Can bilateralism ease the pains of multilateral trade liberalization?

*Corresponding author.E-mail addresses: [email protected] (O. Cadot), [email protected] (J. de Melo),

[email protected] (M. Olarreaga).

European Economic Review 45 (2001) 27}44

Can bilateralism ease the pains ofmultilateral trade liberalization?

Olivier Cadot!,#, Jaime de Melo",#,*, Marcelo Olarreaga#,$

!INSEAD, bd de Constance, 77305 Fontainebleau, France"University of Geneva, De&partement d+e&conomie politique, 102 bd Carl Vogt,

1211 Geneva 4, Switzerland#Centre for Economic Policy and Research, London, UK

$The World Bank, 1818, H. st. NW, Washington, DC 20433, USA

Received 1 April 1998; accepted 1 April 1999

Abstract

Using the in#uence-driven approach to endogenous trade-policy determination, weshow how a free-trade agreement (FTA) with rules of origin can work as a device tocompensate losers from trade liberalization. The FTA constructed in this paper ischaracterized by external tari! structures that are negatively correlated across membercountries, ensuring e$ciency gains and, through reduced average protection, compatibil-ity with the multilateral trading system's requirements. It is also politically viable, and wedemonstrate that, in the countries concerned, governments are willing to include itsformation in the political agenda in spite of the fact that, in equilibrium, politicalcontributions from producer lobbies decline after the agreement. ( 2001 ElsevierScience B.V. All rights reserved.

JEL classixcation: F11; F13; F15

Keywords: Free-trade areas; Political economy

0014-2921/01/$ - see front matter ( 2001 Elsevier Science B.V. All rights reserved.PII: S 0 0 1 4 - 2 9 2 1 ( 9 9 ) 0 0 0 3 3 - 1

1Speech delivered at the annual dinner of the North Sta!ordshire Chamber of Commerce atStoke, December 15th 1897, and reported in Gi!en (1904).

During the American Civil War a friend of mine, who was a careful studentof the military operations, used often to remark that the war was one inwhich the South had all the victories, and the North, all the substantialsuccesses. 2. Protectionists do score many victories, or so-called vic-tories. What I wish to point out is the continual and substantial success ofFree Trade throughout the world.

Sir Robert Gi!en1

1. Introduction

Regionalism is mushrooming and appears to be here to stay. The "rst wave,which appeared in the 1960s, was made up primarily of North}North andSouth}South regional trading arrangements (RIAs), the latter being generallyrecognized to have been less successful than the former. Di!erent, primarilyNorth}South arrangements appeared during the 1980s in a second wave whichsparked a vast literature on the role of regionalism in a multilateral world. Thisliterature, which was recently surveyed by Winters (1996) (see also the editedvolumes by Anderson and Blackhurst, 1993; de Melo and Panagariya, 1993;Bhagwati and Panagariya, 1996), has largely focused on two issues: (i) Shoulda division of the world economy into regional trading blocs be expected to raiseor to lower welfare? (ii) Does regionalism help or hamper multilateral e!orts attrade liberalisation?

In most of the literature addressing issue (i), the common rationale forregional and multilateral trading arrangements is to internalize terms-of-tradeexternalities, as in Krugman (1991). This approach is probably best suited toanalyze the "rst wave of regionalism, although it is not clear that optimal-tari!considerations weigh heavily in trade-policy decisions. In addressing issue (ii),the arguments have instead been couched in a political-economy setting empha-sizing institutional di!erences between various types of regional arrangements.On the negative side, Levy (1997) uses a political-economy approach a la Mayer(1984) to argue that RIAs may jeopardize the multilateral trading system if theyo!er disproportionately large gains to agents in integrating countries, raisingtheir reservation utility over the multilateral free-trade level. Using the politi-cal-support approach to endogenous policy formation (but di!erent modellingassumptions) Krishna (1998) and Grossman and Helpman (1995) show thatFTAs are more likely to be adopted if they are trade diverting, suggesting thatpolitically successful FTAs are likely to be of the harmful type. On the positiveside, Ethier (1998) interprets the new wave of regionalism as an endogenousresponse to the development of the multilateral system, suggesting that regionalintegration can spur multilateral liberalization by facilitating coordination.

28 O. Cadot et al. / European Economic Review 45 (2001) 27}44

2A &pressured stance' is de"ned as a decision that the government takes partly in response to o!ersof support by interest groups.

Hillman et al. (1995) also argue that governments exchange market access in theWTO or in RIAs because the mutual political gains conferred by reciprocitylower the political costs of trade liberalization.

Although the weight of theoretical conjectures suggests that there is someground for concern about the spread of regionalism, it is fair to say that thedebate is still open, if for no other reason because the historical evidence is itselfambiguous (see Irwin, 1993). This paper takes the view that whether or notregionalism should be a source of concern, it is probably here to stay; therefore,in the words of de Melo and Panagariya (1993, p. 9), a more constructiveapproach is to ask whether mechanisms can be devised to ensure that regional-ism complements multilateralism and does not fragment the world into inward-looking blocks. In this spirit, we show how the external tari! structure ofa free-trade agreement (FTA) can be designed so as to generate welfare gainswhile preserving the ability of member countries to protect import-competinginterests from the adverse consequences of trade liberalization, thus making theagreement politically viable.

An FTA is politically viable if it satis"es two criteria. Following Grossmanand Helpman (1995), we "rst assume that the formation of an FTA is on thepolitical agenda and show that the agreement under consideration is politicallyviable using an extension of their notion of a &pressured stance'.2 We also verifythat governments are willing to put the formation of an FTA on the politicalagenda: that is, that any decline in equilibrium contributions after the agree-ment's signature would be o!set by e$ciency gains. Using this two-step ap-proach, we show (by construction) how an FTA with rules of origin (rules oforigin prevent the trans-shipment of goods originating from the rest of the worldthrough the area's internal borders) can enable both countries to cooperatetowards a selective reduction of external tari!s while simultaneously maintain-ing producer prices in import-competing sectors at their initial level, a keycondition for continued political support.

An example will illustrate the gist of the argument. Consider two countries (Aand B) whose textiles and electronics producers compete with imports from therest of the world, and suppose that each country is too small in each sector tomeet its partner's import demand at the initial (tari!-ridden) domestic prices.Let the two countries form an FTA with rules of origin in which A maintains anunchanged external tari! on textiles but eliminates protection in electronics,while B does the reverse. Free trade within the area enables B's textiles pro-ducers to ship their output tari!-free to A, enjoying A's protection; likewise, A'selectronics producers can export to B and enjoy B's protection. Under suitabledemand conditions, producer prices are unchanged, so that import-competing

O. Cadot et al. / European Economic Review 45 (2001) 27}44 29

3MFN (Most-Favored-Nation) tari!s are non-discriminatory tari!s.4The symmetry assumption is discussed in Section 4.

interests do not oppose the FTA. Moreover, as we show that equilibriumpost-FTA contributions (based on external tari!s) are inferior to pre-FTA contributions (based on MFN tari!s)3 by no more than the valuation ofthe FTA's e$ciency gains in the objective function of governments, the latterhave no reason to oppose placing the formation of the FTA on the politicalagenda.

The arrangement's most interesting aspect is that it leads to reduced consumerprices, as B's consumers can now obtain electronics at their world price, whileA's consumers can obtain textiles at their world price. (Consumer-price di!er-ences between A and B cannot be arbitraged because of rules of origin.)Although tari! revenue shrinks in both countries, we show that the consumer-price reductions raise welfare in A and B; with no opposition from import-competing lobbies, these welfare gains are su$cient to make the agreementpolitically viable.

The remainder of the paper is organized as follows. Section 2 introducesthe model used in Section 3 to establish an FTA that is politically viableand globally welfare enhancing. Section 4 introduces extensions (no compensa-tion, asymmetric partners), discusses consistency with GATT article XXIV andgives prima facie evidence for the Canada}US FTA (CUSFTA). Section 5concludes.

2. The model

Two small (price-taking) and symmetric4 countries, A and B, produce andconsume three goods, labelled 1, 2 and 3. Consumers have identical, quasilinearpreferences. Goods 1 and 2 are made with sector-speci"c capital and inter-sectorally mobile labor. The presence of a "xed endowment of speci"c capitalgenerates diminishing returns in Sectors 1 and 2, whereas Sector 3 employs onlylabor under constant returns to scale. Thus, the productivity of labor in Sector3 determines the economywide wage rate, as in Grossman and Helpman (1994).Both countries import Goods 1 and 2 from the rest of the world and exportGood 3 to it, with no transportation costs between any pair of countries. Good3 serves as numeH raire.

We model the formation of an FTA in a two-period framework. This inter-temporal aspect is not introduced for its own sake, but only to highlight animportant distinction between two types of political contributions. In Period 1,each country's trade policy is nondiscriminatory and is determined as theoutcome of a domestic political game a% la Grossman}Helpman, in which


5 In each country, industry lobbies simultaneously face their government with contributionfunctions conditioned on the vector of domestic prices. Those functions satisfy Bernheim andWhinston's (1986) &truthfulness' property by which, in a neighborhood of the equilibrium, thederivatives (with respect to domestic prices) of the industry's contributions and pro"t functions areequal. The resulting &truthful Nash equilibrium' is unique and coalition proof.

6Lobbies consider that their share of aggregate consumer surplus is of negligible magnitudecompared to their share of aggregate producer surplus. Therefore, they do not internalize theexternality that their own protectionist demands exert on consumer expenditure, nor do they lobbyagainst the demands of other sectors.

lobbies representing owners of speci"c capital bid for protection with truthfulcontribution schedules.5 We will call these contributions &Type-1' contributions.At the end of Period 1, governments decide whether or not to put the formationof an FTA on the political agenda in A and B; if they do, lobbies contribute, onceand for all, in favor of the agreement or against it. We will call these one-o!contributions &Type-2' contributions. If the agreement is adopted, the govern-ments of A and B seek again Type-1 contributions in Period 2, but nowconditioned on external rather than MFN tari!s, and they act cooperatively,maximizing a joint objective function. If the agreement is rejected, Period 2 isidentical to Period 1. The political viability of the FTA, if it is on the politicalagenda, depends on the welfare gains it is able to generate and on the net sum ofaggregate Type-2 contributions. Whether or not governments are willing to putit on the agenda depends not only on Type-2 contributions and welfare e!ects,but also on a comparison of the level of Type-1 contributions before and afterthe agreement.

With a "nite time horizon and no links between periods in either demand orsupply functions, the intertemporal aspects of the game can be disregarded andthe equilibrium calculated separately for each period. Accordingly, two prob-lems must be considered: one in which governments use MFN tari!s, character-ized below, and one in which they use external tari!s in an FTA, solved inSection 3.

To rule out unnecessary complications, as in Cadot et al. (1999), we assumethat capital ownership is su$ciently concentrated to ensure that industrylobbies disregard the e!ect of trade protection on the cost of living.6 Sector 3 isnot allowed to lobby for export subsidies. Normalizing world prices to unity inall sectors and letting ti

kbe country i's tari! in sector k, the domestic price of

good k in i is pik"1#ti

k(consumer and producer prices are equivalent). We will

denote by yik(pi

k) the domestic supply of good k in country i (as a function of its

domestic price pik) and by ci

k(pi

k) its domestic consumption; thus, imports are

given by mik"ci

k!yi

k. Type-1 contributions Ci

kfrom producers of good

k (k"1, 2) in i can be written as Cik(tik, bi

k) for some constants bi

k, whose

determination is discussed in Section 3. Expressed as a function of domestictari!s, welfare is =i(ti

1, ti

2); it can be equivalently written as a function of


domestic prices when world prices are "xed. Using vector notation, ti"(ti1, ti

2),

bi"(bi1, bi

2), and the government's problem in country i can be written as

maxti

<i(t i, bi),Ci1(ti1, bi

1)#Ci

2(ti2, bi

2)#a=i(ti) (1)

for some constant a. Let tM i be the solution to Problem (1); we use bars toindicate variables evaluated at their Period-1 (pre-FTA) equilibrium values. Letalso ni

k(pi

k) be aggregate producer pro"ts in sector k and country i, and note that,

by Hotelling's Lemma, Lnik/Lpi

k"yi

k. As truthful contribution schedules satisfy

LCik/Lti

k"Lni

k/Lti

k,Lni

k/Lpi

k, it follows that LCi

k/Lti

k"yi

k. Moreover, it is shown

in Cadot et al. (1999) that, under quasilinear preferences, L=i/Ltik"(mi

k)@ti

k,

where (mik)@ is the own-price derivative of country i's import-demand function in

sector k. Thus, ignoring corner solutions at free trade (import subsidies are notallowed) tM i satis"es the "rst-order condition

L<i

Ltik

"

Lnik

Ltik

#aL=i

Ltik

"yik#a(mi

k)@ti

k"0. (2)

We need to impose the technical condition that at the Period-1 equilibrium levelof tari!s tM i, the combined output of A and B does not exceed the demand fromeach one of them individually. Formally, let tI i

kbe de"ned by

yAk(1#tI i

k)#yB

k(1#tI i

k),ci

k(1#tI i

k) where the terms in parentheses designate

arguments of the supply and demand functions. The condition can be statedconcisely as:

Assumption 1. tM ik4tI i

kfor k"1, 2 and i"A,B.

Given that A and B are symmetric, Assumption 1 means that in Period-1(pre-FTA) equilibrium, import-penetration ratios in Sectors 1 and 2 are at leastequal to one half (as yA

k#yB

k"2yi

k4ci

kimplies that mi

k/ci

k51/2).

3. Compensating losers in an FTA

Before we turn to the formal analysis of the agreement's political viability, weneed to state a technical result "rst established by Richardson (1993); namely, thatan FTA introduces a wedge between the consumer and producer prices ofimported goods, provided that rules of origin prevent consumer arbitrage. (Wede"ne rules of origin as regulations preventing the transshipment of goods fromthe rest of the world through the area's internal borders.) More precisely, in anFTA with rules of origin, producer prices are necessarily equalized across the area,


7The reader may wonder why symmetric countries would ever want to set di!erent tari!s. In theinitial (Period-1) equilibrium, no strategic interaction exists between A and B; in such a context,symmetry implies that they must set equal tari!s. But in a strategic context like an FTA, asymmetrictari!s can emerge both in cooperative and noncooperative settings (see, e.g. Cadot et al., 1996).

8As A and B set their external tari!s separately, the agreement is not quite a customs union, sincethe latter would require them to agree on a common external tari!. On the other hand, as A andB agree to set external tari!s cooperatively, formally the agreement involves more than an FTA; infact, some enforcement mechanism, such as supergame strategies, is formally necessary to implementthe agreement. As the necessary conditions for Folk Theorem-type results to hold are well-known,we do not discuss them here. But we showed in Cadot et al. (1999) that a very similar external-tari!structure emerges as the unique equilibrium of a non-cooperative game between member countriesunder a condition similar to Assumption 1.

whereas consumer prices are not.7 We will denote producer prices by pik

andconsumer prices by qi

k; let also t"(tA

1, tA

2, tB

1, tB

2) be the the area's external tari!

vector.

Lemma 1 (Richardson, 1993). In an FTA with rules of origin and satisfyingAssumption 1, the consumer price of good k is qi

k"1#ti

kin country i, but its

producer price is pik"1#maxMtA

k, tB

kN in the entire area.

Proof. If tAk"tB

k, the lemma holds trivially; suppose that tA

kOtB

k, and label

countries so that tAk'tB

k. As transportation costs are nil between A and B, B's

producers sell in A; as tAk4tI A

k, they can sell their entire output without depress-

ing k's market price in A. Thus, the price relevant to B's producers is not 1#tBk,

but 1#tAk: producer prices are equalized at pB

k"pA

k"1#tA

k. As B's output is

shipped entirely to A, B's consumption is met by imports from the rest of theworld priced at qB

k"1#tB

k. In A, rules of origin prevent consumers from

purchasing quantities of good k imported by B from the rest of the world;consequently, qA

k"1#tA

k("pA

k). Thus, qB

k"1#tB

k(qA

k"1#tA

k. h

Lemma 1 highlights the key role played by rules of origin in an FTA. Togetherwith Assumption 1, rules of origin ensure that the price of k in the more protectedmarket (A) is sheltered from downward pressure arising from the relatively lowprice prevailing in the less-protected market (B). As a result, if B forms an FTAwith A while simultaneously reducing its external tari! on k, its producers canavoid being exposed to the competitive pressure of imports from the rest of theworld by simply diverting their output to A's protected market. Stated di!erently,rules of origin enable A to &export its protection' to B, as Krueger (1993) "rst noted(although Krueger was referring to the protection extended to intermediate-goods producers when "nal goods are subject to rules of origin in an FTA). Thisis the mechanism on which we rely to protect losers from trade liberalization.

Suppose now that the formation of an FTA is on the political agenda, andconsider the following proposal. Barriers to intra-bloc trade (i.e. to tradebetween A and B) are, of course, to be removed. But in addition, A and B agreeto set their external tari!s cooperatively:8 A maintains an external tari! on


9Because the opportunity to form an FTA occurs only once in our two-period framework,participating governments are necessarily committed to staying in the FTA once it is formed. In anin"nite-horizon model, some mechanism } like an entry-and-exit cost } would be needed to preventgovernments from renegotiating the participation issue with their lobbies. We are grateful to RaquelFernandez for attracting our attention to this issue.

Good 1 just equal to its pre-FTA MFN tari! but eliminates its external tari! onGood 2, while B does the reverse. Loosely speaking, an FTA will be said to bepolitically viable if the sum across partners of contributions for and against theFTA and economic e$ciency (properly weighted) is positive. More precisely, thefollowing must hold. In each country, lobbies o!er Type-2 contributions for oragainst the FTA to their government.9 Grossman and Helpman (1995), whopioneered this approach, showed that truthful contributions are in this contextsimple amounts related to the loss or gain that the lobbies expect from the FTA.Thus, if lobby k expects to gain from the agreement, it will o!er a contribution04Fi

k4Dni

k(the change in industry pro"ts attributable to the FTA) in its

favor; if it expects to lose, it will o!er a contribution 04Nik4!Dni

kagainst it.

We will call Fi the set of sectors such that Dnik50 and Ni the set of sectors such

that Dnik(0. If the FTA is implemented, in equilibrium the government of each

country must be just indi!erent between implementing it or not, otherwiselobbies on the winning side could reduce their contributions without altering theoutcome. Thus, lobbies on the losing side (those who oppose the agreement)must bid the whole valuation of their expected losses, implying that Ni

k"!Dni

kfor all k3Ni.

In this section, we assume that A and B can use inter-country transfers tocompensate for tari!-revenue losses due to the agreement; Section 4 shows howthe FTA can be implemented without such transfers. Let D=i denote the welfarechange attributable to the agreement.

Dexnition 1. An FTA is politically viable with compensatory transfers if thereexist contributions Fi

k3[0;Dni

k] such that

+i/A,B

A +k|Fi

Fik! +

k|Ni

Dnik#a D=iB50.

De"nition 1 extends Grossman and Helpman's de"nition of a &pressuredstance' to a context where compensatory transfers are possible; we use it in thefollowing proposition to construct an external tari! structure that ensures theFTA's politically viability and maximizes the sum of the member countries'domestic objective functions <A and <B, given truthful contribution schedules.Let tK"(tM A

1, 0, 0, tM B

2); that is, tK A

1"tM A

1(A's initial MFN tari! in sector one); tK A

2"0;

tK B1"0; tK B

2"tM B

2(B's initial MFN tari! in sector two).


Proposition 1. An FTA with rules of origin and with an external tariw vector equalto tK is politically viable with compensatory transfers.

Proof. By symmetry, the initial (Period-1) tari!s of A and B on k are identical:tM Ak"tM B

kfor k"1, 2; consequently, p6 A

k"p6 B

k. As tK A

1'tK B

1"0, B's output of good

1 is sold in A; applying Lemma 1, p( A1"p( B

1"1#tK A

1. Moreover, as tK A

1"tM A

1, it

follows that p( A1"p6 A

1"p( B

1"p6 B

1. By the same argument, p( A

2"p( B

2"p6 A

2"p6 B

2.

Thus, producer prices are unchanged in both countries, implying thatFik"Ni

k"0 for all k and i. If follows that the agreement is politically viable if

and only if it raises the area's aggregate welfare.As producer prices are unchanged, so is producer surplus. Therefore the

variation in welfare is the sum of the changes in consumer surplus and tari!revenue. Consider country A. The change in A's consumer surplus is tM A

2c6 A2#dA

2,

where dA2'0 is the deadweight loss induced in A by a tari! tM A

2. The change in A's

tari! revenue has two components. In Sector 1, trade diversion reduces tari!revenue by an amount equal to tM A

1y6 B1. In Sector 2, tari! revenue equal to

tM A2(c6 A

2!y6 A

2) is eliminated. Thus, the total variation in A's welfare is

D=A,=A!=M A"tM A2c6 A2#dA

2!tM A

1y6 B1!tM A

2(c6 A

2!y6 A

2)

"dA2!tM A

1y6 B1#tM A

2y6 A2.

Similar calculations for B yield D=B"dB1!tM B

2y6 A2#tM B

1y6 B1. Thus, the aggregate

welfare change for the FTA is

D=,D=A#D=B"dA2#dB

1#(tM B

1!tM A

1)y6 B

1#(tM A

2!tM B

2)y6 A

2

"dA2#dB

1'0.

This establishes that tK satis"es the political-viability constraint ofDe"nition 1. h

Given that aggregate Type-2 contributions in favor of the FTA are nil, itremains to be shown that member governments are willing to put it on thepolitical agenda in the "rst place, i.e., that the net e!ect of the agreement'se$ciency gains and of any variation in Type-1 contributions after its signaturewould leave the governments of A and B at least as well o! as before in terms of<. Strictly speaking, governments could trade o! positive net Type-2 contribu-tions against the discounted value of any drop in Type-1 contributions after theagreement. However, we will show below that such an intertemporal trade-o!need not be envisaged here.

In each period, Type-1 truthful contribution schedules are of the formCi

k"maxMni

k!bi

k; 0N, where the constants bi

kdetermine how the rents gener-

ated by protection at the expense of consumers are shared between lobbies andthe government. Let<M i and<K i stand respectively for the equilibrium value of thegovernment's objective function before and after the FTA (i.e. in Period 1 and in


10This principle has close parallels in agency theory. A detailed analysis of the endogenousdetermination of the constants bi

kcan be found in Grossman and Helpman (1994).

Period 2). Let also <M i~k

and <K i~k

stand for the pre- and post-FTA values of<i when sector k is not allowed to lobby; for instance,<M i

~1is the pre-FTA value of

<i when only Sector 2 is active in lobbying. Finally, let=iF

be the free-trade levelof i's welfare, and note that the functions<M i,<K i, and a=i all coincide at free trade.

Proposition 2. The FTA's ezciency gains are appropriated by lobbies throughreduced contributions; i.e., <M i"<K i"a=i

F.

Proof. Sector k's constant bik

is set endogenously to verify <M i(bik,.)"<i

~k; in

words, sector k sets bik

so as to leave the government just indi!erent betweenchoosing the equilibrium tari!s tM i and ignoring sector k altogether.10 Solving byinduction, if Sector 1 is not allowed to lobby, Sector 2 behaves as a singleprincipal and sets bi

2so as to leave the government just indi!erent between free

trade and the tari! maximizing Ci2(ti2, bi

2)#a=(ti). Thus, <i

~1"a=i

F; it fol-

lows that Sector 1 sets bi1

so as to satisfy<M i"a=iF. (Starting the induction with

Sector 1 leads to the same result.) It is now a simple matter to establish thesecond equation, namely <K i"a=i

F. In the FTA, by Lemma 1, producers of

Good 1 sell only in A and producers of Good 2 sell only in B; therefore, the onlyactive lobbies are those of Sector 1 in A and Sector 2 in B. Using an argumentalready made, each lobby acts as a single principal and leaves the relevantgovernment indi!erent between free trade and the distorted equilibrium,by an appropriate choice of the constant bi

k. It follows directly that

<K i"a=iF("<M i). h

Proposition 2 demonstrates that the utility of governments, inclusive ofType-1 contributions, is una!ected by the formation of the FTA. This resultseems at odds with that of Proposition 1, which established that producers wereindi!erent to the FTA but that governments were not because of welfare gains.The reason for this apparent discrepancy is that lobbies are able to appropriateall the e$ciency gains generated by the move from MFN tari!s to a regionalarrangement through reduced Type-1 contributions rather than increasedpro"ts. In turn, the ability of lobbies to appropriate the entire rents fromprotection (and consequently the e$ciency gains generated by an FTA) stemsfrom the absence of inter-lobby rivalry in a model where the wage rate is "xedand cost-of-living externalities are not recognized. Should some element ofrivalry be present, the basic result of this paper } namely, that an appropriatelydesigned FTA is politically viable } would remain valid (indeed, it would bestrengthened), but member governments would be able to retain some of theinduced e$ciency gains.


Note also that the welfare-maximizing tari! vector that is politically viable inthe sense of De"nition 1 will be such that the tari! in the protected sector incountry A will be smaller than tM . To see this note that by lowering the tari! in theprotected sector, welfare unambiguously increases but at the same time, "rmsstart opposing the FTA. As at tM the FTA is &strictly' politically viable, in the sensethat the inequality in De"nition 1 is strictly positive (as shown in Proposition 1),then by continuity there exists a tari! lower than tM that satis"es the politicalconstraint and procures further welfare gains.

4. Extensions and applications

4.1. Implementation without compensatory transfers

The FTA constructed in Proposition 1 relies on monetary transfers betweenmembers. In the present framework, where governments trade o! domesticwelfare against monetary contributions from lobbies, inter-country transfers tocompensate for welfare losses are quite natural. Moreover, such transfers arecommon in RIAs (a prominent example is the structural-funds system in theEuropean Union). Indeed, Sapir (1993) stressed the role of compensation fundsin the success of the EC's second enlargement to Southern European countries.By contrast, the fact that many of the early South}South regional agreements ofthe 1960s failed has been largely attributed to the lack of compensatory mecha-nisms between unequal partners (see, e.g. Foroutan, 1993). However, one mayobject that there is a logical inconsistency in ruling out lump-sum transfersdomestically (such transfers would enable governments to solve income-distri-bution problems without having recourse to distortionary trade taxes) while atthe same time using them to solve distributional problems between countries.Therefore, we consider in this section how the FTA characterized in Proposition1 can be implemented without compensatory transfers. Let us call &importgoods' the goods imported by the regional bloc from the rest of the world(Goods 1 and 2 in our example).

Corollary 1. The FTA characterized in Proposition 1 is politically viable withoutcompensatory transfers whenever the price elasticity of demand for import goods issuzciently large in both member countries.

Proof. As Type-2 contributions are nil, the FTA is politically viable withoutinter-country transfers if and only if D=A50 and D=B50. The formercondition is equivalent to tM A

2y6 A25tM A

1y6 B1!dA

2; the latter to tM B

2y6 A24tM B

1y6 B1#dB

1. As

symmetry implies that tM Ak"tM B

kand y6 A

k"y6 B

k, superscripts can be disposed of and

the political-viability condition can be rewritten, after slight rearrangement, as!dA

24tM

2y62!tM

1y614dB

1. It can be seen by inspection that, as deadweight-loss


11Note that if Markets 1 and 2 are symmetric, in the sense that tM1"tM

2and y6

1"y6

2, then the FTA

without transfers is always welfare increasing, regardless of relative demand elasticities.12An important caveat is in point here: the symmetry assumption can be relaxed only in as much

as Assumption 1 still holds. This may well be a problem with partners of very di!erent sizes, such asCanada and the US: whereas the US market is likely to be large enough to absorb the output of bothcountries in many industries, the same is not true of Canada. If the &liberalizing' country is the samein too many sectors (because it is too small to absorb the area's combined output) compensatorytransfers may be necessary to ensure that its partner shares in the consumer-surplus gains generatedby the agreement. We are grateful to the editor for attracting our attention on this issue.

terms dikare increasing in the own-price elasticities of demands, higher values of

those elasticities enlarge the range of pairs (y61,y6

2) satisfying the political viabil-

ity condition. h

Corollary 2 has a normative implication beyond that of Proposition 1. Thewelfare e$ciency of any given FTA can be enhanced at no political cost byselective elimination of external tari!s, pairwise, in sectors with high demandelasticities.11 Such a move, while preserving across-the-board internal free trade,would allow member countries to exchange protection among themselves whilegenerating consumer-surplus gains. It is also worth noting that, if trade protec-tion took the form of quantitative restrictions, importing countries wouldtypically retain only a fraction of the quota rents, so that the issue of compensa-tion for loss of such rents would lose importance.

4.2. Asymmetric FTA partners

While the symmetry assumption permits the construction of a clearcut argu-ment at minimum computational cost, it is too strong to provide a basis forpolicy prescriptions. Its role in our analysis is to ensure that Period-1 MFNtari!s are equal, so that producers in import-competing sectors are indi!erentbetween selling at home or abroad at the initial tari! vector. Suppose, then, thattari!s are not equal in Period 1. The FTA's external tari! structure must provideundiminished protection to producers in the whole area while opening up onecountry in each sector to external competition. This dual requirement, which isthe essence of our argument, can be achieved by setting, in each sector, anexternal tari! equal to maxMtM A

k, tM B

kN in one country and zero in the other one.

(The availability of compensatory transfers makes the identity of each countryirrelevant.) Then, in both of the area's import-competing sectors, the commonproducer price is at least as high as it was in either country before the FTA. Asa result, import-competing interests do not oppose the agreement; in fact,producers of good k in the country that was initially the least protectionist insector k support the agreement, as internal free trade enables them to share inthe other country's high level of protection.12 However, one country's


13Although most RIAs among developing countries make use of the WTO's enabling clauseexempting them from the constraints of Article XXIV, FTAs among OECD countries are stillsubject to its requirements.

producer-surplus gains are the other's tari!-revenue losses, and the net e!ect onthe agreement's political viability depends on the parameter a, introducing anindeterminacy that symmetry avoids. As for consumer-surplus gains, they areensured, as in the symmetric case, by the elimination of one member's externaltari! in each sector. Thus, relaxing the symmetry assumption does not a!ect thelogic of our argument but clouds it with an indeterminacy of little practicalsigni"cance, since it hinges on the unmeasurable parameter a.

4.3. Implications for GATT 's article XXIV

It is easy to verify that the external tari! structure set up in Proposition 1, i.e.,tK"(tM A

1, 0, 0, tM B

2), is in conformity with article XXIV of the GATT and its 1994

understanding: As required, the FTA covers all trade, and the area's import-weighted average external tari! is smaller than before, as no tari! is raised whiletwo are eliminated. Thus, the FTA is WTO consistent.13

The argument can be carried one step further. Article XXIV's requirementthat no tari! be raised applies on a country-by-country basis, rather than on thearea's average external tari!. Thus, the fact that some countries reduce theirexternal tari!s when joining an FTA cannot be used by others to raise theirs,even if the area's average external tari! ends up lower than before the agreement.By contrast, when countries form a customs union, Article XXIV allows forcompensatory changes at the tari!-line level among member countries accord-ing to the mechanism stipulated in article XXVIII. That is, a country can raiseits external tari! provided that other member countries reduce theirs by a &su$-cient' amount when converging to the common external tari!. Although theformation of preferential trading agreements under Article XXIV has beenconsidered by some observers as a &loophole' in the MFN principle that ought tobe eliminated, our analysis suggests that a case might in fact be made for anextension of Article XXIV } speci"cally, we suggest that the more #exiblecriterion applied by the WTO to customs unions should be extended to FTAs.The reason is that whereas the trade diversion making piecemeal externalliberalization palatable to import-competing lobbies is costless in our model, itis unlikely to be so in reality. Thus, import-competing sectors forced to redirecttheir activity away from their domestic market to a more protected partnercountry may demand increased external protection in the latter. Such anincrease in external protection in one of the member countries would violateArticle XXIV in its present form, even if it were compensated by the eliminationof external protection in other member countries. This, in our view, creates


14Finger (1993) discusses the appropriateness of GATT rules on RIAs and GATT's lack of successin enforcing these rules.

15Note that comparing tari! data before and after the Uruguay Round is biased by the fact thatduring the agriculture negotiations, countries agreed to transform their non-tari! barriers into tari!barriers.

unnecessary hurdles on the political viability of FTAs regardless of their poten-tial to raise world welfare.14

4.4. Prima facie evidence: The Canada}US FTA

If compensating the losers from trade liberalization matters in the political-economy process leading to the formation of regional trading agreements,Proposition 1 suggests that the changes in external tari! structures following theformation of an FTA should be negatively correlated across members. That is,whenever a member country maintains substantial external protection ina given sector, other members can a!ord to reduce their own protection in thatsector while returning the favor in other ones (provided that the supply anddemand conditions characterized in Assumption 1 hold).

We checked directly for this prediction using Canada and US data on tari!sprovided by the WTO Secretariat. Data is for the 96 sectors of the HarmonizedSystem 2-digit classi"cation and is available for 1988 and 1996 (i.e., before andafter the formation of CUSFTA). Again, Proposition 1 predicts that after thesignature of CUSFTA the correlation between Canadian and American tari!sshould decline. Data shows that the Spearman rank correlation between thetari! structure of the US and Canada went down from 0.69 in 1988 to 0.62 in1996. Both were statistically signi"cant at the 99% level, but their di!erence wasnot. Thus, tari! data suggests a very weak e!ect and probably more disag-gregated data will be needed to clearly identify some of the expected results.

However, if e!ective, this exchange of protection among member countriesshould be re#ected in trade patterns: the growth of intra-bloc trade in import-competing sectors (with goods #owing from relatively open member countries tomore protected ones) should be correlated with the growth of imports from therest of the world to those relatively open member countries.

We checked for those predictions using Canada}US SITC trade data at the4-digit level (816 tari! lines) before and after the signature of the Canada}USFTA (CUSFTA) in 1987. Although trade #ows measure changes in protectiononly indirectly (and no doubt pick up other e!ects as well), they have theadvantage of also capturing the e!ect of non-tari! barriers, which wouldotherwise be di$cult to quantify.15

We estimated panel regressions on changes of exports and imports at thetari!-line level, using a &between-regressor' technique (to get rid of general time


16We tested in all regressions for a structural change in the year 1988 using a Chow test and itturned out to be signi"cant at the 99% level.

17Note that coe$cient estimates cannot be directly interpreted here or compared with oneanother, as regressions are not run in double logs (i.e., coe$cients cannot be interpreted aselasticities).

Table 1Exchange of protection in CUSFTA!

D CanRow" D CanRow D USRow# D USRow83}87 88}96 83}87 88}96

Constant 4026 5538H 20 468 18 705(2110) (2812) (11 671) (10 120)

D USCan$ !0.2E-6 0.29HH(0.3E-5) (0.06)

D CanUS% !128.83 3.28HH(1471) (0.21)

R2 0.4E-5 0.03 0.9E-5 0.23d obs. 3260 7335 3260 7335

!Data source is COMTRADE. We use only import data as it tends to be more reliable. Figures inparenthesis are standard errors."CanRow stands for Canadian imports from the rest-of-the-world (excluding the US) and D standsfor change.#USRow stands for US imports from the rest-of-the-world (excluding Canada).$USCan stands for US imports from Canada.%CanUS stands for Canadian imports from the US.HSigni"cant at 95%.HHSigni"cant at 99%.

trends) for two sub-periods: 1983}1987 (3264 observations) and 1988}1996(7344 observations).16 First, we regressed the change in Canadian exports to theUS on the change in Canadian imports from the rest of the world at thetari!-line level for the period 1983}1987, and then ran the same regression for1988}1996. Similarly, we regressed the change in US exports to Canada on thechange in US imports from the rest of the world, for the same two sub-periods.Again, if our theoretical predictions hold, one would expect a (more) positivecorrelation in the second period after CUSFTA was signed and the UruguayRound was concluded. This is con"rmed by the data as shown in Table 1.

Indeed, before CUSFTA was signed the correlation for both countries be-tween the change in exports to the partner and the change in imports from therest of the world was close to zero and statistically insigni"cant. However, afterthe bilateral trade agreement was signed, the correlation becomes positive andstatistically signi"cant at the 99% level.17


18We are grateful to a referee for suggesting this interpretation of our result.19 In the language of trade negotiations, tari! concessions in multilateral rounds deviate from

straight-line proposals in order to compensate losers, usually labor-intensive industries. Similarexceptions are also found in FTAs, as &sensitive' sectors are usually exempted from tari! reductions,at least during the early stages of implementation.

These correlations are only suggestive and should, of course, be interpretedwith caution, as many factors other than CUSFTA and the Uruguay Roundnegotiations may have contributed to shape trade patterns between Canada andthe US. Nonetheless, they appear to be at least consistent with the implicationsof the kind of political-economy considerations analyzed in this paper.

5. Concluding remarks

Whereas FTAs are often considered, for a variety of reasons, as worsearrangements than customs unions (see, e.g. Krueger, 1997), this paper pur-ported to show that the ability of member countries to maintain di!erentdegrees of external protection in an FTA can in fact be used to mitigate theimpact of existing protectionist pressures. The argument is, in essence, a verysimple one: if it is generally acknowledged that members of an FTA exchangemarket access for their domestic exporters, we show that they can also exchangeprotection for the bene"t of their import-competing interests. When they do so,one member country's external tari! becomes like a producer subsidy for itspartner's producers, who then do not need protection at home;18 dismantlingthe partner's own tari!, in turn, generates welfare gains. Thus, producer interestscan be protected at lower welfare costs than if each member had to protect eachand every one of its import-competing industries. The reason is that, unlikea producer subsidy, the external tari! creates a distortion in consumption.Clearly, this is a second-best argument requiring some sort of political-viabilityconstraint. But constraints of that sort are prevalent in actual policy-making,and our preliminary check on US}Canada data suggests that they may indeedexert a signi"cant in#uence on trade-policy outcomes, along the lines predictedby our results.

As a by-product of our analysis, we have highlighted three types of advant-ages o!ered by FTAs which have hitherto been neglected in the literature. First,they provide a way of compensating losers, an aspect largely overlooked (exceptfor Hillman et al., 1995), but of key practical importance.19 Second, followingthe logic of Kemp and Wan (1976), FTAs can be designed so as to yieldwelfare-enhancing reductions in protection (this issue is analyzed formally inKrishna and Panagariya (1998)). Finally, by reducing the average externalprotection of member countries, they can help meet the requirement of &o!ers'


that WTO members face when joining regional agreements (for instance, Argen-tina and Uruguay decreased their average external tari! from 40% in 1986 to12% in 1996 while forming the MERCOSUR with Brazil and Paraguay). Thiscan represent a non-negligible advantage from the many newcomers to theWTO.

Acknowledgements

We are grateful to Alan Winters, the editor, and two anonymous referees forhelpful suggestions. We also thank Arvind Panagariya, Anthony Venables andparticipants at seminars at the universities of Bologna and Lausanne and theCEPR-IUV workshop on Globalization, Regionalism and FDI for comments.This paper was produced as part of a WTO Research Analysis Division researchprogram on the political economy of regionalism; Cadot also gratefully acknowl-edges "nancial assistance from INSEAD under grant 2010-5308R. The viewsexpressed here are those of the authors and do not necessarily represent those oftheir a$liated institutions.

References

Anderson, K., Blackhurst, R. (Eds.), 1993. Regional Integration and the Global Trading System.St. Martin's Press, New York.

Bernheim, D., Whinston, M.D., 1986. Menu auctions, resource allocation and economic in#uence.Quarterly Journal of Economics 51, 1}31.

Bhagwati, J., Panagariya, A. (Eds.), 1996. Free Trade Areas or Free Trade? The Economics ofPreferential Trading Agreements. AEI Press, Washington, DC.

Cadot, O., de Melo, J., Olarreaga, M., 1999. Regional integration and lobbying for tari!s againstnon-members. International Economic Review 40, 635}658.

de Melo, J., Panagariya, A. (Eds.), 1993. New Dimensions in Regional Integration. CambridgeUniversity Press, Cambridge, UK.

de Melo, J., Panagariya, A., Rodrik, D., 1993. The new regionalism: A country perspective.In: de Melo, J. Panagariya, A. (Eds.), New Dimensions in Regional Integration. CambridgeUniversity Press, Cambridge, UK, pp. 159}193.

Ethier, W.E., 1998. Regionalism in a multilateral world. Journal of Political Economy 106 (6),1214}1245.

Finger, M., 1993. GATT's in#uence on regional arrangements. In: de Melo, J., Panagariya, A. (Eds.),New Dimensions in Regional Integration. Cambridge University Press, Cambridge, UK.pp. 128}147.

Foroutan, F., 1993. Regional integration in Sub-Saharan Africa: Past experience and futureprospects. In: de Melo, J., Panagariya, A. (Eds.), New Dimensions in Regional Integration.Cambridge University Press, Cambridge, UK, pp. 234}271.

Gi!en, R., 1904. Economic Inquiries and Studies. George Bell and Sons, London.Grossman, G., Helpman, E., 1994. Protection for sale. American Economic Review 84, 833}850.Grossman, G., Helpman, E., 1995. The politics of free-trade agreements. American Economic

Review 85, 667}690.


Hillman, A., Long, N.V., Moser, P., 1995. Modelling reciprocal trade liberalization: The politicaleconomy and national welfare perspectives. Swiss Journal of Economics and Statistics 131,503}515.

Irwin, D., 1993. Multilateral and bilateral trade relations in the world trading system: An historicalperspective. In: de Melo, J. Panagariya, A. (Eds.), New Dimensions in Regional Integration.Cambridge University Press, Cambridge. pp. 90}119.

Kemp, M., Wan, H., 1976. An elementary proposition concerning the formation of customs unions.Journal of International Economics 6, 95}97.

Krishna, P., 1998. Regionalism and multilateralism: A political economy approach. QuarterlyJournal of Economics 113, 227}251.

Krishna, P., Panagariya, A., 1998. On the existence of necessarily welfare-enhancing free trade areas.Mimeo., University of Maryland.

Krueger, A., 1993. Free trade agreements as protectionist devices: Rules of origin. Working paper4352. NBER, Cambridge, MA.

Krueger, A., 1997. Free-trade agreements versus customs unions. Journal of DevelopmentEconomics 54, 169}187.

Krugman, P., 1991. Is bilateralism bad? In: Helpman, E., Razin, A., (Eds.) 1991. International Tradeand Trade Policy, MIT Press, Cambridge, MA.

Levy, P.I., 1997. A political-economy analysis of free trade agreements. American Economic Review87, 513}526.

Mayer, W., 1984. Endogenous tari! formation. American Economic Review 74, 18}20.Panagaryia, A., 1998. The regionalism debate: An overview. Mimeo., University of Maryland.Richardson, M., 1993. Endogenous protection and trade diversion. Journal of International

Economics 34, 309}324.Sapir, A., 1993. Discussion. In: de Melo, J. Panagariya, A. (Eds.), New Dimensions in Regional

Integration. Cambridge University Press, Cambridge. pp. 230}233.Winters, A., 1996. Regionalism versus multilateralism. Discussion paper 1525. CEPR, London.


Can bilateralism ease the pains of multilateral trade liberalization?

Documents

Transcript of Can bilateralism ease the pains of multilateral trade liberalization?