Harmonizing External Quotas in an FTA: A Step Backward?

&Blackwell Publishers Ltd 2002. Published by Blackwell Publishers, 108 Cowley Road,Oxford OX4 1JF, UK and 350 Main Street, Malden MA 02148, USA. 259

HARMONIZING EXTERNAL QUOTAS IN AN FTA:A STEP BACKWARD?

OLIVIER CADOT, JAIME DE MELO, AND MARCELO OLARREAGA*

This paper explores how political-economy forces shape quantitativebarriers against the rest of the world in an FTA. We show that whereasthe dilution of lobbying power in an FTA typically leads to a relaxation ofexternal quotas, this result is likely to be overturned as integration deepens.In particular, we show that cooperation among member countries on thelevel of their external quotas, cross-border lobbying by import-competinginterests in the free-trade area, and the consolidation of national externalquotas into a single one, all lead to stiffer restrictions against imports fromthe rest of the world. We also show that, unlike tariffs, endogenous quotasare not crucially affected by the presence of rules of origin.

1. INTRODUCTION

IN SPITE of the failure to launch a new round of trade negotiations in Seattle in

December 1999, the multilateral trading system remains very strong – perhaps

stronger than ever – in its capacity to prevent discriminatory treatment, to settle

disputes, and to promote gradual trade liberalization. At the same time,

regionalism is also thriving: the European Union is – so far successfully –

managing unprecedented ‘‘deepening’’ and enlargement, while Mercosur and

other regional agreements are faring better than earlier South–South agree-

ments. Is the second of these trends (regionalism)1 reinforcing or sapping the

first (multilateralism)? A voluminous and growing literature has so far provided

an ambiguous answer (see the recent surveys by Winters, 1996; Frankel, 1997;

Panagariya, 1999, 2000; see also the edited volumes by de Melo and Panagariya,

1993; Anderson and Blackhurst, 1993; Bhagwati and Panagariya, 1996a; and

Bhagwati et al., 1999). Over time, concerns expressed in the literature have

ECONOMICS AND POLITICS 0954-1985

Volume 14 November 2002 No. 3

*We are grateful to Rupa Duttagupta, Aaditya Mattoo, Maurice Schiff, participants at seminars atthe WTO and the World Bank, two anonymous referees, and the editors for very helpful comments.An earlier version of this paper was written while Cadot was at INSEAD and was financed undergrant 2010-5308 R. The views expressed are those of the authors and do not necessarily reflect those ofthe institutions to which they are affiliated. Contact addresses: Olivier Cadot, HEC, University ofLausanne, 1015 Dorigny, Switzerland, CERDi and CEPR. E-mail: [email protected] de Melo, University of Geneva, 102 bd Carl Vogt, 1211 Geneva 4, Switzerland, CERDi andCEPR. E-mail: [email protected]. Marcelo Olarreaga, The World Bank, 1818 H Street,NW, Washington, DC 20433, USA, and CEPR. E-mail: [email protected].

1Although we use here the term ‘‘regionalism’’, the expression ‘‘preferential trade agreements’’ is abetter characterization of the type of arrangements we are dealing with in this paper, as some of them,like the US–Israel FTA, have little ‘‘regional’’ dimension. We are grateful to J. Bhagwati forattracting our attention to this point.

shifted from what Bhagwati and Panagariya (1996b) called the static Vinerian

analysis of trade diversion to other issues. First, a growing number of

overlapping preferential trading agreements (PTA) and the resulting criss-

crossing rules of origin have generated what Jagdish Bhagwati aptly called a

‘‘spaghetti’’ problem. Second, a number of authors have analyzed ‘‘dynamic

time-path’’ issues (the expression is from Bhagwati et al., 1999) including

whether PTAs will tend to expand by opening up to new members (on this, see

inter alia Baldwin, 1996; Levy, 1997; or Krishna, 1998). Third, how PTAs affect

the incentives of member countries in setting external trade barriers against non-

members has been explored in a number of papers (see e.g. Richardson, 1993,

1994, 1995; Grossman and Helpman, 1995; Panagariya and Findlay, 1996; or

Cadot et al., 1999, 2001). The present paper is a further contribution to this third

strand of the literature. If a wealth of theoretical arguments suggest potential

sources of conflict between regionalism and multilateralism, sources of

complementarity have also been identified, and both the empirical and historical

evidence are somewhat ambiguous.2 As long as the debate is not finally settled

one way or another – and it may never be: in Frankel’s words (1997, p. 226),

‘‘trading blocs have both building-block attributes and stumbling-block

attributes’’ – further work is needed on particular aspects of the interaction

between preferential trade liberalization and the incentives to seek protection

against non-members. In this perspective, the present paper examines the effect

of harmonizing quantitative restrictions (QR) within a trading bloc on the

endogenous level of the bloc’s external QRs.

How important are quantitative restrictions? There is little doubt that one of

the Uruguay Round’s successes was to contribute to a general reduction in their

use as a trade policy instrument, in particular by industrial countries. However,

in spite of efforts toward the ‘‘tariffication’’ of agricultural QRs and the planned

phasing out of Multi-Fiber Agreement (MFA) quotas in textile and clothing,

they are still widespread. Tariff-quotas are still important in agriculture

(Hathaway and Ingco, 1996). In textiles and clothing, although industrial

countries are formally on schedule in meeting their obligations under the

Uruguay Round, the selective elimination of QRs on categories that account for

a small share of textile trade means that, of the 35 percent elimination of QR

incidence that should logically have been completed by January 1998 under the

planned phasing-out of the MFA, only 1 percent had actually been achieved in

the US at that date (see Finger and Schuknecht, 1999, for details). Also, many of

the special actions allowed under the General Agreement on Tariffs and Trade

(GATT) authorize the use of QRs (e.g., safeguards and balance of payments

measures).3 Voluntary Export Restraint (VER) agreements, a form of QR, have

260 CADOT, DE MELO, AND OLARREAGA

2On the empirical evidence, see Frankel (1997) and references therein; on the historical evidence,see Irwin (1993).

3Many of the restrictions allowed in the General Agreement on Trade in Services (GATS) can alsobe seen as quantitative restrictions; for example, quotas for foreign maritime transport or for foreigncultural products (see Mattoo, 2000).

&Blackwell Publishers Ltd 2002.

also been negotiated under the threat of antidumping measures, as was the case

in March 1999 between the US and Brazil. Thus, notwithstanding commendable

efforts to get rid of QRs, their prevalence simply cannot be overlooked.

When QRs are in use, their harmonization can be a significant step in the

deepening of regional integration. In Europe, for instance, progressive QR

harmonization was undertaken as part of the 1992 Single Market program and

involved the dismantling of national QRs that member countries had hitherto

maintained under the umbrella of Article 115 of the Treaty of Rome (see

Schuknecht, 1992). In the automobile sector, drastic harmonization led the

European Commission to negotiate with Japan the complete phase-out by 1999

of a web of bilateral quotas and VERs, including the extremely restrictive Italian

and French ones. In the much-publicized case of bananas, by contrast,

harmonization took a more protectionist turn, as Germany was unable to

retain the free-trade regime that it had secured under the special ‘‘banana

protocol’’ of the Treaty of Rome. The European banana regulation led to a

series of complaints, first at the GATT, then at the WTO, which the European

Union lost (see, e.g., Schohe, 1997, or Cadot and Webber, 2002). The

automobile and bananas examples show just how diverse the outcomes of QR

harmonization can be. The question of this paper is whether, in general, the

harmonization of national QRs in a trading bloc and their ultimate

consolidation into a single area-wide external QR should be expected to

reinforce or weaken the political pressure for protection against non-members.4

In the case of tariffs, whether internal free trade, in the form of a Free Trade

Area (FTA) or Customs Union (CU), would reduce or enhance political

pressure for external protection was first explored formally by Richardson

(1993, 1995). Both papers argued that FTAs are more likely to bring external

tariff reductions and highlighted the importance of tariff revenue considerations

behind this result. In Richardson (1993), he argued that the importing country in

an FTA would tend to reduce its external tariff if competition within the FTA

reduced the domestic price, thus making a high external tariff redundant. By

lowering its external tariff, the importing country would raise revenue from rest-

of-the-world imports instead of transferring the rent to exporters in the FTA

partner. In Richardson (1995), he also highlighted the role of competition for

tariff revenue within an FTA as a force that could lead to the complete

elimination of external tariff barriers. By lowering its tariff below the partner’s

level in an FTA a country could raise tariff revenue not only on its imports, but

also on (part of ) its partner’s deflected imports. In equilibrium these incentives

could lead to the complete elimination of external tariffs by both countries. We

proposed a generalization of Richardson’s analysis in Cadot et al. (1999), where

we assessed the different political-economy forces released by preferential trade

HARMONIZING QUOTAS 261

4The mirror question is whether the existence of QRs rather than tariffs makes regional integrationmore feasible. This is explored in Panagariya and Duttagupta (2000) within a political-economysetting.


liberalization under FTAs and CUs. But as tariff-revenue competition is unlikely

to play a crucial role in actual policy-making – at least in industrial countries –

its importance in the game-theoretic analysis of tariff equilibria is a somewhat

unappealing feature. QRs are more well-behaved in this regard as they do not

create the kind of discontinuity in payoffs which are encountered with tariffs.

That is, whereas a country reducing its external tariff to just below that of its

FTA partner will see its imports (and consequently tariff revenue) jump up by a

discrete amount, the same does not apply to QRs, as the point of equality of

external QRs has no special significance. Thus, the analysis of QRs is both

empirically relevant and analytically better suited to the understanding of

strategic interaction between FTA partners.

We tackle this question using the Grossman–Helpman (1994) model of

influence activities, in which lobbies representing owners of specific factors in a

Ricardo–Viner economy face the government with contribution schedules

conditioned on the vector of trade-policy instruments (here, the vector of quotas

in import-competing sectors). Under Bernheim and Whinston’s (1986)

‘‘truthfulness’’ restriction, the derivative of a lobby’s contribution function is

the derivative of the industry’s profit function; thus, the political equilibrium is

directly related to some of the economy’s fundamentals (e.g. import-demand

elasticities, levels of domestic output, or the fraction of quota rents retained by

the importing country).5 Unlike Syropoulos (1999), who considers strategic

interaction between a trading bloc and the rest of the world, we focus our

analysis on what happens inside the bloc. Internal trade liberalization in the bloc

induces a reallocation of real resources. This reallocation also redistributes the

cards politically: winning sectors are better able to lobby for external protection,

losing sectors less so. If the winning sectors happen to be those that compete

with the rest of the world, external protection will go up.

We consider first the pure effect of the elimination of internal trade barriers in

an FTA on the external quotas of member countries. That is, we suppose at this

stage that member countries keep independent external quotas and set them

non-cooperatively (the import-competing good’s domestic price is common to

the free-trade area, though, because barriers to internal trade have been

removed). In this context, we show that member countries endogenously choose

to relax their external quotas compared to their pre-FTA levels. The reason is

that in an FTA, the effect of a given change in a member country’s external

quota on the import-competing good’s domestic producer prices is less than

before the FTA’s formation, because the domestic price is determined in a larger

area after the FTA’s formation. This cuts the marginal benefit of the quota

change for the import-competing lobby, thereby reducing the incentive to lobby.

However, as integration deepens, taking the form of a cooperative determination


5Given that lobbying here only implies an income transfer from the firm to the government, anddoes not imply any use of productive resources, there is no possibility for lobbying to paradoxicallylead to higher levels of welfare as it draws away resources from the import-competing sector (seeBhagwati and Srinivasan, 1980, 1982).


of external quotas, and ultimately of their consolidation into a single one, we

show that this pro-free-trade effect is reversed. The reason is that cooperation

and deeper integration enable lobbies to internalize a ‘‘protection spillover’’

(namely, when member country i tightens its external quota under pressure from

its domestic import-competing lobby, the price rises in the whole area, benefiting

import-competing lobbies in other member countries as well), leading to tighter

external quotas.

The paper is organized as follows. Section 2 sets up the model; section 3

derives basic results that hold in the context of symmetric countries. Section 4

extends the model to asymmetric countries and explores the role of rules of

origin. Section 5 concludes.

2. THE MODEL

Three countries, A, B, and C, produce and consume three tradable goods,

labeled respectively one, two and three, and a non-tradable, labeled zero, which

serves as numeraire. Each country is the least-cost producer – and hence the

exporter – of one of the three tradables: A of one, B of two, and C of three, so

that international trade follows the ‘‘Meade pattern’’ (Meade, 1955); however,

none is specialized in either production or consumption. All tradable-good

sectors use specific capital and intersectorally mobile labor, while the non-

tradable-good sector uses only labor under constant return to scale; thus, the

wage rate is fixed at the level of its productivity in the non-tradable sector.

Consumers in all countries have identical preferences represented by the quasi-

linear and additive function

U ¼ c0 þX3

k¼1

uðckÞ, ð1Þ

where the function u satisfies the usual properties.

To avoid the complexities arising when terms of trade are endogenous (see,

e.g., Grossman and Helpman, 1995), we assume that all countries are small;6

moreover, the rest of the world, which we call ‘‘C’’, is itself an aggregate of small

countries; thus, strategic interaction is limited to A and B, in which trade policy

is determined a la Grossman–Helpman (1994). That is, lobbies representing the

residual claimants of import-competing sectors (there are two of them in each

country) simultaneously face their government with ‘‘truthful’’ (Bernheim and

Whinston, 1986) political contribution schedules conditioned on the vector of


6As shown by Syropoulos (1999), the introduction of terms-of-trade effects as determinants ofprotection may introduce some ambiguity in our results regarding the relative levels of protectionbetween CU and the pre-FTA situation. On one hand, a CU will internalize terms-of-tradeexternalities, and that will induce it to adopt a more aggressive external trade policy. On the otherhand, trade diversion generated by the CU’s internal trade liberalization raises the price elasticity ofthe ROW demand for its exports. This second effect may more than compensate the first and result ina CU with a lower level of protection.


trade-policy instruments.7 The government, acting as the lobbies’ common

agent, chooses its trade policy in order to maximize a weighted average of

political contributions and aggregate welfare. Trade policy consists of non-

discriminatory quotas on import goods (goods two and three in A, and goods

one and three in B); by ‘‘non-discriminatory,’’ we mean that import licenses do

not restrict the origin of imported goods subject to quotas. Export taxes and

subsidies are ruled out.

In this setup, each lobby is concerned exclusively with the protection it can

obtain for itself, regardless of the protection simultaneously sought by other

sectors, so that the common agency problem degenerates into a simpler

collection of separate principal–agent relationships.8 All world prices are

assumed to be unity. Let pi ¼ ðpi0, : : :, pi3Þ be the vector of domestic prices in

country i; as each country exports one tradable good, two of the elements of pi

must be unity: namely, pA0 and pA1 in A, and pB0 and pB2 in B. Let qi be the two-

dimensional vector of i ’s import quotas: thus, qA ¼ ðqA2 , qA3 Þ and qB ¼ ðqB1 , q

B3 Þ;

let also Ki be the set of i ’s import goods. We assume that a uniform fraction

b 2 ð0, 1Þ of quota rents is retained by importing countries and redistributed

lump-sum to domestic residents. That quota rents be retained, at least in part, by

the importing country is crucial to the calculus, as it is a necessary condition for

the government’s optimization problem to have an interior solution in any QR

setting. However, by virtue of our ‘‘double-smallness’’ assumption (namely, that

owners of specific capital in import-competing industries, i.e. lobbies, are small

in the population, and that countries A and B, whose trade policy we are

interested in, are small compared to the collection of countries aggregated in

‘‘C’’), we do not have to worry about the effect of retained quota rents on the

incentive to lobby, nor about the effect of unretained quota rents on ROW

welfare. In practice, arrangements on the sharing of rents generated by QRs

differ. In VERs, the rents are retained by exporters; indeed, it is often argued

that those rents are what ‘‘buys’’ the consent of the exporting parties in such

arrangements. By contrast, when import licenses are auctioned out, the rents are

retained by the importing government.

Because preferences are additive and quasi-linear and the wage rate is fixed,

unconstrained expenditure and revenue functions can be used, with the

understanding that the domestic price of good k is a function of the import

quota. For country i, these two functions are respectively Eiðpi, WiÞ and

Riðpi, liÞ, where Wi is i ’s welfare and li its labor endowment. Finally, mikðpikÞ is

country i ’s import-demand function in sector k. The government’s problem is:


7We assume that specific-factor ownership is sufficiently concentrated to ensure that lobbiesdisregard the effect of trade barriers on the price index; moreover, labor is not represented in thelobbying process. See Cadot et al. (1997) for a discussion and extensions.

8 For models with lobby interaction and second-best policy formulation, see Grossman andHelpman (1994) and Cadot et al. (1997). For the problem at hand, such issues are peripheral.


maxqiV i �

Xk2Ki

CikðqikÞ þ aW

i ð2Þ

subject to

Eiðpi, WiÞ ¼ Riðpi, liÞ þ bXk2Ki

ðpik � 1Þqik, ð3Þ

mikðpikÞ ¼ q

ik 8k 2 Ki, ð4Þ

where a is a constant assumed common to all countries. Along (4), dq ik ¼

ðmikÞ0dpik, whereas, along (3), dWi ¼ ½ðb� 1Þqik=ðm

ikÞ

0þ bðpik � 1Þ�dq ik. (The last

expression makes use of the fact that @Ei=@Wi ¼ 1 under quasi-linear

preferences and of the fact that the price of the good entering the utility

function additively is taken as the numeraire.) Substituting these, using the fact

that mik ¼ qik, and applying Bernheim and Whinston’s (1986) ‘‘truthfulness’’

property whereby @Cik=@qik ¼ @pik=@q

ik, we obtain the first-order condition

corresponding to problem (2)–(4):

@Vi

@qik�

@pik@qik

þ a@Wi

@qik¼yik � að1� bÞq

ik

ðmikÞ

0 þ abðpik � 1Þ ¼ 0, ð5Þ

where yik is the output of sector k in country i. As abðpik � 1Þ > 0, it follows that,

at the optimum,

ffik � ½yik � að1� bÞqik�=ðm

ikÞ

0 < 0. ð6Þ

Expression ffik will be useful later on when comparing quota levels before and

after an FTA. The second-order condition is ½ðyikÞ0=ðmikÞ

0þ að2b� 1Þ � ffikðm

ikÞ

00�=

ðmikÞ0 < 0; it is verified provided that the fraction of quota rents retained by the

importing country, b, is at least one-half and that a is sufficiently large. For

instance, with linear demand and supply functions ½ðmikÞ

00¼ 0�, if b ¼ 1 then the

second-order condition boils down to a > jðyikÞ0=ðmikÞ

0j > 0: We assume from

now on that the second-order condition is satisfied.

Let qq ik be the solution to (5). Its comparative-statics properties are

straightforward. Given that the second-order condition holds, @qqik=@a has the

sign of the cross-partial derivative

@2Vi

@qik@a¼ �

ð1� bÞqikðmikÞ

0 þ bðpik � 1Þ > 0.

In accordance with intuition (since A and B, being small countries, have nothing

to gain, in aggregate, from trade protection) a higher value of the weight on

welfare in the government’s objective function leads to a relaxation of the

optimum quota. Conversely, a parametric increase in the level of good k’s

domestic output yik would lead to a tightening of the quota on good k: this is

because an increase in the domestic output of good k raises sector k’s incentive

to lobby for a higher domestic price – or, equivalently, for a tighter quota.



3. EXTERNAL QUOTAS IN AN FTA

Supposenow thatA andB formanFTAwherebyA eliminates its quota on imports

of good two from B, while B eliminates its quota on imports of good one from A.

Meanwhile, both determine their external quotas on imports of good three fromC

independently, at a new level which reflects the change in trade flows triggered by

the FTA. Two questions arise: First, if A and B determine their quotas non-

cooperatively, how does the presence of regional free trade affect the optimal

level of their quotas on the area’s import good (good three)? Second, would

cooperation in the determination of A and B’s national quotas – a step forward

in regional integration – make them more or less protectionist vis-a-vis the rest

of the world? These two questions are tackled in the following two sections.

As far as external quotas are concerned, the action is confined to sector three,

which competes in A and B with imports from the rest of the world. Therefore,

the game played after the formation of an FTA has only four players: the lobbies

representing sector three in A and B respectively, and the two governments. In

the first stage, the lobbies set their contribution schedules; in the second stage,

the two governments set external quotas that are best responses vis-a-vis each

other given their domestic lobby’s contribution schedule. In the game’s simplest

form, the import-competing lobby of each member country takes the other

member’s policy decision as given. Therefore, q j enters as a parameter in the

contribution function of i ’s lobby, but i ’s lobby does not attempt to influence it.

As integration deepens, each country’s lobby may carry out influence activities

in the partner country or, alternatively, members may cooperate to harmonize

their external quotas. Ultimately, economic integration may lead to a complete

harmonization of external trade barriers, with lobbies in both countries merged

into one big lobby spanning the entire area. Each of these possibilities is

considered in the following subsections.

3.1 Non-cooperative FTA

When two countries form an FTA, their ability to keep external trade barriers at

unequal levels creates opportunities for arbitrage, as goods originating from the

rest of the world can be imported into the low-protection country and then re-

exported to its more protectionist partner, undermining the effectiveness of the

latter’s external protection. Because of this, FTAs are generally riddled with

rules of origin preventing the transshipment of imported goods through the

area’s internal borders. When external protection takes the form of tariffs, such

rules of origin are crucial because, in their absence, unlimited quantities of

imports could be brought in from the rest of the world for arbitrage purposes

(see Krueger, 1993, and Cadot et al., 1999). By contrast, when external protection

takes the form of quantitative restrictions, rules of origin are less crucial,

because the quantities of foreign goods available for arbitrage are, in any case,

limited. Because such rules introduce significant complications into the analysis

without modifying the main results, we relegate their analysis to the Appendix.



Once A and B liberalize trade in sectors one and two, the only remaining trade-

policy issue is the determination of external quotas (affecting imports from C) in

sector three. In that sector, the optimization problem faced by the government of

each member country is somewhat similar to what it was before the FTA, except

that good three has now a single price p3 in the whole free-trade area, so that

constraint (4) must be suitably modified. Without general-equilibrium interaction

in either welfare or political contributions, the government’s maximization

problem involves only variables pertaining to the area’s import-competing sector

(sector three). This enables us to eliminate sector subscripts, keeping only

country superscripts. Accordingly, country i ’s government (i ¼ A, B) solves

maxqiV i � CiðqiÞ þ aWi ð7Þ

subject to

Eiðpi, Wi Þ ¼ Riðpi, liÞ þ bðp� 1Þqi, ð8Þ

mAðpÞ þmBðpÞ ¼ qA þ qB. ð9Þ

The first-order condition for problem (7)–(9) implicitly defines country i ’s

reaction function in ðqi, q j Þ space:

@Vi

@qi¼yi � aðmi � bqi Þ

ðmAÞ0 þ ðmBÞ0þ abðp� 1Þ ¼ 0. ð10Þ

Let �qqiðq jÞ be the solution to (10). Note that all strategic interaction betweenA and

B takes place through the adjustment of the area’s common price p.9 Condition

(10) differs from (5) in two (related) ways. First,mi can be in excess of qi in (10) if i

also imports from its FTApartner, or less than qi if i re-exports to its FTApartner.

Second, because (4) has been replaced, in (7), by the aggregate relation (9), the first

term’s denominator is now the price derivative of the aggregate import-demand

function. Let us call this first term �ffi, by analogy with (6).

We can now tackle the first question, namely how the FTA affects quota

policy towards the rest of the world. Suppose that A and B are symmetric in the

import-competing sector (we maintain this simplifying assumption throughout

this section, and relax it in subsection 4.2). Under the second-order condition,

the two objective functions VA and VB are globally strictly concave in own

strategies (i.e. in qA and qB respectively); therefore, (10) characterizes the game’s

unique pure-strategy equilibrium.10 Moreover, this unique equilibrium is


9 It is shown in the Appendix that qA and qB are strategic complements whenever a > 1; however,the results below do not hinge on whether they are strategic complements or substitutes.

10 See Friedman (1986, p. 43). Strictly speaking, (10) characterizes only the equilibrium of thesecond-stage subgame whose strategic variables are the external quotas; a full characterization of thegame’s equilibrium would also involve the contribution schedules. However, ‘‘truthfulness’’ pinsdown the contribution schedules up to a constant (whose role is to determine how the economic rentsgenerated by the quotas are shared between the government and the lobbies). As the policy outcome isunaffected by the rent-sharing issue, we will not provide here a complete characterization of thecontribution schedules (see Grossman and Helpman, 1994; or Cadot et al., 1997).


symmetric, so after the FTA’s formation, A and B have identical imports and

domestic prices, and (10) simplifies to

@Vi

@qi¼yi � að1� bÞqi

ðmAÞ0 þ ðmBÞ0þ abðp� 1Þ ¼ 0, ð11Þ

which can be readily compared to (5). We will show that the equilibrium

characterized by (11) entails larger (looser) quota levels than before the FTA. To

see this, let bars denote variables evaluated at their post-FTA values, and

suppose that qqi ¼ �qqi. Then qqA þ qqB ¼ �qqA þ �qqB, so ppA ¼ ppB ¼ �pp, and the only

difference between (5) and (11) is that, for any value of qi, ffiðqi Þ < �ffiðqi Þ(because the latter’s denominator is larger in absolute value and both are

negative numbers); but this implies that qqi < �qqi, a contradiction. Thus, qqi 6¼ �qqi;

moreover, as @2Vi=@ðqi Þ2 < 0 at qqi, it follows that qqi < �qqi. Therefore, we have:

Proposition 1. If A and B are symmetric in the area’s import-competing sector,

the formation of an FTA without rules of origin leads them to relax their

external quotas.

Proposition 1 shows that regional trade liberalization creates a force working to

reduce the level of the area’s trade barriers against the rest of the world. The

reason is as follows. A tighter quota benefits the import-competing lobby only

inasmuch as it leads to a higher domestic price. Prior to the formation of an

FTA, the link between the national quota and the domestic price in sector three

was described by (4), along which, omitting again industry subscripts,

dpi ¼ dqi=ðmiÞ0. After the FTA’s formation, the relationship between quota

and domestic price (the domestic price is now common to the entire area) is

described by (9), along which dp ¼ dqi=½ðmAÞ0 þ ðmBÞ0�. Clearly, the marginal

benefit of a tighter quota is reduced in an FTA, because the common domestic

price, being determined in a larger area, is less sensitive to changes in a member

country’s individual quota. This reduces the optimal level of the quota.

This result can be related to earlier ones by Richardson (1994) and Panagariya

and Findlay (1996). Both papers argue that the power of domestic lobbies is

diluted more effectively in a CU than in an FTA because in the former, lobbies

need to capture a larger political constituency in order to influence policy

decisions. Here, the power of the import-competing lobby is diluted even in an

FTA, but the source of the dilution is economic rather than political.

Proposition 1 is also in accordance with the results of Richardson (1995) and

Cadot et al. (1999). Both papers showed that, in the absence of rules of origin,

tariff-revenue competition in an FTA would give rise to a sort of ‘‘Bertrand

paradox’’ in which the only non-cooperative tariff equilibrium would be the

outright elimination of external tariffs by both member countries. Whereas the

real-life importance of tariff-revenue competition can be debated, we show here

that a weaker version of the result – namely, that an FTA without rules of origin

weakens lobbying pressure – still holds true in a setting where member countries



have only quotas and have accordingly no revenue-driven incentive to undercut

each others’ trade barriers.11

3.2 Quota Harmonization and Integration

Consider first the possibility of cross-country influence activities within the area.

If j ’s import-competing lobby can undertake influence activities in i, the political

equilibrium in i is determined by

@ ~VVi

@qi¼

@pi

@qiþ@p j

@qiþ a

@Wi

@qi¼ 0. ð12Þ

Let ~qqi solve (12); as @p j=@qi ¼ y j=½ðmAÞ0 þ ðmBÞ0� < 0 whenever y j > 0, (12)

implies that ~qqi must be lower than �qqi whenever y j > 0. Moreover, under

symmetry and with a common price in the area, we have yA ¼ yB, cA ¼ cB,

whence mA ¼ mB and ðmAÞ0 ¼ ðmBÞ0. Let respectively y, c, m, and m0 denote these

common values. Writing (12) in full, we have

@ eVVi

@qi¼

2y� að1� bÞqi

2m0þ abð p� 1Þ ¼ 0. ð13Þ

In order to compare (13) with (5), let us evaluate (13) at qqi, the solution to (5).

Putting hats on variables evaluated at qqi and taking the difference between the

two first-order conditions, we have�@ eVVi

@qi�@Vi

@qi

��qqi¼

2yy� að1� bÞqqi

2 bmm0 �yy� að1� bÞqqibmm0

¼ ~ff j � ff j ¼að1� bÞqqi

2mm0< 0,

which implies that ~qqi < qqi: Thus, under symmetry, an FTA with cross-border

lobbying is characterized by an external quota that is not only tighter than that

of an FTA without cross-border lobbying, but also tighter than the pre-FTA

non-discriminatory quota. Intuitively, cross-border lobbying cancels out the

price dilution effect of the FTA formation on firm’s profits. In other words, after

the FTA formation, for a given quota increase, price falls are smaller, which

implies less lobbying by domestic firms. However, once we allow for cross-

border lobbying this effect vanishes. On the other hand, as price changes are

smaller, it also becomes less socially costly to tighten the quota in an FTA. This

effect is not canceled out by cross-border lobbying and therefore leads to a

tighter quota compared to the pre-FTA situation.

Summing up, we have:


11 Technically, the quota game of this paper has continuous payoff functions and is in this sensemore akin to a Cournot game, whereas tariff games have, like Bertrand games, discontinuous payoffs.


Proposition 2. Under symmetry, cross-country influence activities within the

FTA lead member countries to tighten their external quotas compared to their

Nash equilibrium value and compared to their pre-FTA value.

We turn now to inter-governmental cooperation, an arguably deeper degree of

integration. The cooperative solution for i’s external quota is defined by

@ðVi þ V jÞ

@qi¼

@Vi

@qiþ@Vj

@qi¼ 0 for i ¼ A, B. ð14Þ

The question is again whether cooperation leads member governments to relax

or tighten their external quotas compared to their Nash equilibrium values. The

answer to this question hinges on the cross-partial derivative of the function V j

with respect to qi, which measures the externality exerted by i ’s quota (through

its influence on the area’s common price for the import-competing good) on j ’s

objective function. If it is positive at the Nash equilibrium, cooperation would

lead i to relax its quota; if it is negative, it would lead i to tighten its quota. This

cross-partial derivative is equal to

@V j

@qi¼dV j

dp

@p

@qi¼y j � að1� bÞq j

ðmAÞ0 þ ðmBÞ0� �ff j,

which is negative at ð �qqi, �qq j Þ; thus,

@ðVi þ V jÞ

@qi

��qqi, �qq j

¼ �ffi þ �ff j þ abðp� 1Þ < �ffi þ abð p� 1Þ ¼ 0, ð15Þ

implying that the cooperative level of i ’s quota should be lower (i.e. stricter)

than its Nash equilibrium level. Moreover, performing an exercise similar to the

previous one, we get�@ðVi þ VjÞ

@qi�@Vi

@qi

��qqi¼ ffi > 0,

so the external quota is now less tight than the pre-FTA quota. Summing up, we

have:

Proposition 3. Cooperation on the FTA’s external quotas leads member

countries to tighten them compared to their Nash equilibrium value, albeit by

less than cross-border lobbying alone does. Under symmetry, the cooperative

FTA’s external quota is also less tight than the pre-FTA quota.

Intuitively, an increase in i ’s external quota has four distinct effects, three of

which apply both domestically and across the border, and one of which applies

only domestically. First, an increase in qi reduces the domestic output and hence

the political contribution of sector three in both countries. Second, it also raises

consumer surplus in both countries. Third, it reduces the wedge between good

three’s domestic price in the area and its world price, thereby reducing quota



rents in both countries. In addition (this is the fourth effect), an increase in qi

raises the import quantities over which the diminished quota rents are collected;

unlike the previous three ones, this effect is purely domestic. At country i ’s

optimum, because the last effect is positive, the domestic component of the

previous three ones, taken together, must be negative; but as they apply equally

to i and j, if they are negative for i, so are they for j. Hence, the externality

exerted by an increased qi on V j is negative at the optimum, and cooperation

leads to a tightening of quotas, as stated in Proposition 3.

Finally, note that if the two economies merge completely and consolidate their

external quotas into a single one, the post-merge quota will be twice the size of

the pre-merge quota. To see this, observe that if they do, the ‘‘consolidated’’ or

harmonized quota is determined by the usual first-order condition, but with

aggregate variables replacing national ones. Because A and B are symmetric and

expression (5) is linear in country-specific variables (yi and qi ), aggregation

leaves it unaffected. (This invariance holds irrespective of the bargaining rule

used to share the rents of the common external quota.)12

4. EXTENSIONS

We explore here the robustness of our results by considering two extensions of

the model: the introduction of general-equilibrium effects and a different trade

pattern. The analysis of the FTA’s political feasibility, if it was to form

endogenously, is relegated to the Appendix.

4.1 General-Equilibrium Effects

The model developed in this paper is essentially a partial-equilibrium one. On

the supply side, general-equilibrium effects are frozen by the assumption that the

numeraire good is produced with labor only and under constant returns to scale.

On the demand side, they are eliminated by the assumption of a quasi-linear and

additive utility function: the quasi-linearity on the numeraire good suppresses

income effects on goods one, two, and three, and additivity implies that goods

are neither substitutes nor complements for consumption purposes.

Allowing for general-equilibrium effects would lead to cross-industry

lobbying, which would alter some of the qualitative results of the previous

sections. General-equilibrium effects on the labor market (i.e. a wage rate that

reacts to tariff changes through implied changes in labor demand in a full-

fledged specific-factor model) would imply that producers of good one are

adversely affected by the level of protection in sector two through their wage bill.

In such a model, internal trade liberalization would reduce employment by

sectors competing with intra-bloc imports; this would put downward pressure

on the equilibrium wage rate, which would be beneficial to the sector competing


12 This result also holds for endogenous tariffs when countries are symmetric but not when they areasymmetric; see Cadot et al. (1999).


with ROW exports. That sector would consequently expand output, and this

output expansion, through the familiar working of the common-agency model,

would encourage it to lobby more aggressively for protection. Thus, the level of

external protection would, ceteris paribus, rise with internal trade liberalization

because of the inter-sectoral reallocation of productive resources within the

liberalizing economies.

The effects of substitutability in demand in our political-economy model

would be ambiguous. On one hand, were goods one, two, and three to be

substitutes in demand, internal QR liberalization would trigger an effect

identified by Syropoulos (1999). Namely, after internal trade liberalization, trade

diversion due to external protection would not generate quota rents on intra-

bloc trade flows anymore. Ceteris paribus, this would reduce the marginal

welfare benefit of external protection to any member country. But in a political-

economy model such as ours, this effect, which would undoubtedly be present,

would not be the end of the story. Internal trade liberalization, leading to lower

internal prices for the goods traded within the bloc, would also, through

substitutability, reduce the demand for the good imported from the rest of the

world. In a QR regime (in contrast to a tariff regime where this effect would be

irrelevant), the equilibrium domestic price and output of the good competing

with ROW imports would go down, thus reducing the sector’s incentive to seek

protection. The endogenous level of the quota would thus, ceteris paribus, tend

to be enlarged. Which of the two effects (supply side or demand side) of

substitutability dominates would be an empirical question.

4.2 An Alternative Trade Pattern

The interest of the Meade trade pattern assumed so far is that FTA partners

import the same good from the rest of the world, so that interesting coordination

issues arise. We explore here what would happen under a different trade pattern.

Suppose, following Syropoulos (1999), that A imports good one and exports the

other two goods, B imports good two and exports the other two, and C imports

good three and exports the other two. Then A and B import different goods from

the rest of the world, and no direct coordination issue arises, although indirect

issues will arise. Consider what happens on market one in A after the formation

of an FTA. In this section, we relax the symmetry assumption of the previous

sections as it could not hold under the present trading pattern. In order to avoid

a taxonomy of cases, we will assume that A’s consumption of good one is

sufficient to absorb B’s output and the quota on imports from C at the free-trade

price. Good one is imported into A freely from B and subject to an external

quota from C. In B, producers of good one, having free access to A’s market, are

not concerned by its domestic price at home and consequently do not lobby for

external protection. Given this and the fact that A’s domestic price is above its

world price, all of B’s output of good one is sold in A, and B’s own consumption

of that good is now covered by imports from C. Thus, the FTA alters the trade



pattern, generating intra-industry trade in good one since B is now both an

exporter (to A) and an importer (from C).

After the FTA has been formed, both A and B import good one from the rest

of the world. But only A’s external trade policy raises an issue in sector one,

since B maintains free trade in the absence of domestic lobbying for protection.

The interesting question is then, is A induced to tighten or to relax its quota on

good one from C compared to its pre-FTA quota? In order to answer this

question, we need to go back to the original policy problem. Note that A’s

imports are now given by

mA1 ¼ qA1 þ yB1 , ð16Þ

whereas mB1 ¼ cB1 : Clearly, for A to maintain protection in sector one while

having selective free trade with B which has itself no protection, there has to be a

rule of origin. We leave the full treatment of rules of origin to the Appendix, but

here such a rule plays a very clear-cut role, simply preventing the unlimited

arbitrage of domestic-price differences within the free-trade bloc. The problem

solved by the government of country A is:

maxqA1

VA � CA1 ðqA1 Þ þ aW

A ð17Þ

subject to

EAðpA, WAÞ ¼ RAðpA, lAÞ þ bðpA1 � 1ÞqA1 , ð18Þ

mAðpA1 Þ ¼ qA1 þ yBðpA1 Þ. ð19Þ

Using (16), the FOC is

@VA

@qA1¼yA1 � a½yB1 þ ð1� bÞqA1 �

ðmA1 Þ0� ðyB1 Þ

0 þ abðpA1 � 1Þ ¼ 0. ð20Þ

Formula (20) is not directly comparable to (5). However, suppose that yA1 , the

level of good one’s output, was the same in both expressions. Compared to (5),

the first term’s numerator would have gone down in absolute value whereas its

denominator would have gone up, also in absolute value. Recall that when the

FOC is satisfied, its first term must be negative since its second one is positive.

As the negative first term has gone down in absolute value, the curve plotting

@VA=@qA1 as a function of qA1 has shifted up, whence the optimal value of qA1 lies

now to the right of its previous location. In other words, country A’s quota on

imports of good one from C is relaxed compared to its pre-FTA value. But then,

if the quota is enlarged, the domestic price goes down, which implies a

contraction in domestic output. Thus, yA1 is smaller in (20) than in (5), which

reinforces the conclusion.

However, note that the comparison is somewhat misleading, since we are

comparing a non-discriminatory quota (pre-FTA) with an external quota (post-

FTA). In order to assess the effect of the FTA on the bloc’s total imports, we must



now aggregate imports of good one by A and B after the pattern of trade has

changed. It turns out that those total imports are up. To see this, observe that,

prior to the FTA, C’s exports of good one to the area, xC1 , are given by

xC1 ¼ qA1 � xB1 ð1Þ

¼ qA1 � ½yB1 ð1Þ � cB1 ð1Þ�,

where xB1 ð1Þ, yB1 ð1Þ, and cB1 ð1Þ are respectively B’s exports, output, and

consumption of good one at free trade (prior to the FTA, B does not protect

good one since it is an exporter of the good). After the FTA’s formation, as

already noted, all of B’s consumption of good one is covered by imports, so

xC1 ¼ �qqA1 þ cB1 ð1Þ,

where �qqA1 denotes A’s post-FTA external quota on good one. But if �qqA1 > qA1 , it is

clear that C’s total exports (the area’s total imports) of good one are up.

This means that the quota-relaxation effect identified in section 2 is robust to

the change in trade pattern considered here. The reason is as follows. Recall that

the quota-relaxation effect was previously due to the fact that the area’s

common domestic price was less sensitive to an arbitrary tightening of national

external quotas than ‘‘national’’ domestic prices before the FTA, thus reducing,

at the margin, the incentive to lobby for tighter quotas. Here, two distinct effects

play in the same direction. First, any tightening of A’s quota would raise imports

from trading partner B along the latter’s supply curve, thus mitigating the price-

raising effect (what the lobby is ultimately interested in). This obviously is a

dilution effect that is fairly similar (although not identical) to what we had

before. But there is a second effect, namely the fact that the equilibrium output

of good one in A has shrunk under the effect of intra-bloc trade liberalization,

contributing to discourage lobbying. In sum, trade diversion plays here the

useful role of reducing the return to lobbying.

Again, cross-border lobbying would anihilate this liberalizing effect, as

producers of good one in B would now enter the game, having an interest of

their own in seeing protection in market A rise. As the logic is similar to that of

the previous section and the intuition is clear, we will not develop the argument

formally here.

5. CONCLUDING REMARKS

From a political-economy perspective, it has been often argued that deep forms

of regional integration are a lesser evil for the multilateral trading system (e.g.

Richardson, 1994; Panagariya and Findlay, 1994; and Bhagwati and Panagar-

iya, 1996). When special interests from different countries must coordinate their

lobbying efforts to influence policies set at a supra-national level, Olson-type

collective-action problems are less likely to be overcome, and lobbying pressures

for trade protection tend to be diluted.



By contrast, we suggested in an earlier paper (Cadot et al., 1999) that,

notwithstanding possibly serious collective-action problems, deep integration

might enable special interests to internalize cross-border externalities in their

lobbying activities, leading to increased external protection. However, our

results were derived in an endogenous-tariff model in which (as in Richardson,

1995) competition for tariff revenue played a significant role, although its role in

actual policy-making is of debatable relevance, at least in rich countries. In this

paper, we extended our analysis to the highly relevant case of quantitative

barriers, where quota rents, even if they are captured by the importing countries,

do not generate competition and therefore do not affect the strategic interaction

between the regional bloc’s member states.

The new results contain both good and bad news for the multilateral trading

system. The good news is that shallow, non-cooperative regionalism (which

eliminates intra-bloc barriers but does not involve the coordination of external

barriers) helps keep import-competing interests in check because of an economic

dilution mechanism that is different from the collective-action problem

previously stressed in the literature. The bad news is that those favorable effects

are likely to reverse themselves as regional integration deepens through the

coordination and harmonization of external barriers, enabling special interest to

internalize cross-border externalities (as in the tariff case, although the nature of

the externalities is different). Thus, the new results confirm qualitatively the

intuition of the tariff-case results, although the equilibria are all interior and

feature more plausible outcomes under quantitative barriers than under tariffs.

In a richer model with general-equilibrium interaction on factor markets,

additional considerations would also come into play. As sectors competing with

imports from within the area (sectors two in A and one in B) shrink under the

effect of regional trade liberalization, the sector competing with the rest of the

world (sector three) enjoys reduced rivalry on factor markets and can therefore

expand. In the logic of the present model, sector three’s ability to gain economic

power translates into enhanced political clout as well, leading to renewed

protectionist pressure against the rest of the world. In sum, it seems that, for a

variety of reasons, our pessimistic conclusions on the effect of ‘‘deep

regionalism’’ on outward protectionism are likely to hold irrespective of the

type of trade-policy instruments available to member countries.

APPENDIX

Rules of Origin

This section extends the characterization of the non-cooperative external-quota

equilibrium of section 3 to FTAs with rules of origin. In an FTA with

endogenous external tariffs, if goods imported from the rest of the world could

transit through a member country with a low external tariff into one with a

higher external tariff, the latter would see its protection flounder. Because

transshipment brings tariff revenue, each country in such a setting has an



incentive to undercut the external tariff of its partners, creating a sort of

prisoners’ dilemma leading to the complete elimination of external tariffs by all

member countries. While being efficient from a welfare point of view, this

outcome is, of course, not the one preferred by import-competing lobbies. Rules

of origin, which prevent the transshipment of goods through the area’s internal

borders, prevent this type of arbitrage, and therefore play a crucial role in the

political equilibrium (see Cadot et al., 1999). We show here that rules of origin

play no such role in an FTA with endogenous quantitative restrictions, as the

unique symmetric equilibrium of Proposition 1 retains its properties when rules

of origin are imposed.

Although the equilibrium is symmetric, a full characterization of payoff

functions requires asymmetric quota pairs (with unequal prices) to be

considered. Ceteris paribus, consumer price differentials can arise in an FTA

with rules of origin only if national external quotas are set at sufficiently unequal

levels. To see this, suppose that pA > pB, and let ciðpiÞ be country i’s

consumption of the import-competing good and yiðpiÞ its domestic output. As

the functions ci and y i are identical in A and B, we will omit superscripts on

them. When pA > pB, B’s producers ship their entire output yðpAÞ to A, but rules

of origin prevent goods imported from C into B from being transshipped to A.

Thus, A’s domestic price pA must satisfy the market-clearing equation

qA ¼ cðpAÞ � 2yðpAÞ. ðA1Þ

Note that a necessary condition for the existence of a positive quota qA satisfying

(A1) is that cð1Þ > 2yð1Þ, i.e. that the import-penetration ratio mð1Þ=cð1Þ be in

excess of one-half at free trade. As B’s consumption is met entirely through

imports from C, B’s domestic price pB is determined by

qB ¼ cðpBÞ. ðA2Þ

Let cAðqBÞ ¼ supfqA : pAðqAÞ > pBðqBÞg; i.e. cA defines an upper bound on the

range of qA compatible with pA > pB given qB. Using (21) and (22),

cAðqBÞ ¼ cðpBÞ � 2yðpBÞ

¼ cðc�1ðqBÞÞ � 2yðc�1ðqBÞÞ

¼ qB � 2yðc�1ðqBÞÞ, ðA3Þ

where c�1ð�Þ is the inverse demand for the import good. Because c and c�1 are

decreasing functions whereas y is an increasing one, cA is an increasing function;

moreover, it is readily seen from (23) that its slope ½cA�0 ¼ 1� 2y0 � ½c�1�0 is more

than one (the second derivative of cA is positive with a linear demand and

indeterminate otherwise). Define qB0 by cðqB0 Þ ¼ 0; using (23), qB0 must satisfy

qB0 ¼ 2y½c�1ðqB0 Þ�; that is, qB0 is a fixed point, if any, of the function 2y½c�1ðxÞ�. We

claim that this fixed point exists, is positive, and is unique. Let pB0 ¼ c�1ð0Þ

whenever c�1ð0Þ is well-defined (otherwise we say that pB0 ¼ 1); pB0 is the ‘‘choke



point’’ (vertical intercept) of B’s inverse demand function, when it has one.

Clearly, 2y½c�1ð0Þ� ¼ 2yðpB0 Þ > 0; hence, if it exists, qB0 must be strictly greater

than zero. Moreover, at qB ¼ cð1Þ (B’s free-trade level of consumption, at which

its external quota ceases to be binding), c�1ðqBÞ ¼ 1, so 2y½c�1½cð1Þ�� ¼ 2yð1Þ,

which is, by (21), less than cð1Þ; therefore, at qB ¼ cð1Þ, 2y½c�1ðqBÞ� < qB. It

follows that the function 2y½c�1ðxÞ� has at least one fixed point strictly between

x ¼ 0 and x ¼ cð1Þ; moreover, because it is monotone decreasing, this fixed

point is unique. Having established the existence of qB0 > 0 completes the

characterization of the function cAðqBÞ. A similar function cBðqAÞ can be

constructed on the assumption that pB > pA, leading to a partition of the space

ðqA, qBÞ into three regions: a central band with price equalization, and two

lateral regions with price differentials of opposite signs.

We can now map quota pairs ðqA, qBÞ into pairs of domestic prices ðpA, pBÞ in

order to construct payoff functions VA and VB. Consider the following three

functions: for all x, z in Rþ,

z ¼ fðxÞ , x ¼ cðzÞ � 2yðzÞ,

z ¼ gðxÞ , x ¼ 2½cðzÞ � yðzÞ�,

z ¼ hðxÞ , x ¼ cðzÞ.

Using (A1), it can be seen that f gives pi as a function of qi when qi is such that

pi > p j [i.e. when qi < ciðq jÞ]; similarly, g gives pi as a function of qA þ qB when

pi ¼ p j [i.e. when ciðq j Þ4 qi4 ðc j Þ�1ðq j Þ]; while h gives pi as a function of qi

when pi < p j [i.e. when ðc j Þ�1ðq jÞ < qi]. A few more pieces of notation are

needed to construct payoff functions. Let sðpi Þ stand for consumer surplus in i,

and define the following three continuously differentiable functions:

VV iðqi Þ ¼ Ci½ fðqiÞ� þ afp½ fðqiÞ� þ s½ fðqiÞ� þ b½ fðqiÞ � 1�qig,

�VViðqi, q jÞ ¼ Ci½gðqi þ q jÞ� þ afp½gðqi þ q jÞ� þ s½gðqi þ q jÞ� þ b½gðqi þ q jÞ � 1�qig,

�VViðqi, q jÞ ¼ afp½ fðqjÞ� þ s½hðqiÞ� þ b½hðqiÞ � 1�qig.

Recalling from section 2 that Ci is sector three’s contribution function in i and

noting that the terms in curly brackets are welfare terms, VV i is i ’s ‘‘payoff ’’

(contributions plus appropriately-weighted welfare) when qi < c iðq jÞ; similarly

for �VVi when c iðq jÞ4 qi4 ðc jÞ�1ðq jÞ; and for �VVi when ðc jÞ

�1ðq jÞ < qi. The

function �VVi differs from the other two in that i’s import-competing lobby does

not contribute because it sells its output in j; for the same reason, the producer

price relevant for the calculation of producer surplus p is fðq j Þ instead of the

domestic price hðqiÞ. Using these, i’s payoff function is

V i ¼

VVi if qi < c iðq jÞ

�VVi if c iðq jÞ4 qi4 ½c j��1ðq jÞ

�VVi otherwise.

8>><>>: ðA4Þ



Equation (A4) shows that Vi may be, in general, neither differentiable nor

continuous. As ViVi and �VVi are identical at qi ¼ ciðq jÞ, but ½VVi �0 > ½ �VVi �0, Vi has a

point of non-differentiability at qi ¼ ciðq j Þ. Moreover, as �VVi is strictly greater

than �VVi at qi ¼ ½c j ��1ðq j Þ, it also has a point of discontinuity (a jump down) at

that point; by construction of the functions ci and cj, this jump must lie to the

right of the point where qi ¼ q j. To the right of its jump, Vi rises monotonically

along �VVi up to the point where qi is no longer binding. In addition, because

sector three is, in each country, the only lobby actively trying to influence the

government, its contribution schedule Ci is adjusted so as to leave the

government just indifferent between adopting �qqi � argmax �VVi and free trade.

The last remaining analytical step consists of characterizing the Nash

equilibrium in ðqA, qBÞ. Because all relevant functions and parameters are

identical in A and B, �qqA ¼ �qqB; hence, any equilibrium must be symmetric.

Maximizing the function �VVi given �qq j yields the first-order condition

@ �VVi

@qi¼

@p@qi

þ a

�@p@qi

þ@s

@qiþ b½gðqi þ qjÞ � 1� þ qig 0ðqi þ q j Þ

�¼ yþ aðy� cþ bqi Þg 0 þ abðg� 1Þ

¼y� að1� bÞqi

2m0þ abðp� 1Þ ¼ 0. ðA5Þ

Note that the last transformation comes from the fact that, by implicit

differentiation, g 0 ¼ 1=2m0 where m ¼ c� y, and that, when pA ¼ pB, no intra-

bloc trade takes place so mi ¼ qi. It is apparent by comparison of (A5) with (11)

that the value of qi which maximizes i ’s payoff is the same whether or not rules

of origin are imposed.

Strategic Complementarity

The second cross-partial derivative of the function Vi is:

@ðVi Þ2

@qi@qj¼

ðyi Þ0 � aðmiÞ0 þ �ffiðmi Þ00þ ab

ðmi Þ0 . ðA6Þ

Let ci be i ’s domestic consumption of the import good. After slight

rearrangement and substitution of the first-order condition, (A6) can be

rewritten as

@ðVi Þ2

@qi@q j¼

ð1� aÞðyi Þ0 þ aðciÞ0 þ ab½1� ðp� 1ÞðmiÞ00�

ðmiÞ0.

Provided that a > 1, the numerator of this expression is negative, so qi and q j

are strategic complements (numerical simulations in Cadot et al., 1997,

suggest that plausible policy outcomes require values of a largely in excess of

unity).



A.1 Political Feasibility of FTA

We consider here the political feasibility of an FTA (i.e. its acceptability for

sector-specific lobbies) by looking at the change it involves in the income of the

lobbies net of political contributions. Clearly, the issue arises only if the pre-

FTA (‘‘MFN’’) equilibrium involves positive protection of sector two or three,

since otherwise there would be no loss of protection and hence no political

feasibility issue to start with. We consider country A, which imports goods two

(from its FTA partner B) and three (from the rest of the world). Let us write

aggregate welfare as a function of the quotas in sectors two and three, i.e.

W ¼Wðq2, q3Þ.

Similarly, let CkðqkÞ be sector k’s political contribution, pkðqkÞ its profits, and

vk ¼ pkðqkÞ � CkðqkÞ its income net of contribution. We will use the notation

qk ¼ 1 to denote the elimination of the quota on good k.

The first task is to calculate sector three’s equilibrium contribution in the FTA.

The government’s objective function V, evaluated at the FTA equilibrium, is

�VV ¼ C2ð1Þ þ C3ð �qq3Þ þ aWð1, �qq3Þ.

Were sector three to get out of the lobbying game, keeping sector two’s contri-

bution unchanged, the government’s objective function V would then become

C2ð1Þ þ aWð1, 1Þ.

Equating these two expressions and lettingW* stand forWð1, 1Þ (the value of

welfare at free trade) and �WW for Wð1, �qq3Þ (the value of welfare in the FTA

equilibrium) gives the indifference condition that pins down sector three’s

contribution; namely,

�CC3 ¼ aðW*� �WWÞ.

In words, the lobby representing sector three transfers to the government the

monetary equivalent (given a, the government’s marginal rate of substitution

between money from contributions and welfare) of the distortion created by

protection in sector three. Lobby three’s income net of its political contribution

is thus

�vv3 ¼ �pp3 ��CC3 ¼ �pp3 þ að �WW�W*Þ.

Similar calculations for lobby three’s net income in the pre-FTA (‘‘MFN’’)

regime in which variables are designated by hats give

vv3 ¼ pp3 � CC3 ¼ pp3 þ a½WW�Wðqq2, 1Þ�.

Thus, the variation in lobby three’s net income attributable to the formation of

the FTA is

Dv3 ¼ �vv3 � vv3 ¼ �pp3 � pp3 þ a½ �WW�W*� WWþWðqq2, 1Þ�.



Similar calculations for sector two give

Dv2 ¼ �vv2 � vv2 ¼ �pp2 � pp2 þ a½Wð1, qq3Þ � WW�.

We now derive a necessary condition for the change in the aggregate income of

lobbies, Dv2 þ Dv3, to be positive. Adding up the last two expressions, we have

Dv¼ �pp2 � pp2 þ �pp3 � pp3 þ a½ �WW�W*� WWþWðqq2, 1ÞþWð1, qq3Þ � WW �.

But note that, by construction, positive protection of sector two in the pre-FTA

equilibrium implies vv2 > v*2, or, after slight rearrangement,

a½Wð1, qq3Þ � WW� < pp2 � p*2.

Substituting this in the expression for Dv above and simpifying gives

Dv < �pp3 � pp3 þ a½ �WW�W*þWðqq2, 1Þ � WW�.

By the same reasoning, vv3 > v*3 implies

a½Wðqq2, 1Þ � WW � < pp3 � p*3;

again substituting this in the previous expression gives

Dv < �pp3 � p*3 þ að �WW�W*Þ,

which implies that, for Dv to be positive, it must be the case that the rent from

protection of sector three at its FTA-equilibrium level ( �pp3 � p*3Þ exceeds the

monetary equivalent of its distortionary cost að �WW�W*Þ. This simply says that the

FTA equilibrium involves positive protection of sector three. Thus, provided that

the FTA equilibrium is an interior one, the necessary condition is always verified.

Next, we derive a sufficient condition for the change from MFN to FTA

regime to be politically feasible. It is easily seen that the change in lobby two’s

net income is negative if the MFN regime involves positive protection of sector

two. Thus, the change in sector three’s net income must be positive and

sufficiently large to ensure that Dv is positive. But Dv3 > 0 requires

pp3 � �pp3 < afð �WW� WWÞ þ ½Wðqq2, 1Þ �W*�g.

The term on the LHS is the loss of protection rent for sector three attributable

to the FTA’s formation. The first term on the RHS (in parentheses) is the

efficiency gain attributable to the FTA compared to the MFN regime, which is

necessary for political feasibility given that lobbies are residual claimants. But

the second term on the RHS (in square brackets) is negative. Given that the term

in parentheses, scaled by a, cannot by itself be larger than the LHS since

otherwise the MFN regime would not be an equilibrium, this condition simply

cannot be met. Thus, under the assumptions of strong separability, fixed wage

rate and Meade trade pattern, the FTA is not politically feasible. However, this

result breaks downundermore general assumptions including general-equilibrium



interaction on either the demand or supply side or an alternative trade pattern

with more than one good imported from the rest of the world. Moreover, other

non-economic objectives may be behind the creation of a preferential agreement

(e.g. national security). Finally, note that it can also be shown that the move

from an FTA to quota harmonization within a CU is politically feasible. The

issue of political feasibility is analyzed in detail in Cadot et al. (2001).

OLIVIER CADOT JAIME DE MELO

University of Lausanne, CERDI University of Geneva, CERDI

and CEPR and CEPR

MARCELO OLARREAGA

The World Bank, and CEPR

REFERENCES

Baldwin, R., 1996, A domino theory of regionalism, in: R. Baldwin, P. Haaparanta, andJ. Kiander, eds., Expanding Membership of the EU (Cambridge University Press,Cambridge, UK) 25–48.

Bernheim, B. D. and M. D. Whinston, 1986, Menu auctions, resource allocation, andeconomic influence. Quarterly Journal of Economics 51, 1–31.

Bhagwati, J. and A. Panagariya, eds., 1996a, The Economics of Preferential TradeAgreements (AEI Press, Washington, DC).— and —, 1996b, The theory of preferential trade agreements: historical

evolution and current trends. American Economic Review (Papers & Proceedings) 86,82–87.— and T. N. Srinivasan, 1980, Revenue seeking: a generalization of the theory of

tariffs. Journal of Political Economy 88, 1069–1087.— and —, 1982, The welfare consequences of directly-unproductive profit-

seeking (DUP) lobbying activities: price versus quantity distortions. Journal ofInternational Economics 13, 33–44.—, P. Krishna, and A. Panagariya, eds., 1999, Trading Blocs: Alternative Approachesto Analyzing Preferential Trade Agreements (MIT Press, Cambridge, MA).

Cadot, O. and D. Webber, 2002, Banana splits and banana slips: the European andtransatlantic politics of bananas. Business and Politics 4, 5–39.—, J. de Melo, and M. Olarreaga, 1997, Lobbying and the structure of protection.

CEPR Discussion Paper No. 1574.—,—, and—, 1999, Regional integration and lobbying for tariffs against

non-members. International Economic Review 40, 635–658.—,—, and—, 2001, Can bilateralism ease the pains of multilateral trade

liberalization? European Economic Review 45, 27–44.de Melo, J. and A. Panagariya, eds., 1993, New Dimensions in Regional Integration

(Cambridge University Press, Cambridge, UK).Finger, J. M. and L. Schuknecht, 1999, Market access advances and retreats: the Uruguay

Round and beyond. Policy Research Working Paper No. 2232, The World Bank.Frankel, J., 1997, Regional Trading Blocs in the New World Economic System (Institute for

International Economics, Washington, DC).Grossman, G. and E. Helpman, 1994, Protection for sale. American Economic Review 84,

833–850.



—and—, 1995, Trade wars and trade talks. Journal of Political Economy 103,675–708.

Hathaway, D. and M. Ingco, 1996, Agriculture liberalization and the Uruguay Round, in:W. Martin and L. A. Winters, eds., The Uruguay Round and the Developing Countries(Cambridge University Press, Cambridge, UK).

Irwin, D., 1993, Multilateral and bilateral trade policies in the world trading system: ahistorical perspective, in: J. de Melo and A. Panagariya, eds., New Dimensions inRegional Integration (Cambridge University Press, Cambridge, UK).

Krishna, P., 1998, Regionalism and multilateralism: a political economy approach.Quarterly Journal of Economics 113, 227–251.

Krueger, A. O., 1993, Free trade agreements as protectionist devices: rules of origin.NBER Discussion Paper No. 4352.

Levy, P., 1997, A political–economic analysis of free-trade agreements. AmericanEconomic Review 87, 506–519.

Mattoo, A., 2000, Shaping new rules for trade in services: lessons from the GATS. Paperpresented at the NBER Conference on Trade in Services, Seoul, June 2000.

Meade, J. E., 1955, Trade and Welfare: Mathematical Supplement (Oxford UniversityPress, Oxford, UK).

Panagariya, A., 1999, The regionalism debate: an overview. The World Economy 22, 477–511.—, 2000, Preferential trade liberalization: the traditional theory and new develop-

ments. Journal of Economic Literature 38, 287–331.— and R. Duttagupta, 2000, Did the multi-fiber agreement make the NAFTA

politically more acceptable? A theoretical analysis. Paper presented at the LeitnerConference on Political and Economic Aspects of Regional Integration, YaleUniversity, April 7–8, 2000.— and R. Findlay, 1996, A political-economy analysis of free-trade areas and

customs unions, in: The Political Economy of Trade Reform: Essays in Honor of JagdishBhagwati (MIT Press, Cambridge, MA) 265–287.

Richardson, M., 1993, Endogenous protection and trade diversion. Journal ofInternational Economics 34, 309–324.—, 1994, Why a free-trade area? The tariff also rises. Economics and Politics 6, 79–96.—, 1995, Tariff revenue competition in a free trade area. European Economic Review

39, 1429–1437.Schohe, G., 1997, The WTO panel report on the EU’s banana regime: will the rule of law

prevail over political arbitrariness? European Food Law Review 23, 245–263.Schuknecht, L., 1992, Trade Protection in the European Community (Harwood, Chur,

Switzerland).Syropoulos, C., 1999, Customs unions and comparative advantage. Oxford EconomicPapers 51, 239–266.

Winters, A., 1996, Regionalism versus multilateralism. Policy Research Working PaperNo. 1687, The World Bank.



Harmonizing External Quotas in an FTA: A Step Backward?

Documents

Transcript of Harmonizing External Quotas in an FTA: A Step Backward?