Articulo - Antony Venables

8/13/2019 Articulo - Antony Venables

1/15

The EconomicJournal, 113 (October),747-761. @ Royal Economic Society 2003. Published by BlackwellPublishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA02148, USA.

WINNERS AND LOSERS FROM REGIONALINTEGRATION AGREEMENTS*

AnthonyJ. VenablesHow are the benefits and costs of a customs union divided between member countries? Out-comesdependon the comparative dvantage f members,relative o eachother andrelative othe rest of the world. Countries with a comparative advantage between that of their partnersand the rest of the world do better than countries with an 'extreme' comparative advantage.Consequently, integration between low income countries tends to lead to divergence ofmember country incomes, while agreements between high income countries cause conver-gence. Results suggest that developing countries are likely to be better served by 'north-south'than by 'south-south' agreements.

How are the gains and losses associated with membership of a customs uniondivided between member countries? Do unions promote convergence of per capitaincome levels amongst member states, or divergence? The theory of economicintegration (from Viner (1950) onwards) tells us that the effects of membershipare ambiguous but gives little guidance on these questions.'The empirical literature is slightly more suggestive. For customs unionscontaining relatively high income countries there is evidence of convergence.For example, the work of Ben-David (1993, 1996) charts convergence withinthe European Union. From the late 1940s to early 1980s he finds that percapita income differences narrowed, falling by about two thirds over the per-iod, due mainly to more rapid growth of the lower income countries.2 Morerecently there has been the strong performance of Ireland, Spain and Portu-gal, which have made substantial progress in closing the gap with richermembers of the EU. Whereas in the mid 1980s these countries' per capitaincomes were, respectively, 64%, 67% and 57% of the income of the large EUcountries,3 by the end of the 1990s the numbers had risen to 113%, 80% and71%. Clearly, the prospect of convergence is motivating the queue of entrantsto the EU.For low income countries there is some evidence that the opposite process isat work, with regional integration promoting divergence. Perhaps the bestdocumented example of this is the concentration of manufacturing in the oldEast African Common Market. In the 1960s Kenya steadily enhanced its posi-tion as the industrial centre of the Common Market, producing more than

* This paper was produced as part of the Globalisation Programme of the UK Economic and SocialResearch Council funded Centre for Economic Performance, London School of Economics. Thanks toreferees and to participants in seminars at Columbia University, the LSE, the University of Sao Paolo,and the University of Sussex for helpful comments.Two recent surveys are Baldwin and Venables (1995) and Panagariya (2000).2 Differences measured by the standard deviation across countries of log per capita incomes.The average of France, Germany, Italy and the UK. All numbers in this and the next paragraph forper capitaincome PPP, from World Bank, World Development Indicators.

[ 747 ]


2/15

748 THE ECONOMIC JOURNAL [OCTOBER70% of the manufactures, exporting a growing percentage of them to its tworelatively less developed partners, and achieving faster GDP growth (Eken,1979). The Common Market collapsed in 1977, partly because of the internaltensions that this divergent performance created. More recent examples in-clude the concentration of industry and services in and around Guatemala Cityand San Salvador in the Central American Common Market, and Abidjan andDakar in the Economic Community of West Africa. El Salvador and Guatemalanow account for over 80% of manufacturing value added in the CentralAmerican Common Market, up from 68% in 1980; over the same time periodtheir per capita incomes have gone from 117% and 112% of the average forCACM to 138% and 116% respectively. In the Economic Community of WestAfrica the combined share of Cote d'Ivoire and Senegal in manufacturingvalue added has risen from 55% in 1972 to 71% in 1997, although Coted'Ivoire's income lead has narrowed.4 Understanding the effects of regionalintegration on the distribution of income in 'South-South' agreements isparticularly important given the recent rapid growth in the number of suchagreements (World Bank, 2000).

Many factors may be driving these changes but this paper concentrates juston comparative advantage and its implications for trade creation and tradediversion. We show how careful consideration of a country's comparativeadvantage - relative to the world as a whole and relative to its partners in thecustoms union (CU) - yields predictions about the winners and losers fromCU formation. It turns out that countries with 'extreme' comparative advant-age in a CU will generally be more vulnerable to trade diversion than arethose with an 'intermediate' comparative advantage. If comparative advantageis associated with per capita income (via physical or human capital endow-ments) then CU membership will lead to convergence of income levels withina union composed of high income countries, and divergence in a unioncomposed of low income members.The argument is based on the comparative advantages of member countries,relative to each other and to the rest of the world. Suppose that countries differin their endowments of skilled and unskilled labour, and that these differencesform the basis of their comparative advantage. Let us take two countries that areunskilled labour abundant relative to the rest of the world (say 'Uganda' and'Kenya'), and suppose that one of them, Uganda, is also unskilled abundantrelative to the other, Kenya. Uganda therefore has an 'extreme' comparativeadvantage, and Kenya an 'intermediate' one. What happens if these two coun-tries form CU? The comparative advantage of Kenya relative to Uganda will causeKenya to export the skilled labour intensive good (say manufactures) to Uganda,which will export the unskilled labour intensive good (agriculture) in return.The first of these flows is trade diverting: Uganda is getting its imports ofmanufactures from Kenya not from the rest of the world, in line with intra-union

4 Another good example is the divergence in economic performance between East and West Pakistanwhich was one of the factors leading to the break up of the country. See World Bank (2000) for fullerdiscussion of these cases.? Royal Economic Society 2003


3/15

2003] REGIONAL INTEGRATION AGREEMENTS 749not global comparative advantage. The second is trade creating: by increasingimports of agriculture from Uganda, Kenya is trading with the global, not justintra-union, lowest cost supplier.The general argument here is that the country with an 'intermediate' com-parative advantage will do better from the union than the one with the 'ex-treme' comparative advantage. Intuitively, interposing an intermediate countrybetween the extreme one and the rest of the world is exactly the circumstancelikely to divert the extreme country's trade. For two poor countries this unequaldivision of costs and benefits causes income divergence; the extreme country isthe one with the least skilled labour, and hence initially poorest. However, fortwo rich economies (both with above world average skilled labour abundance)the extreme country is the one with the highest skilled-unskilled labour ratio.Thus, exactly the same force that drives income divergence in a CU betweenKenya and Uganda, leads to income convergence in a CU between, say, Franceand Spain.The remainder of the paper is devoted to developing these ideas more fully.First (Section 1), we present a two-good diagrammatic analysis of the relationshipbetween comparative advantage and trade creation/ diversion. Then we developthe argument in a multi-good Ricardian trade model (Section 2). Finally (Section3), we present a simulation based exploration of a two-factor and two-sectormodel which combines a Heckscher-Ohlin structure with product differentiationby location of production (Armington). The model shows how, given theendowment of the rest of the world, the gains and costs of CU membershipdepend on each country's own endowment and that of its partner. We also usethis model to look at the question of South-South versus North-South CUs,arguing, as do Spilimbergo and Stein (1998), that the latter are likely to bepreferable for Southern countries.How does the present paper, with its focus on comparative advantage, relate toexisting literature? Much recent work analyses regional integration in models withproduct differentiation and intra-industry trade, generally abstracting from com-parative advantage. This is partly because of the intrinsic importance of someproduct market issues (eg, competition effects) and partly because these modelsprovide a tractable framework in which to analyse dynamic effects and politicaleconomy considerations (Krugman, 1993; Krishna, 1998; Baldwin, 1995). In thecompetitive equilibrium tradition, the Kemp-Wan-Ohyama (Kemp and Wan, 1976)analysis establishes sufficient conditions for gain, but does not investigate thedistribution of costs and benefits when these conditions do not hold. Much of therest of the competitive equilibrium literature is devoted to models in which a veryspecific structure of trade is assumed; for example, the three-country and three-good models (following Meade (1955) and reviewed in Lloyd (1982)), in whichintegrating countries are simply assumed to export different goods. These modelshave the drawbacks that a very large number of trade configurations are possible(Lloyd 1982), and that the failure to connect with underlying determinants ofcomparative advantage makes it impossible to ask questions such as, is a South-South agreement better than a North-South agreement? The objective of thispaper is precisely to show how analysis of countries' comparative advantage can? Royal Economic Society 2003


4/15

750 THE ECONOMIC JOURNAL [OCTOBERproduce insights on the extent to which they experience trade creation or tradediversion.5

The fundamental difficulty in the development of the literature perhaps arisesfrom the fact that, in the obvious benchmark trade model, CU formation has noeffect whatsoever. If the integrating countries are small and their pattern of tradewith a large 'rest of the world' is the same before and after formation of the CU,then prices of all goods so traded are set in the rest of the world and unchanged byformation of the union; formation of the CU then has no effect. An interestingmodel must therefore have one of the following characteristics. Either goods mustswitch source of supply, or terms of trade effects must be introduced, so that pricechanges can occur. In this paper, we develop a family of models in order to pursueboth approaches. In the diagrammatic analyses (Sections 1 and 2) CU formationhas the effect of causing changes in the sourcing of imports. In the Heckscher-Ohlin-Armington model (Section 3) product differentiation means that fixedprices of rest of the world goods are consistent with variation of the prices of goodsproduced and exported by the integrating economies. We have no completelygeneral theorem but show how consideration of countries' comparative advantage,relative to their partner and relative to the rest of the world, yields some importantinsights about the costs and benefits of CU membership, insights that are robustacross the family of models studied. The focus is on the relationships betweencomparative advantage, trade creation, and trade diversion, and in the last Sectionof the paper we link this back to income levels and the convergence-divergenceissue.

1. Internal and External Comparative Advantage: A DiagrammaticExampleThe basic argument can be made most simply - yet rigorously - through a dia-grammatic example. Figure 1 describes a world with two goods, A and M, andthree countries, a large rest of the world (country 0), and two small countries, 1and 2. The Figure has on the axes quantities of goods A and M, consumption ofwhich takes place in fixed proportions, along the consumption line illustrated.The world price of good M in terms of A is po-6Production possibilities for countries 1 and 2 are illustrated by the solid linesAIM1and A2M2.They are constructed such that both 1 and 2 have comparativeadvantage in good A relative to the rest of the world, and 2 also has a comparativeadvantage in A relative to 1. Thus, with free trade and prices po countries 1 and 2

5 A recent paper that does address this issue and makes explicit the endowments of the countries isSpilimbergo and Stein (1998). They undertake numerical analysis of a model in which there are twoidentical low income (capital-scarce) countries and two identical high-income (capital-rich) countries,investigating the effects of alternative trading arrangements. Our analysis provides for a larger set ofconfigurations of comparative advantage, permitting analysis of gains and losses within South-South andNorth-North agreements. Levy (1997) uses a factor endowment based model to analyse the politicaleconomy of CU formation but allows only for situations of completely free trade or autarky: thus,countries in a CU are assumed to have no trade with the rest of the world.

6 po is the relative price on international markets. There are no trade or transport costs, and internalprices differ from po only because of tariffs.? Royal Economic Society 2003


5/15

2003] REGIONAL INTEGRATION AGREEMENTS 751A

PO 0 T QI*+ Q2A2

,,+ C2

F2Al PO

P\T . Consumption ineF\

QCC2

SM2 MIO MFig. 1. Preferentialiberalisation

would produce at points F1 and F2. They would both export good A, country 2more than country 1, since it has the more extreme comparative advantage (likeUganda in our earlier example).The initial situation is not free trade but a position in which all imports bycountries 1 and 2 are subject to tariffs at rate T > 1.7 We set this rate sufficientlyhigh that country 1 is self sufficient at point Ci = QI, with the domestic price ofgood M in terms of good A given by the gradient of the production possibilityfrontier at this point. This price ratio lies between the domestic price ratio thatwould rule if good M were to be imported (poT), and that which would rule if goodA were to be imported (po/ 7), so confirming that trade is not profitable. Country 2has the same tariffs, but its more extreme comparative advantage means that it hassome trade in the initial situation, producing at Q2and consuming at C2.It importsgood M, meaning that the domestic price ratio is poT, at which Q2 is profit max-imising. The budget constraint holds at world prices, po, so country 2's trade vectoris Q2C2.What happens if these two countries form a customs union? In the Figure, thenew equilibrium has the CU as a whole continuing to import M from the rest ofthe world, so the internal price ratio is poT. Production then occurs at points Q*and Q. (Adding these production vectors to point Q1*+ Q confirms that the CUimports good M, since the CU's production is less M intensive than consumption.)The trade of each of the countries is as follows. Country 1 has a comparative

We use ad valorem ariff factors throughout, so 7'= 1 is free trade.@ Royal Economic Society 2003


6/15

752 THE ECONOMIC JOURNAL [OCTOBERadvantage in M relative to country 2, so trades vector Q Cj with country 2, at priceratio PoT. Correspondingly, country 2 has internal trade vector QE = Q?C1*.It also has external trade vector EC2, this being conducted at world pricepo, and equalling the external trade of the CU as a whole, vectorEC2 = (Q +? Q2) - (Cl + C;).The welfare effects of the CU are given by comparison of consumption points.Country 1 gains from the union as C* is above C1;it now has some gains fromtrade, where previously it had none. Notice that this arises despite the fact thatcountry 1's production structure has moved in the opposite direction from the wayit would go under free trade. In contrast, country 2 loses, going from G2 to C*.Thereason is trade diversion: it was getting all its imports of M from the rest of theworld and is now getting some of them from its higher cost partner. As we arguedin the introduction, the imports of the extreme country (2) are diverted to apartner country with comparative advantage between it and the rest of the world.However, for the intermediate country (1), trade with the partner and with the restof the world are less close substitutes, and therefore less vulnerable to tradediversion.

This diagrammatic analysis provides a rigorous argument but perhaps seemsrather contrived - one of the countries is in autarky in the initial situation andtrades only with its partner once the CU is in place. This reflects the problemnoted at the end of the introduction, and is why we now turn to more generalmodels.

2. A Generalised Ricardian ModelIf there are many goods with technical coefficients varying across countries, thenCU formation will generally bring both trade creation and trade diversion as goodschange source of supply. How is the distribution of these effects related to coun-tries' comparative advantage? Once again, a diagrammatic approach provides themain insights.The vertical axis of Figure 2 measures the cost of producing a good in country 2,and the horizontal the cost in country 1. The points labelled with Greek lettersrepresent goods, and their coordinates the costs of producing them in eachcountry. These costs are composed of the wage in each country, wi, times the unitlabour coefficients, bi,which varyacross goods and countries reflecting Ricardianefficiency differences. All goods have rest of the world price 1 (by choice of units)and initially face country 1 and 2 import tariffs at rate T.

Looking first at efficiency levels in country 1, we see that good a has the lowestcountry 1 unit labour requirement.8 This good will therefore be exported bycountry 1 and, since the world price of the good is unity, this sets the country 1wage at wlbl () = 1. In the initial situation where all imports bear tariff T, country1 is self sufficient in goods /, y, and Esince domestic costs (wlbl) are less than theprivate unit costs of import (= T) and greater than unit export receipts (= 1).Goods 6 and ( are imported from the rest of the world, since for each of them

8 Is furthest to the left on the horizontal axis.? Royal Economic Society 2003


7/15

2003] REGIONAL INTEGRATION AGREEMENTS 753

w2b2 w1b1= w2b2

T -- --~f~------- .I-I

1 T wlb1SC:Countryrade DI:ountryradecreation: diversion:Fig. 2. Sourcingof Manufactures

wlb > T. The analogous configuration for country 2 can be read off the verticalaxis. Good # has the lowest unit labour requirement, so the country 2 wage is set bywb,2(fl) = 1. Country 2 is self sufficient in goods 6 and e(1 < w2b2< T) and im-ports c, y, and C(W2b,2 T).Formation of a CU will change the pattern of trade in some goods and notothers. Wages in both economies remain constant, because they continue tosupply their respective export goods to the rest of the world. However, country 1will now buy from country 2 any good for which w2b2< wlba (below the 450 line)and zwbb2< T (so it is cheaper to import duty free from the partner than importfrom rest of the world). As illustrated, this includes two goods. Good 6 goes frombeing imported from the rest of the world to being imported from the partnercountry; this is trade diversion with additional cost per unit of w2b(6) - 1. Good #goes from country 1 self sufficiency to being imported from 2; this is trade cre-ation, with cost saving per unit of wlbl(fl) - W2b2(f).Analogously, country 2 now imports from 1 any good for which wlbl < w2b2andw1bl< T. Good y therefore experiences trade diversion, now being supplied bycountry 1 instead of the rest of the world (since T > wabl(y) > 1). Good Cgoes frombeing locally produced in 2 to imported from 1, and this is trade creation, sincewlbl(e) < w2b2(e),bringing unit cost saving equal to this cost difference.These effects are summarised in Table 1 (in which 0 denotes rest of the world),and the regions of product space within which country 1 experiences trade cre-ation and diversion are illustrated by the shaded areas on Figure 2; (analogouscountry 2 zones are not marked).@ Royal Economic Society 2003


8/15

754 THE ECONOMIC JOURNAL [OCTOBERTable 1

The Direction of TradeInitial CU Welfare Change

Country 1 Country 2 Country 1 Country 2 Country 1 Country 2c Exp. to 0 Imp. from 0 Exp. to 0 Imp. from 0fl No trade Exp. to 0 Imp. from 2 Exp. to 0 & 1 1, TCy No trade Imp. from 0 Exp. to 2 Imp. from 1 L,TD6 Imp. from 0 No trade Imp. from 2 Exp. to 1 , TDE No trade No trade Exp. to 2 Imp. from 1 1, TCC Imp. from 0 Imp. from 0 Imp. from 0 Imp. from 0

This approach can be generalised to arbitrarily many commodities. To calculatethe overall welfare effect we then need to know the distribution of commoditiesover the space and the consumption levels of each, before and after CU formation.If products are indexed Z and consumed in fixed quantity c(Z) then the change incountry 1 welfare is simply

Awl = E[1 - w2b(Z)]c(Z) [wib,(Z) - wzb2(x)]c(X)XED1 ZECIwhere D1and C1are the sets in which there is trade diversion and trade creation.Can we now link this to our discussion of countries' comparative advantagerelative to each other and relative to the rest of the world? Suppose that the setof products that exist are uniformly distributed within the ellipse shape area onFigure 2. Then, intuitively, country 1 is 'more like' the rest of the world than iscountry 2. Country l's production costs relative to the rest of the world (withunit cost = 1) vary at most by an amount equal to the width of the ellipse, andon average by half of this; in contrast, country 2's production costs varyaccording to the height of the ellipse. More precisely, country 1 has comparativedisadvantage relative to the world but comparative advantage relative to country 2(1 < wlbl < w2b2), n all products in the ellipse and above the 450 line. Thus, forthis majority of commodities, it lies 'between' country 2 and the rest of theworld.

Comparing the shape of the ellipse with the regions of trade creation anddiversion completes the argument. As illustrated, a low proportion of country I'sgoods change source of supply (the intersection of the ellipse and the shadedareas) and for most of those that do, this is trade creation. For country 2, a muchhigher proportion of goods change source of supply, and most of these changesare trade diversion. Thus, this multi-commodity framework confirms our earlierfinding. The 'extreme' country does worse than the 'intermediate' one.A formal model of this case can be easily developed if the set of products isrestricted to lie on a line in blbspace, as is usual in such a model, e.g. Dornbuschet al. (1977). Results confirm the conclusions drawn from Figure 2.9

9 Such a model is developed in Venables (2000).? Royal Economic Society 2003


9/15

2003] REGIONAL INTEGRATION AGREEMENTS 7553. Income Divergence and Convergence: A Heckscher-Ohlin-ArmingtonModelWe now turn to a variant of the standard trade model in which comparativeadvantage arises from differences in factor endowments. We add to the standardHeckscher-Ohlin model an assumption of product differentiation at the nationallevel, in order to maintain non-specialisation of production and to allow outputprices to change rather than being set by supply of homogeneous products fromthe rest of the world. Analysis of this model requires numerical simulation,although most of the intuition comes from Heckscher-Ohlin.The model structure is as follows. There are two factors of production, skilledand unskilled labour (Sand U), and two sectors, differing in factor intensity. Thereare three countries, one of which, the rest of the world, is large and is endowedwith equal quantities of the two factors. The other countries, 1 and 2, have factorendowment ratios different from each other and from the rest of the world, andthese differences are the basis of their comparative advantage. All countries havethe same technology and preferences, although we assume some national ('Arm-ington') product differentiation. Thus, products in each sector are differentiatedby location of production, although the amount of differentiation is set at minimallevels - the elasticity of substitution between different countries' products in eachsector is 50 in the examples that follow. For ease of interpretation we impose asymmetric structure on production and consumption, assuming that consumerexpenditure is equally divided between sectors, and that the factor intensity in oneindustry is the reciprocal of that in the other industry (using Cobb-Douglas tech-nologies, see Appendix for details).Because of the symmetry that is built into the model the equilibrium price ratioof output produced in the rest of the world is unity; this world price ratio is heldconstant in all experiments that follow. In the initial equilibrium all imports facethe same tariff rate (set at 30%). The internal price ratios and trade patterns ofcountries 1 and 2 reflect these tariffs and each country's factor abundance. Theexperiment we study is the removal of the tariff between countries 1 and 2, and weshow how outcomes depend on their endowments, relative to each other and tothe rest of the world.

Results are illustrated on Figures 3-5, which give contours of welfare change as afunction of the factor endowments of countries 1 and 2. Axes measure the factorabundance ratios of each country, and in Figures 3 and 4 are constructed withSi + Ui= 1, i = 1,2. Each country's factor abundance relative to the rest of theworld depends on whether Si/ Ui is greater or less than unity, while intra-unioncomparative advantage is measured relative to the 450 line, above which country 1is S abundant relative to country 2.We look first at relative utility changes, by mapping on Figure 3 the contours ofthe proportionate change in utility in country 2 minus that in country 1. The boldstraight lines marked 00 are the zero contours, and the surface is saddle shaped,with regions in which country 2 does better than country 1 marked +. Consider thecase in which both countries' endowment ratios are on the same side of those ofthe rest of the world. In the bottom left quadrant both countries are unskilled? Royal Economic Society 2003


10/15

756 THE ECONOMIC JOURNAL [OCTOBERSi/U

(log scale) 0 + 0

2.0

,+

0.5

0 00.5 1 2.0 S2/U2(log scale)

Fig. 3. Relative Changesin Welfare country 2 minus country 1)

labour abundant relative to the world as a whole. Then country 2 loses relative tocountry 1 if and only if it is the more unskilled labour abundant of the twoeconomies, i.e. at a point such as A, above the diagonal, with S1/U1 > S2/U2.Conversely, in the upper right quadrant both countries are skilled labour abun-dant (relative to the world) and the more skilled labour abundant country suffersthe relative loss - at point B this is country 2. Both these cases illustrate that, as withthe previous models, the 'extreme' country fares relatively badly, the reason beingthat its imports are diverted to the 'intermediate' country in line with intra-CUcomparative advantage but not global comparative advantage.In the upper left and lower right quadrants countries 1 and 2 lie on oppositesides of the rest of the world endowment ratio. Along the downward sloping line00 countries 1 and 2 are symmetric, so experience the same welfare change.'0 Offthis line the country that has endowment closer to the world average is the relativegainer. Thus, at point C country 2 is the relative loser, since S2/ U2 is further fromunity (horizontal distance) than is S1/U1i vertical distance). The intuitive reason isthat although country 2's union partner is S abundant relative to the world, the CUas a whole is S scarce. Trade diversion is experienced by country 2 as it switches itssourcing of the S intensive good from the rest of the world to the CU.How general are these results? Providing country differences arise only fromendowments of two factors, then the surface is quite generally a saddle with a zero-contour on the 45% line from the origin, along which the two economies are

1 Country 1's endowment ratio is the reciprocal of country 2's.? Royal Economic Society 2003


11/15

2003] REGIONAL INTEGRATION AGREEMENTS 757identical. However, the second zero-contour line need not be straight, nornecessarily downwards sloping everywhere; its linearity in Figure 3 is a conse-quence of the symmetry built into the model.The symmetry of Figure 3 derives from the fact that it gives utility change incountry 2 minus that in country 1. Figure 4 focuses just on country 2 welfarechange, expressed as a proportion of 2's initial welfare. This, like the relativechange of the previous Figure, forms a saddle on endowment space. The linesmarked 00 are zero contours, and the plus and minus signs indicate regions ofcountry 2 gain and loss. There are verysmall gains along the 450 line, arising fromthe product differentiation in the model. However, the main message of the Figureis that the relative losses we saw in Figure 3 can also be associated with absolutelosses. These regions of absolute loss are subsets of regions in which country 2suffers relative loss (Figure 3), and the intuition is as before. Thus, at the left andright hand sides of the Figure country 2 has 'extreme' endowments, and tradediversion causes it to lose from CU membership.In Figures 3 and 4 the factors Sand Uenter the model symmetrically,so to refer tothem as skilled and unskilled labour is a misnomer - the wage of S is on average nohigher than that of Uand countries with much Sare on average no richer than thosewith much U. To capture the idea that S abundant economies are relatively highincome we now modify the analysis in the following way. In Figure 3 and 4, if aneconomy gained a unit of Sit lost a unit of U(since Si + Ui 1). Now, in Figure 5, wehold U constant (= U), and simply varythe amount of S. Perhaps the best way tothink of this is that countries 1 and 2 both have U workers, a fraction Si/s of whom

Sl/U1(log scale)

2.0

A

0.50 \0.5 1 2.0 S2/U2(log scale)

Fig. 4. Country2 WelfareChange Contours? Royal Economic Society 2003


12/15

758 THE ECONOMIC JOURNAL [OCTOBERare endowed with one unit of unskilled labour and s units of skilled labour, while thefraction 1 - Si/s have no skill. Thus, at a high value of Si/ Uia higher proportion ofthe population are endowed with skilled labour. Sand Uenter production as beforebut S abundant economies will now tend to be richer. For example, moving fromSi/ Ui = 0.5 to Si/ Ui = 2 in the example of Figure 5 raises the proportion of thepopulation with skills and approximately doubles country i per capita income.Contours in Figure 5 illustrate, like Figure 4, the country 2 welfare change dueto CU formation. Two main messages come from the Figure. The first is theoriginal argument, that CU formation between two poor countries tends to lead toincome divergence and between rich countries leads to convergence. Considerpoint A. At this point country 2 is poorer than country 1 (it is S scarce relative to itspartner), and suffers a welfare reduction, while country 1 experiences a welfaregain, causing divergence. (The country 1 gain is not illustrated directly, but can beseen by reversing country labels and looking at point A', the reflection of A aroundthe 450 line). Conversely, at point B both countries are S abundant, but country 2relatively more so, and therefore relatively rich. It is now country 2 that loses andcountry 1 that gains, causing convergence of their real incomes.The second point concerns the attractiveness of 'North-South' agreements forlow income countries. Let us take a fixed and low value of S2/ U2,and ask: what typeof partner is country 2 best off forming a CU with? The answer is a skilled labourabundant economy (high S1/ UI). There are two forces driving this. One is that tradecreation is maximised and trade diversion minimised with such a partner (as in

S1/U1 ....log scale) 02.0

A0.5

A' +0

0.5 1 2.0 S2/U2(log scale)Fig. 5. Country2 WelfareChange Contours(Ui = constant)

? Royal Economic Society 2003


13/15


14/15

760 THE ECONOMIC JOURNAL [OCTOBERAppendixThere are two goods, x and y, (indicated by superscripts), two countries 1 and 2 (indicatedby subscripts) and the rest of the world (indicated by subscript 0). Factor endowments are Siand Uj with respective prices vi and wi. Technologies are described by cost functions,

c =wv , c = w v , = 0.25. (Al)Factor market clearing takes the formdcx dOc OcX OcYSi = q + qY, Ui - + ,i qi , i=1,2, (A2)

where qk denotes the quantity of good k produced in country i.Preferences are described bymi = ui(GxG) /2, (A3)

where mi is income, ui is utility, and Gf is the price index of good k in country i, defined byGik [(pik),-

+ (tpk 1-1 + T- /-,i,j = 1, 2, i j. (A4)where pk denotes the price of good k produced in country i, equal to unit cost, t denotes theinternal tariff and T the external tariff. a is set at 50, and t and Tboth take initial value 1.3, tdropping to 1 when the customs union is formed. Demands are derived from utilitymaximisation, and income is given by

mi= wiUi+viSi+p qX(t - 1)+ pq(t- 1) + qoXi(T-1)+qoi(T- 1), ij=1,2, (A5)where qij denotes the quantity of good k produced in j and sold in i. In addition, country 0has demands qok which have price elasticity a and are scaled such that in the initialequilibrium an average of 10% of the output of countries 1 and 2 are exported to country 0.In Figures 3 and 4 endowments vary in the interval Si = [0.25, 0.75] with Ui = 1 - Si.In Figure 5 endowments vary in the interval Si = [0.1667, 1.5] with Ui = 0.5.

ReferencesBaldwin, R. (1995). 'A domino theory of regionalism', in (R. Baldwin, P. Haaparanta and J. Kiander,eds.), ExpandingMembershipf theEuropeanUnion,Cambridge: Cambridge University Press.Baldwin, R. and Venables, A. J. (1995). 'Regional economic integration', in (G. Grossman andK. Rogoff, eds.), Handbookof InternationalEconomics, ol. 3, Amsterdam: North Holland.Ben-David, D. (1993). 'Equalizing exchange; trade liberalization and income convergence', Quarterly

Journal ofEconomics, ol. 108, pp. 653-79.Ben-David, D. (1996). 'Trade and convergence among nations', Journal ofInternationalEconomics,ol. 40,pp. 279-98.Dornbusch, R., Fischer, S. and Samuelson, P. A. (1977). 'Comparative advantage, trade and payments ina Ricardian model with a continuum of products', AmericanEconomicReview,vol. 67, pp. 823-39.Eken, S. (1979). 'Breakup of the EastAfrican Community', Financeand Development,ol. 16, pp. 36-40.Kemp, M. and Wan, H. (1976). 'An elementary proposition concerning the formation of customsunions', Journalof InternationalEconomics, ol. 6, pp. 95-7.Krishna, P. (1998). 'Regionalism and multilateral; a political economy approach', Quarterlyournal ofEconomics, ol. 113, pp. 227-51.Krugman, P. R. (1993). 'Regionalism versus multilateralism; analytical notes', in (J. de Melo andA. Panagariya,eds.), NewDimensions n Regional ntegration,Cambridge: Cambridge UniversityPress.Levy, P.I. (1997). 'A political economic analysis of free-trade agreements', AmericanEconomicReview,vol. 87, pp. 506-19.Lloyd, P.J. (1982). '3 x 3 theory of customs unions', JournalofInternationalEconomics, ol. 12, pp. 41-63.Meade, J. E. (1955). The Theory f CustomsUnions,Amsterdam: North-Holland.Panagariya,A. (2000). 'Preferential trade liberalization; the traditional theory and new developments',JournalofEconomicLiterature, ol. 2, pp. 287-331.

@ Royal Economic Society 2003


15/15

2003] REGIONAL INTEGRATION AGREEMENTS 761Puga, D. and Venables, A.J. (1997). 'Preferential trading arrangements and industrial location',Journalof InternationalEconomics, ol. 43, pp. 347-68.Puga, D. and Venables, A. J. (1998). 'Trading arrangements and industrial development', WorldBankEconomicReview, ol. 12, pp. 221-49.Spilimbergo, A. and Stein, E. (1998). 'The welfare implications of trading blocs among countries withdifferent endowments', in (J.Frankel, ed.) 'TheRegionalization ftheWorldEconomy',Chicago: NBERand Chicago University Press.Venables, A. J. (2000). 'Winners and losers from regional integration agreements', CEPR DiscussionPaper no 2528.Viner, J. (1950). TheCustomsUnionIssue,New York:Camegie Endownment for International Press.World Bank (2000). TradeBlocs,Policy Research Report, Washington DC.

? Royal Economic Society 2003

Articulo - Antony Venables

Documents