labor, countries with flexiblewage rates (America) should experience a risein wage inequality, while wage inequalityshould rise considerably less in countries withwage rigidities (Europe) at a cost of increasedunskilled unemployment.1 Although Krug-man’s argument is, in principal, based on ageneral-equilibrium model of internationaltrade, the classical Heckscher-Ohlin (HO)model, Donald R. Davis (1998) challenges thisapproach as inconsistent with a global general-equilibrium approach. He argues that compara-tive studies based on separate applications ofthe general-equilibrium framework to differentregimes of wage flexibility ignore the factor-price equalization (FPE) property that applies ina global equilibrium context. Because of theglobal equilibrium’s FPE property, an exog-enously given minimum wage in one part of thetrading world ties down all prices for goods andfactors all over the world, and thus implies realwage rigidity for all countries producing acommon set of goods. The commitment ofEuropean labor-market institutions to highwages then fully protects competitive Amer-ican labor markets against both global exog-

enous shocks and U.S.-specific labor-supplyshocks (like, e.g., immigration from Mexico).The link between national factor markets es-tablished by integrated world markets forgoods thus generates quite counterintuitiveresults which are rather contrary to those fromcomparative studies.

The present paper develops a model that in-tegrates institutionally set minimum wages intoa global HO-type model without generatingDavis’s counterintuitive results, although it pre-serves the global-equilibrium FPE properties.2

It extends Davis’s approach by accounting forindividuals with differences in productivitiesboth within and between different skill groupsof labor. This generates differences in individ-ual demand for education and endogenizes therelative labor supply.3 As long as workers arerewarded according to their individual produc-tivity, this heterogeneity also disentangles theeffective wage (the firms’ costs per unit ofeffective labor input) and the hourly wage(the wage per physical unit of labor). Withminimum-wage legislation referring to thehourly wage (since this is the only wage that canbe fixed), an exogenously determined wagefloor does not tie down all prices. As a result,the present model preserves the results fromthe standard HO model in a qualitative sense,and it extends them for unemployment effectsand endogenous adjustment of relative skillsupplies.

* Department of Economics, Justus Liebig UniversityGiessen, Licher Str. 66, D-35394 Giessen, Germany, andInstitut fur die Zukunft der Arbeit (IZA), Bonn, Germany(e-mail: juergen.meckl@wirtschaft.uni-giessen.de). I thankMax Albert, Benjamin Weigert, Stefan Zink, and two anon-ymous referees for helpful suggestions and comments. Fi-nancial support from the German Research Foundation(FOR 454) is gratefully acknowledged.

1 Of course, Krugman’s argument is a bit more differen-tiated. Wage inequality increased considerably in the UnitedStates and the United Kingdom, whereas we observe moreeffects on unemployment in continental European econo-mies. For sake of simplicity, however, we stick to thestylized U.S.-Europe dichotomy.

2 Other authors (cf., e.g., Anthony B. Atkinson, 2001, orPaul Oslington, 2002) criticize Davis’s approach because itimplies FPE. They suggest alternative solutions to integrateunemployment in an HO-type model that dispense withFPE. In contrast, the present approach preserves the FPEproperty while at the same time confirming the results fromcomparative cross-country studies.

3 Davis and Trevor A. Reeve (2002) also extend theDavis (1998) framework by endogenizing the choice ofeducation. Since they do not account for within-groupworker heterogeneity, however, wage rigidities still deter-mine all prices as long as the FPE property prevails.

1924

I. Davis’s Insulation Hypothesis

Davis considers two countries—America andEurope—sharing identical constant-returns-to-scale production technologies. Each countryproduces two goods (X and Y) that are freelytraded. Production uses two types of labor in-puts, skilled and unskilled labor. Both factorsare assumed to be indispensable in production.Factor supplies are exogenously given in eachcountry. The key difference between Americaand Europe is in terms of labor-market flexibil-ity. Whereas the U.S. market for unskilled laboris characterized by perfect wage flexibility, theunskilled labor market in Europe is character-ized by wage rigidities (caused by, e.g., unions,explicit minimum wages, or social security sys-tems). The market for skilled labor is assumedto be perfectly competitive in both countries.Finally, consumer preferences are assumed tobe homothetic and identical in both countries,with both goods being essential in consumption.

The basic argument can be stated by lookingat the production side of an economy. Assumethat, at any common factor price ratio, good X isskill intensive relative to Y. Denote the wage ofunskilled labor by w, the wage of skilled laborby r, and goods prices by pX and pY. Competi-tive cost conditions then ensure that for eachactive sector, price equals unit costs:

(1) pX ! cX !w, r", pY ! cY !w, r".

With fully flexible wages and a fully diversifiedproduction structure, the zero-profit conditions(1) determine factor prices (w, r) solely as afunction of the goods’ prices (pX, pY).

With wage rigidities, however, the path of de-termination is different. By our assumption ofhomothetic consumer preferences, the unskilledworkers’ real wage is w/e(pX, pY), where the con-sumer price index used to deflate the nominalwage rate is given by the unit utility expenditurefunction e(pX, pY).4 Normalizing consumer prices

according to pX # p and pY # 1, we can write thatreal wage as z $ w/e(p), where e(p) # e(p, 1).The zero-profit conditions then read

(2) p ! cX !ze!p", r", 1 ! cY !ze!p", r".

For competitive firms in both sectors to pay aminimum wage z, this wage rate must be sup-ported by an appropriate goods price p! . Other-wise, factor payments either exceed revenues inthe labor-intensive sector for any positive out-put Y % 0 (in the case of p % p!), or theminimum-wage rate is not a binding constraint(in the case of p & p!). A global equilibriumfeaturing (a) diversified production in bothAmerica and Europe and (b) a binding mini-mum wage in Europe thus implies that the termsof trade are uniquely determined by Europe’sminimum wage. With perfectly competitive U.S.labor markets, American wages are then alsodetermined by Europe’s minimum wage rate.The FPE property of the diversified globalequilibrium implies that American wages arefully determined by European labor-marketinstitutions.

FPE in the presence of an exogenously givenminimum wage rate in one part of the worldimplies that this minimum wage determines allprices for goods and factors all over the world.Real wage rigidity in Europe actually impliesreal wage rigidity for all countries in the worldproducing a common set of goods. Adjustmentto exogenous shocks (such as exogenous laborsupply shocks or integration of newly industri-alized countries into world markets) can then beaccomplished solely by appropriate adjustmentsof production quantities that are consistent withfixed prices. The reason is that, to support agiven minimum wage w! under full diversifica-tion, the unique equilibrium price is p! . Withhomothetic consumer preferences, any exoge-nous shock in the relative supply of goods onworld markets has to be exactly compensated bysectoral adjustment in Europe in order to sup-port p! . The sole degree of freedom allowing forthe required sectoral adjustments is the (un)em-ployment of unskilled labor. The commitmentof European labor-market institutions to highwages wholly insulates competitive Americanlabor markets both from global exogenousshocks and—even more surprising—from U.S.-specific labor-supply shocks (e.g., immigrationfrom Mexico) as long as (a) the minimum wage

4 In principle, it is not necessary to distinguish producers’real wages and consumers’ real wages in the present context.This follows because fixing one relative price and assuming adiversified production structure ties down all relative prices. Innext section’s model, however, setting a real minimum wagedoes not tie down all relative prices. In order to ensure con-sistency of the exposition, we assume that minimum-wagesetting applies to the consumers’ real wage throughout.

1925VOL. 96 NO. 5 MECKL: NATIONAL LABOR MARKETS AND GLOBAL TRADE: COMMENT

is a binding constraint in Europe both beforeand after the shock and (b) Europe and theUnited States remain fully diversified after theshock.5

II. The Model

A. Heterogeneous Labor and Effective Wages

The present approach reconciles the commonview on wage determination and the global gen-eral-equilibrium analysis. We develop a modelthat allows for integration of institutionallycaused wage rigidities into an otherwise stan-dard HO-type model of international tradewhich does not generate the implausible resultsabout international wage determination orcross-country specialization patterns of theDavis model.6 The basic difference is that thepresent model allows for heterogeneous laborand its effect on individual wage incomes. Westart from the observation emphasized in theempirical labor-market literature that individualwage incomes differ because individuals differ(a) in their skill levels and (b) in their levels ofinnate abilities (cf. Christopher R. Taber, 2001).To capture (b), we distinguish between physicalunits of labor supply and effective units of laborsupply. The difference between labor in physi-cal and in effective units originates in differ-ences in individual abilities.

The distinction between labor supply in phys-ical units and in effective units is the main dif-ference between our approach and that followedby Davis, who implicitly assumes that individ-uals are homogeneous with respect to their abil-ities. As a result, both effective and physicallabor inputs and the effective wage (the factorprice of effective labor units) and the workers’hourly wage (the wage firms pay per physicalunit of labor) are also identical in Davis’smodel. In the present model, however, the ef-fective wage and the hourly wage are different.

Since only the hourly wage can be fixed insti-tutionally, the effective wage is still free toadjust in the present model. But it is the effec-tive wage that enters the firms’ cost minimiza-tion problems. Denoting the effective wage forthe unskilled by w and for the skilled by r,competitive cost conditions are again given by(1). As in the standard HO model with compet-itive factor markets, these conditions uniquelydetermine the factor prices w and r as a functionof goods’ price p:

(3) w ! w! p", r ! r! p";

w#! p" " 0, r#! p" # 0.

The signs of the derivatives in (3) reflect ourassumption that good X is relatively skillintensive.

B. Educational Decisions and AggregateFactor Supplies

We follow Meckl and Benjamin Weigert (2003)in modeling individual and aggregate factorsupplies.7 There is a continuum of agents. Eachagent has some ability a ! [0, 1]. The density ofagents with ability a is f(a). The mass of thedensity function is normalized to 1, that is, $0

1

f(a) da % 1. The composition of labor supply isendogenously determined by decisions of indi-viduals with different abilities. For the moment,let us dispense with any form of institutionalwage rigidities.

An individual with ability a can either enterthe labor force as unskilled, thereby supplying(1 & a) units of unskilled labor, and earn thewage rate w per unit of effective labor.8 Alter-natively, an individual can choose to spend anexogenously given fraction $ of time in trainingto become a skilled worker. Education is as-sumed to raise individual abilities. For simplic-ity, we assume individual abilities of skilledworkers to be ba, where b ' 1 can be inter-preted as a measure of the efficiency of theeducational system. Thus, a skilled worker with5 I suspect that most labor economists would disagree

with Davis’s results. For example, George J. Borjas et al.(1997, p. 62) find that immigration into the United Statesduring the 1980 –1995 period accounts for about one-quarter to one-half of the rise in wage inequality.

6 Cf. Max Albert and Meckl (2001) for an efficiencywage–based approach to introduce unemployment into ageneral-equilibrium trade model that preserves decisiveproperties of the standard HO model.

7 In contrast to the present paper, Meckl and Weigert(2003) emphasize the impact of educational decisions onmeasured wage inequality in an HO-type model of a smallopen economy.

8 Note that we implicitly normalize working hours tounity.

1926 THE AMERICAN ECONOMIC REVIEW DECEMBER 2006

ability a supplies (1 ! !)(1 " ba) units ofskilled labor and earns the wage rate r per unitof effective labor. The wage income of an indi-vidual with ability a then either is (1 " a)w asan unskilled worker, or (1 ! !)(1 " ba)r as askilled worker.

An individual chooses to become skilled ifand only if his ability is not smaller than somethreshold value t determined by

(4) t# p$ " %a : #1 # a$w# p$

$ #1 # ba$#1 $ !$r# p$ " 0}.

Since both skilled and unskilled labor are indis-pensable in production, we must focus on inte-rior solutions t ! (0, 1). This requires 2/(1 "b) & (1 ! !)%(p) & 1, where %(p) ' r(p)/w(p). Furthermore, given our assumptionsabout factor intensities, t is a function of p witht((p) & 0.

Individual education decisions determine theaggregate supplies of unskilled and skilled labor(L and H) as functions of p. Factor supplies aregiven by

(5) L# p$) !0

t# p$

#1 # a$f#a$ da,

H# p$) !t# p$

1

#1 $ !$#1 # ba$f#a$ da,

with L((p) & 0, and H((p) * 0. As a result,relative labor supply h(p) :) H(p)/L(p) is anincreasing function of the goods price p. Fullflexibility of factor prices then ensures that L(p)and H(p) are always fully employed, implyingthat h(p) also measures relative labor employ-ment.

Suppose now that there is an institutionallyset wage floor z below which the real hourlywages paid in an economy must not lie. Forsimplicity, we take the value of z as givenexogenously. Furthermore, we assume that thewage floor is sufficiently low, such that mini-mum-wage legislation is not relevant for skilledlabor. With respect to the unskilled, however,the wage floor may be binding. An unskilledworker with ability a receives a real hourlywage (1 " a)w/e(p). Existence of a real wagefloor then requires

(6)#1 # a$w# p$

e# p$& z.

Given that w((p) & 0 and e((p) * 0, the con-straint (6) is not binding for p ' p0, where p0solves

(7) w# p$ " ze# p$.

At prices p ' p0, all unskilled workers havesufficient productivity to earn the minimumhourly wage at the respective real return toeffective unskilled labor w(p)/e(p). With abinding wage floor, however, there are individ-uals with a productivity that is too low to guar-antee them a real hourly wage of z at the realreturn to effective unskilled labor w(p)/e(p).Consequently, these individuals are not em-ployed by any of the firms. The threshold ability( to become employed is then determined by

(8) ( #p, z$ " %a : #1 # a$w# p$ " ze# p$+.

Our assumptions about factor intensities implythat ( is increasing both in p and in z.

With wage rigidities, the aggregate unemploy-ment rate amounts to

(9) u# p, z$ " !0

( #p,z$

f#a$ da.

The unemployment rate rises with z and—dueto our assumptions about factor intensities—isalso increasing in p. Aggregate employment ofunskilled labor is

(10) L# p, z$ " !( #p,z$

t# p$

#1 # a$f#a$ da,

with )L(p, z)/)p & 0, and )L(p, z)/)z & 0.Consequently, relative labor employment h(p,z) :) H(p)/L(p, z) is increasing in both p and z.

C. Equilibrium Prices and Unemployment

Consider a world consisting of two countrieswith free trade in goods and no internationalfactor mobility. Both countries are alike in ev-ery respect except for labor-market institutions.Labor markets in one country—Europe—are

1927VOL. 96 NO. 5 MECKL: NATIONAL LABOR MARKETS AND GLOBAL TRADE: COMMENT

characterized by institutionally fixed wagefloors. Labor markets in the other country—America—are fully competitive. From our anal-ysis in the preceding subsection, we get that, forany goods price compatible with (weak) diver-sification and a binding wage floor, incentivesto acquire education are identical in both coun-tries.9 This implies identical supplies of bothtypes of labor in both countries. With no wagerigidities in America, the supply of unskilledlabor is fully employed there. Due to the wagefloor, however, Europe’s employment of un-skilled workers is less than America’s. Conse-quently, Europe’s relative employment ofskilled labor exceeds that of America, whileaggregate income in Europe is less than inAmerica because of unemployment.

Now we analyze the global equilibrium by ap-plying the graphical technique shown in Figure 1.The downward-sloping curve CC shows the rela-tive demand for goods as a function of p that iscommon to Europe and America (reflecting ourassumptions about consumer preferences). Theupward-sloping curve AA depicts the supply of Xrelative to Y of the U.S. economy. In contrast tothe standard HO model, an increase in p raisesrelative supply by two effects. In addition to theusual sectoral restructuring in favor of X at givenfactor supplies, the increase in the relative factorprice ! additionally lowers the threshold for ac-quiring education. Thus, the relative supply of

skilled labor rises reinforcing sectoral adjustmentsby a standard Rybcynski effect. In Europe thatsecond effect is stronger than in the US since anincrease in ! additionally raises the threshold ",thus reducing the aggregate employment of un-skilled labor by even more. As a result, relativeemployment of skilled labor exceeds relative em-ployment in the United States at any price thatgives rise to a binding minimum wage. For allprices p ! p0, Europe’s relative supply curve liesto the right of America’s, and its slope is less thanthe slope of the American relative supply curve.10

It is obvious from Figure 1 that equilibriumautarky prices (determined by the intersectionof national relative demand and relative supplycurves as pA and pE, respectively) differ, andhence international differences in institutionallyset wage floors give rise to trade. The openingup of trade implies a striking contrast betweenlabor markets in Europe and in America. Sup-pose that p* is the price in the free-trade equi-librium. In Europe, the rise in p reduces w andtherefore drives up unemployment (for a givenz). In America, the decline in p raises w. Inprincipal, these results are in line with that fromDavis’s analysis: opening up trade between Eu-rope and America raises both American wagesand European unemployment rates. The mech-anism driving these results, however, is com-pletely different. In our model, it is the changein goods prices and the resulting changes infactor prices that alter production structures,educational decisions, and the impact of wagefloors on unemployment. Basically, our mech-anism is similar to the well-known adjustmentmechanism from the standard full-employmentHO model. The only difference with respect tothe full employment version is that quantitativereactions additionally account for (a) endoge-nous labor supply adjustments that are gener-ated by a change in educational decisions ofindividuals and (b) employment adjustmentsfrom the change in the unemployment thresh-old. Eventually, our results provide a soundfoundation for the arguments made in the com-parative cross-country studies that have beencriticized by Davis.

9 This implicitly assumes that the parameters b and # areidentical for both countries.

10 In contrast, the Davis approach implies that Europe’srelative supply becomes infinitely elastic at p " p! (implyinga flat segment in Europe’s import demand curve). All hisrelatively drastic results are caused by this infinite elasticity.

FIGURE 1

1928 THE AMERICAN ECONOMIC REVIEW DECEMBER 2006

III. Shocks from the South: Implications forEurope and America

The insulation thesis, stating that exogenouschanges in world factor supplies affect neitherU.S. wages nor the aggregate global productionquantities, is the most controversial implicationof European wage institutions in Davis’s anal-ysis. The present model shows that the general-equilibrium link does not necessarily imply thatthe part of the world with wage rigidities has tobear the complete burden of adjustment to ex-ogenous shocks.

Consider immigration of unskilled labor fromthird countries into the United States. Supposefurther that immigrants cannot acquire educa-tion in the United States.11 Since decisionsabout acquiring education depend on the goodsprice, the nonmigrants’ investment incentives atany given p do not change after immigration.Immigration, however, reduces the relative sup-ply of skilled labor for each p, thus raising thedifference in effective relative labor suppliesbetween America and Europe. This change inrelative employment alters the U.S. relativesupply of goods in favor of Y at any admissibleprice. In terms of Figure 1, immigration shiftsthe American relative supply curve AA to theleft. Consequently, the new goods price sup-porting global equilibrium is some p! " p*.Thus, we arrive at the results prospected byempirical labor-market studies on the effect ofimmigration: immigration of unskilled laborinto the United States reduces the wage incomeof the unskilled there, while driving up theincentives to invest in education within the na-tive population. In Europe, the effects are iden-tical with respect to wage incomes of theunskilled, and with respect to educational incen-tives. Additionally, unskilled unemployment rises.

IV. Conclusions

This paper has shown that minimum-wage-in-come constraints can be introduced into the clas-sical Heckscher-Ohlin framework of internationaltrade without altering the results derived form thestandard full-employment version of that model ina fundamental way. Contrary to what is expected

from Davis’s (1998) analysis of minimum wagesin a HO-type model, wage rigidities in Europe donot insulate flexible-wage economies like Amer-ica from exogenous shocks by shifting the com-plete burden of quantity adjustments to Europeanlabor markets. In our model, each exogenousshock renders price adjustments that generate ad-justments in national production structures for alltrading partners. Nevertheless, European unem-ployment does indeed prop up U.S. wages. This isa consequence of national labor markets beinglinked by integrated goods markets in a globaleconomy. Global general equilibrium effects donot, however, equalize wage incomes in countrieswith completely different labor-market institu-tions. Our results, and the mechanisms drivingthem, are more in line with those derived fromcomparative studies (cf. Krugman, 1995), thusproviding a consistent common framework forthese results.

Our model also overcomes another problem-atic implication of Davis’s approach concerningspecialization patterns within European econo-mies. In European countries sharing wage rigid-ities, minimum wages are by no means identical.But if minimum-wage rates differ between Eu-ropean economies, Davis’s HO-type frameworkimplies either that only the highest minimumwage is actually binding (with no unemploy-ment in all other European economies) or thatcomplete specialization of production on theskill-intensive good occurs in all economieswith binding minimum-wage rates,12 but theeconomy with the lowest minimum wage. Thisimplication, however, is clearly contrary to fact.Only if one allows for specific technologicaldifferences within European countries it is the-oretically possible that different minimum-wage rates are supported by an identical goodsprice p! , and that diversification occurs in allcountries. But this is a highly specific case ofminor practical importance. Since in our modelthe minimum-wage–income constraints do notcompletely determine factor prices, the model iscompatible with diversification in all econo-mies, irrespective of the specific value of theminimum-wage income.

11 Alternatively, one can analyze immigration of individ-uals at the bottom end of the ability scale.

12 This corresponds to the minimum-wage–induced spe-cialization patterns in a small-open-economy version of theHO model discussed by J. Peter Neary (1985).

1929VOL. 96 NO. 5 MECKL: NATIONAL LABOR MARKETS AND GLOBAL TRADE: COMMENT

REFERENCES

Albert, Max, and Jurgen Meckl. 2001. “Efficien-cy-Wage Unemployment and Intersectoral WageDifferentials in a Heckscher-Ohlin Model.” Ger-man Economic Review, 2(3): 287–301.

Atkinson, A. B. 2001. “A Critique of the Trans-atlantic Consensus on Rising Income In-equality.” World Economy, 24(4): 433–52.

Borjas, George J., Richard B. Freeman, andLawrence F. Katz. 1997. “How Much DoImmigration and Trade Affect Labor MarketOutcomes?” Brookings Papers on EconomicActivity, 1: 1–90.

Davis, Donald R. 1998. “Does European Un-employment Prop up American Wages? Na-tional Labor Markets and Global Trade.”American Economic Review, 88(3): 478–94.

Davis, Donald R., and Trevor A. Reeve. 2002.“Human Capital, Unemployment and Rela-tive Wages in a Global Economy.” In Trade,Investment, Migration and Labour MarketAdjustment, ed. David Greenaway, Richard

Upward and Katharine Wakelin, 7–27.Houndmills, UK: Palgrave Macmillan.

Krugman, Paul R. 1995. “Growing WorldTrade: Causes and Consequences.” Brook-ings Papers on Economic Activity, 1: 327–77.

Meckl, Jurgen, and Benjamin Weigert. 2003.“Globalization, Technical Change and theSkill Premium: Magnification Effects fromHuman-Capital Investments.” Journal ofInternational Trade and Economic Devel-opment, 12(4): 319 –36.

Neary, J. Peter. 1985. “International FactorMobility, Minimum Wage Rates, and Factor-Price Equalization: A Synthesis.” QuarterlyJournal of Economics, 100(3): 551–70.

Oslington, Paul. 2002. “Trade, Wages and Un-employment in the Presence of Hiring andFiring Costs.” Economic Record, 78(241):195–206.

Taber, Christopher R. 2001. “The Rising Col-lege Premium in the Eighties: Return to Col-lege or Return to Unobserved Ability?”Review of Economic Studies, 68(3): 665–91.

1930 THE AMERICAN ECONOMIC REVIEW DECEMBER 2006