On the Viability of the Multilateral Trade Agreement: A ...

On the Viability of the Multilateral Trade

Agreement: A Political-Economy Approach�

Danilo R. Trupkiny

Texas A&M University

March 2008

Abstract

The big picture issue this paper intends to address is on the de-

terminants that have led to the proliferation of numerous preferential

trade agreements around the world. The paper builds on a framework

originally introduced in Grossman and Helpman�s (1995) The Politics

of Free-Trade Agreements. That paper was designed to explain the

viability of FTAs between two countries. A simple but appealing ex-

tension to a three-country setting allows one to analyze the viability of

a grand accord. An illustration with speci�c functional forms serves to

�nd conditions under which FTAs are somehow partial building blocs,

i.e., an FTA can be feasible when multilateral liberalization is not.

This exercise can be considered as an advance in the direction of how

a model of trade agreements a-la Grossman-Helpman could answer the

dynamic path question of trading blocs.

Key words: International Trade; Free-Trade Agreements; Multilat-

eral Trade Agreement; Political-Economy.

JEL Classi�cations: F13, F15, D78.

�I thank the Private Enterprise Research Center at Texas A&M University for giving methe opportunity to work on this project. I am grateful to Kishore Gawande for numerousconversations on the topic, and to Amy Glass and seminar participants at Texas A&MUniversity for helpful comments. All errors are my own.

yDepartment of Economics, Texas A&M University, College Station, TX 77843-4228.E-mail: [email protected].

1

1 Introduction

A large number of preferential trade agreements do exist nowadays all around

the world - more than 200 are in e¤ect today according to the World Trade

Organization (WTO) -, almost all major countries are members of at least

one preferential trade agreement, and, most importantly, the majority of

these agreements were originated during the last decade. Bhagwati (1995)

has called this scenario a �spaghetti bowl�of tari¤s in which countries charge

the same good at di¤erent tari¤ rates depending on its alleged origin.

One may want to address this issue by focusing on the sources leading

to the proliferation of such a number of preferential agreements. Or, as in

Bhagwati (1991), the discussion may be stated on the grounds of whether

free-trade agreements (FTAs) are building blocs or stumbling blocs.1 That

is, whether they are a step stone or simply an obstacle to the achievement

of a broad multilateral agreement.

Since Viner�s (1950) analysis on the welfare e¤ects of customs unions,

research interests on this topic have considerably caught the attention of

international trade theorists. Viner pioneered the static analysis of trade

agreements providing the message that these arrangements could harm both

a member country and world welfare, i.e., they can certainly be �trade di-

verting�.

There has been academic concern recently in regards to the viability of

successful multilateral trade liberalization under a world that is largely pop-

ulated by potentially harmful FTAs.2 In this sense, Pravin Krishna asks

the following, political-economy question: �are there incentives for FTAs

to keep expanding with more members so as to move toward multilateral

free trade eventually, or will there be incentives instead to keep new mem-

1 In what follows, we will use the term FTA as a preferential trade agreement. In theliterature, the latter is considered more general in the sense that it implies a reductionof external tari¤s among members, whereas the former implies the simple elimination ofmembers�tari¤s. Since this work will treat any agreement as simply removing barriers,then both agreements are identical. Note that, in addition, one may consider another typeof trade agreement such as a customs union (agreements whereby all members impose acommon external tari¤), yet we restrict the analysis here to the FTA case.

2See Panagariya (2000) for a survey of this literature.

2

bers out?� - Krishna (1998), pp. 228. Or, as in Aghion et al. (2007), the

big question of regionalism versus multilateralism could be posed in asking

whether multilateral bargaining or sequential bargaining are more likely to

lead to global free trade. Under multilateral bargaining, all countries simul-

taneously participate in a single round of negotiations. Under sequential

bargaining, subsets of countries negotiate separately through many rounds.

The stumbling view has been supported, among others, by Bhagwati

(1991, 1993), who �nds that even though FTAs may generate static welfare

gains; they �nally reduce the incentives to seek posterior trade liberaliza-

tion. Yet, others such as Summers (1991) argue that deeper integration

within a subset of countries may raise the chances for successful multilateral

negotiations.

The focus of this paper is both on the incentive aspects that explain

the formation of trade blocs and on the viability of the multilateral agree-

ment. The approach followed here belongs to the political-economy theory

of international trade. In particular, this paper builds on a framework �rst

introduced in Grossman and Helpman�s (1995) The Politics of Free-Trade

Agreements. They analyze the viability of an FTA between two countries

using a political-economy approach that emphasizes the interaction between

lobbies and an incumbent government. The paper considers both the case

when the FTA must cover all bilateral trade and the case when some polit-

ically sensitive industries can be excluded from the agreement.

In Grossman-Helpman (1995), international relations involve two dis-

tinct stages of strategic interaction. In the initial stage, political competition

among special interests in each country - the lobbies - determines the govern-

ment�s trade policy preferences. The second stage involves the bargaining

process between the governments, which ultimately shapes the equilibrium

agreement. The initial stage in which lobbies in a single country compete

for the government�s external policy makes use of the analytical framework

previously introduced in Grossman and Helpman�s (1994) Protection for

Sale. In this model, lobbies o¤er policy-contingent campaign contributions

to politicians, who make decisions on the ground of their own objectives.

Thus, a country�s policy stance re�ects, on the one hand, the relative force

3

of its special interests and, on the other hand, the government�s concern on

the welfare of the average voter.

According to this approach, an agreement may emerge when it creates

pro�t gains for exporting sectors in excess of the losses coming both from

import-competing sectors and from any welfare harm caused to the average

voter. Naturally, an FTA requires that the two governments consent the

agreement. Grossman-Helpman�s study concludes that this is most likely

when there is relatively balanced trade between both countries, and when the

agreement provides mostly enhanced protection - exporting sectors receiving

high domestic prices in the partner country - rather than reduced protection

- import-competing sectors receiving a lower domestic price product of the

FTA.

It turns out that reduced protection may involve some trade creation,

whereas enhanced protection generates only trade diversion. Thus, Grossman-

Helpman�s analysis of FTAs establishes that the features raising the viability

of an agreement also raise the likelihood that this agreement �nally reduces

social welfare. Perhaps even more appealing is the paper�s development of

a model in which exceptions from an agreement increase the prospects of

that agreement. They show that a set of industry exclusions might make

an otherwise infeasible FTA politically viable. Moreover, they study the

determinants of the size of this set, and also identify the sensitive industries

that belong to it.3

Following Grossman-Helpman�s tradition, this paper develops a three-

country, political-economy representation in order to �nd conditions for the

viability of a multilateral agreement. Roughly, this model takes the two-

country, FTA-game settled in Grossman and Helpman (1995), and utilizes

it as the starting point to evaluate the prospects of ulterior multilateral

negotiations.

By explicitly introducing a third country into the game, it does not

merely constitute a simple extension to Grossman-Helpman�s approach, but

it represents what I believe is an important change from the original idea.

3 In practice, the size of the set of exceptions should not violate the WTO clause statingthat an FTA should cover �substantially all trade�.

4

Grossman and Helpman look for determinants of the viability of an agree-

ment between two countries, whereas a three-country set up allows us to

move the focus of the analysis toward the question of whether trade blocs

are building blocs or stumbling blocs.

The paper proceeds as follows. Section 2 provides some comments on the

related literature. Section 3 describes the Grossman and Helpman�s model

of free trade agreements in order to set up the framework I will use all along

the paper. Section 4 extends this model to a three-country setting in which

the viability of successful multilateral negotiations is analyzed. This section

extends formally the general analysis that is presented in Grossman and

Helpman. Section 5 assumes particular functional forms so as to illustrate

how Grossman-Helpman works in a world with the possibility of multilateral

liberalization. Section 6 concludes.

2 Related Literature

Krishna (1998) distinguishes between two questions on the issue of whether

trade blocs are building blocs or stumbling blocs. The �rst question refers

to the e¤ects on welfare of any individual country, or group of countries, as

FTAs arbitrarily grow in number until worldwide free trade is reached. The

second question studies the issue of regionalism versus multilateralism by

looking at both the incentives of forming FTAs and the incentives for their

expansion to include new members.

This paper deals with the latter question, as Krishna (1998) does. In

answering this question under a political-economy set up, Krishna �nds ba-

sically two conclusions: �rst, FTAs that divert trade away from the more

e¢ cient countries are more likely to be supported politically - this is in line

with Grossman and Helpman�s �ndings -; and second, such agreements will

reduce the incentives for multilateral liberalization.4 As I will show later,

4Nonetheless, under a related approach Ornelas (2005) shows that, while political mo-tivations may obstruct the multilateral trade agreement, membership in an FTA makessuch a possibility less likely to happen if we allow for external tari¤s to be endogenous.Under this setting, strategic and distributive e¤ects of an FTA reduce the importance ofpolitics in de�ning trade policy.

5

this second result may also be obtained in the present study, under cer-

tain conditions. In other words, it is shown that trade-diverting bilateral

agreements may critically hinder multilateral liberalization e¤orts. The key

di¤erence between mine and Krishna�s approach is that the latter uses an im-

perfect competition model with segmented markets a-la Brander-Krugman,

whereas this paper�s approach makes use of a perfect competition, small-

union model. In addition, unlike Krishna (1998) - and as far as I am aware

unlike other studies in this literature -, I allow for a list of exceptions to the

trade agreements, both into the bilateral and into the multilateral stances,

once again in line with Grossman and Helpman.

In turn, Levy (1997) asks the following two questions: (i) can the option

to form a trade bloc make a previously infeasible multilateral liberalization

feasible? (ii) can this additional option render a previously feasible multi-

lateral liberalization infeasible?

Building a median-voter model, Levy answers the �rst question nega-

tively, stating that it is indeed the infeasibility of multilateral liberalization

at the �rst place what implies that the median voter enjoy higher utility

under autarky than under free trade. Thus, a trade bloc which increases

their utility only generates an increase of their �reservation utility�, and

hence a lower willingness to accept multilateral liberalization. To answer

the second question, he uses two distinct models: �rst, a classic, standard

Heckscher-Ohlin model, and second, a variant of it with one sector produc-

ing a di¤erentiated, monopolistically competitive good. Levy shows that in

the �rst model, the FTA does not obstruct a previously feasible multilateral

agreement but, in the second one, it can. It is worth noting, though, that

Levy (1997) only examines the extreme choice between free trade and au-

tarky, giving thus no role to the endogenous determination of tari¤s and its

relation with the viability of both regionalism and multilateralism.

More recently, and focused on the e¤ects of endogenous coalition forma-

tion of FTAs, Aghion et al. (2007) developed a dynamic model of bargaining

with transferable utility between countries in which a leading economy de-

cides whether to sequentially negotiate free-trade agreements with subsets

of countries, or to simultaneously engage in multilateral negotiations with

6

all countries at once. They show that global free trade is not achieved when

the political-economy motive for protection is su¢ ciently large, and describe

the determinants that may lead to both building and stumbling bloc e¤ects

of FTAs.

This paper is closely related to Aghion et. al. (2007) by the fact that

both intend to address the dynamic path issue of FTAs under a Grossman-

Helpman�s framework. However, they emphasize the endogenous con�gura-

tion of alternative free trade agreements in a model that allows transfers.

This paper, in contrast, treats FTAs as being con�gured exogenously, but

it explicitly introduces the possibility of industry exclusions to the agree-

ments - as Grossman and Helpman (1995) did. This, indeed, can somehow

be understood as a sort of transfers of privileges between trade partners,

especially delivered to the import-competing sectors.

In the tradition of Aghion et al. (2007), previously Riezman (1985, 1999)

and Burbidge et. al. (1997), among others, have also studied the endogenous

coalition formation of trade blocs, and their e¤ects on member and outside

countries.

Riezman (1985, 1999) has analyzed whether trade blocs help or hinder

the formation of a grand coalition. Riezman (1985) was in fact a pioneer

paper examining the formation of stable customs unions using the core as

a solution concept - although it precludes the possibility of inter-country

transfers. Riezman (1999) concludes, on the one hand, that banning bi-

lateral trade agreements can lead to more protection when there is a large

trading bloc along with some smaller ones. On the other hand, allowing

bilateral negotiations could lead to more protection when trading blocs are

initially of similar size. The �rst result appears from the fact that, in this

model, smaller trading blocs may use the threat of forming FTAs to prevent

a large country from imposing tari¤s in an original free-trade situation. The

second result arises because bilateral agreements allow blocs to more e¤ec-

tively monopolize world trade, thus making free trade less likely. Riezman�s

models have the limitation that equilibria are not derived analytically, and

so numerical methods are needed to get the results aforementioned.

Burbidge et. al. (1997) develop an alternative coalition formation game

7

in which the economies can capture e¢ ciency gains from policy coordination.

This paper �nds that with only two countries, the unique equilibrium struc-

ture is a fully coordinated area. However, with more than two, they show

that the grand coalition can fail to form even with side payments within

the area. Although the grand coalition always captures all e¢ ciency gains,

they �nd that there may be no point that constitutes a simultaneous Pareto

improvement for every alternative structure at the same time.

Within the literature of regionalism versus multilateralism, and in par-

ticular within the literature of endogenous coalition formation of FTAs, one

also �nds an interesting and recent work by Saggi and Yildiz (2006). They

develop an equilibrium theory of FTA formation deriving stable Nash equi-

libria of two non-cooperative games of trade liberalization between three

countries. In one game - the FTA game -, a subset of countries can form

FTAs without imposing restrictions on the resulting coalition structure,

whereas in the other game - the No FTA game -, each country must only

choose between no agreement and global free trade, thus forbidding the for-

mation of FTAs. They �nd both stumbling bloc and building bloc e¤ects,

and cases of partial building bloc in which FTAs can improve welfare when

multilateral liberalization is not attainable.

Notice that the equilibrium results found in the papers abovementioned

rely to a great extent on the di¤erent solution techniques applied in each

of them. For instance, Riezman (1985, 1999) uses the core solution concept

of coalitional games, Burbidge et. al. (1997) and Saggi and Yildiz (2006)

apply the concept of coalition-proof Nash equilibrium (CPNE) developed

by Bernheim et. al. (1987), while Grossman and Helpman (1995) uses the

CPNE concept in the �rst-stage game between lobbies and the government,

and a Nash bargaining solution concept in the second-stage game of negotia-

tions for exclusions between countries. Regarding Grossman and Helpman�s

paper, it is worth noting that the previous 1993�s NBER version describes

this inter-country game as a bargaining game in strategic form. That is,

whereas the �nal 1995�s AER version describes the bargaining game for

excluded industries using axiomatic approach, i.e, a Nash bargaining prob-

lem, the previous version uses a strategic approach - an alternating o¤ers

8

bargaining a-la Rubinstein (1982) and Binmore et. al. (1986).5

3 The Grossman-Helpman Model of FTAs

3.1 Analytical Framework

Consider two small countries and the rest of the world. Each country

charges non-discriminatory, most-favored-nation (MFN) tari¤s, following

WTO rules. There is a numeraire good 0 that is untaxed in each coun-

try, and n other goods. Initially, some of these goods are imported by one

or both of these countries, while others may be exported. Neither export

subsidies are allowed in the model, in recognition of WTO rules, nor are

export taxes. International prices are normalized to 1, thus domestic export

prices are simply 1. As for the import goods, these may be subject to import

tari¤s. Let � ij represent 1 plus the initial tari¤ rate on good i on country j,

for j = A;B. By normalizing international prices to 1, then these are the

domestic prices on import goods in country j.

Assume that, after the conclusion of an FTA, the initial external tari¤

levels continue to apply to imports from nonmember countries, once again

according to WTO rules which establish that these tari¤s must be no higher

than those imposed before the agreement. Article XXIV of the General

Agreement on Tari¤s and Trade (GATT) permits certain exceptions to the

principle of MFN, provided that the agreement covers �substantially all

trade�. Although the interpretation of this article could be somewhat mis-

leading, a �rst part of this model analyzes the viability of an FTA requiring

that all goods be freely traded between the members. A second part will

consider exclusions of sectors that might be sensitive to the agreement - the

term sensitive will be de�ned below.

Each country has a (voting) population of 1. Individuals in each country

5See, for example, Osborne and Rubinstein (1990, 1994) on bargaining and solutionconcepts in cooperative game theory.

9

have identical preferences

u(c) = c0 +nXi=1

ui(ci)

where ci denotes consumption of good i, and ui(�) is an increasing andconcave function. Consistent with these preferences, one gets per capita

demands for the n goods, Di(qi), and per capita demand for good 0, y �Pni�1 qiDi(qi), where qi is the domestic consumer price of good i and y is

the individual�s spending.

The production of a unit of good 0 requires only one unit of labor,

whereas each of the n goods is produced under constant returns to scale

with labor and a sector-speci�c factor. Since the domestic price of good 0 is

normalized to 1, the competitive wage must equal 1 provided there is some

production of this good initially. Denote with �i(pi) the pro�ts earned by

the speci�c factor in sector i, where pi is its domestic producer price. Denote

by Xi the aggregate supply of good i, which is given by Xi(pi) = �0i(pi) > 0

for each i.

Assume that the ownership of speci�c factors is highly concentrated, and

take the extreme case where these factor owners are a negligible fraction of

the population. Also, assume that they form a special-interest group (a

lobby) which takes political action so as to maximize their joint welfare.

The incumbent government has as its only function the decision on trade

policy, which in this case reduces to whether accept or reject the FTA.

Politicians in the government receive contributions from the lobbyists, who

seek to in�uence their decision. The government does not only value these

contributions, but also cares about aggregate welfare.

Assume that the government�s objective has a simple linear form

G =

nXi=1

Ci + aW

where Ci is the contribution of lobby i, W is aggregate welfare, and a

is a non-negative parameter re�ecting the government�s welfare maximizing

10

behavior relative to contributions. In turn, the individual�s surplus from

consuming good i is

Si(qi) � ui[Di(qi)]� qiDi(qi)

The consumer also receives a lump-sum transfer from the government

determined by the equally rebated, total tari¤ revenues. Finally, aggregate

welfare is given by

W = L+

nXi=1

�i(pi) +

nXi=1

(� i � 1)Mi +

nXi=1

Si(qi)

where the �rst term, L, is the aggregate labor supply, andMi are imports

in sector i. Remember that the wage is 1, then W will be the sum of labor

income, pro�ts, tari¤ revenues, and total consumer surplus.

Since factor owners capture only a negligible fraction of the consumer

surplus, and receive only a negligible fraction from tari¤ revenues, the ob-

jective of these owners may be approximated by �i(pi)�Ci, i.e., pro�ts netof contributions.

The domestic, political game comprises two stages. In the �rst stage,

lobbies make their move, o¤ering �nancial support - the contributions -

to the incumbent government. These contributions are directly linked to

the government�s policy, which here is simply the acceptability or rejec-

tion of the trade agreement. In the second stage, the government has two

choices: whether to pursue the FTA or not. In the end, a factor-owner

policy-contingent contribution schedule will only encompass two numbers:

CiN , representing the contributions in favor of the status-quo, and CiF ,

representing the contributions in favor of the FTA.

Finally, assume each lobby o¤ers in contributions no more than what it

stands to gain if the government were to follow its bidding.

3.2 E¤ects of an FTA

Consider the e¤ects of an FTA on the interests of the di¤erent economic

agents. To begin with, focus on a single industry i where, without loss of

11

generality, assume that �Ai > �Bi � 1.That is, the domestic price in country

A is greater than that in country B, which in fact is no lower than that

prevailing in the rest of the world. Figure 1 (shown in the appendix) depicts

the demand for imports by country A and three possible locations of country

B�s supply curve - locations that depend on the endowments of the speci�c

factor used in this industry.

Grossman and Helpman remark two e¤ects of an FTA on country mem-

bers. On the one hand, if country B�s supply of good i is relatively small

to su¢ ce country A�s import demand, as the case of XBi [1], then the equi-

librium producer price received by industries in the lower-tari¤ country will

rise towards the tari¤ rate �Ai . This is the case of enhanced protection. In

the opposite side, the endowment of speci�c factor in country B may be

large enough so that it su¢ ces to satisfy A�s import demand at the lower

price �Bi - situation depicted by supply curve XBi [3]. In this case, A�s import-

competing industries see their price falling as well as their pro�ts. This is

the case of reduced protection. There is also an intermediate case whenever

the supply in country B lies between the higher and the lower price, yet the

results are merely a combination of the two previous e¤ects.

In order to illustrate these two cases, consider now an industry that ex-

periences enhanced protection. Producers in B bene�t from their preferential

access to A�s market, whereas producers in A are una¤ected, given that their

domestic price does not change. Regarding the welfare in A, the only e¤ect

is the decrease in tari¤ revenues, which is an adverse e¤ect product of the

diverted trade from the more e¢ cient rest of the world to the partner B.

The total welfare change in A is

�WAi = �(�Ai � 1)XB

i (�Ai )

In country B, aggregate welfare necessarily rises because of two com-

ponents: �rst, pro�ts of the export sector increase given the higher price

received in market A; second, the country substitutes domestic supply with

international supply, thus gaining from an increase in tari¤ revenues. The

12

total welfare change in B is

�WBi = ��Bi + (�

Bi � 1)XB

i (�Bi )

It is straightforward to show that the welfare loss in A exceeds the welfare

gain in B, which reveals the global e¢ ciency cost of the trade diversion e¤ect.

As for the reduced protection case, one can see that exporters in B gain

nothing from the agreement, while producers in A bear the increased import

competition. Since all imports in A come from B now, tari¤ revenues from

this sector fall to zero. However, average voters in this country gain as

consumers, since the domestic price in A decreases toward �Bi . The change

of aggregate welfare in A is

�WAi = ��

Ai � (�Ai � 1)MA

i (�Ai ) + S

Ai (�

Bi )� SAi (�Ai );

which may be positive or nonpositive.

Country B receives only the extra tari¤ revenue generated by the partial

substitution of domestic supply with foreign supply. Welfare change in B is

�WBi = (�Bi � 1)MA

i (�Bi )

It is easy to see that welfare gains for the FTA members are assured in

the reduced protection case. This arises from the trade creation e¤ect of

trading blocs in this set up.

About the intermediate case, it simply combines features of the �rst two

scenarios, and the result is ambiguous in terms of the joint welfare of the

bloc.

3.3 Equilibrium FTAs Without Exceptions

First in this sub-section I look for conditions under which an FTA will

be unilaterally supported in each country. Then, I study the negotiations

between countries that may or may not lead to an equilibrium agreement.

Suppose that no exclusions (or side-payments) are allowed, so that ei-

ther all industries are included in the FTA or the status quo prevails. Un-

13

der this setting, an FTA occurs only if both countries unilaterally favor it.

Thus, consider �rst in each country a game between lobbies and the govern-

ment determining the conditions for a unilateral stance that may favor the

agreement (outcomes will be denoted with subscript F ), or the status quo

(outcomes will be denoted with subscript N ).

I will skip here the technical discussion made in Grossman and Helpman

(1995) with respect to the description of the problem in terms of �pressured�

and �unpressured�stances�.6 Instead, I will apply a simpli�ed model devel-

oped by Goldberg and Maggi (1999), who use a Nash bargaining problem. In

this version, unilateral stance equilibrium results simply from the maximiza-

tion problem of the joint surplus of the government and domestic lobbies.

Under this setting, the government unilaterally will favor an FTA if and

only if, Xi�iF + aWF �

Xi�iN + aWN (1)

In words, the government will simply favor the agreement if the �politi-

cal�welfare under the FTA exceeds the �political�welfare under the status

quo. It is certainly possible that inequality (1) fails to hold in either of

the two countries, making the FTA not feasible. Grossman and Helpman

(1995) shows that the viability of an FTA that does not allow for industry

exclusions depends on how balanced is the trade between potential members

of the agreement. Following Gawande et. al. (2005), one can summarize

this statement with the following proposition: an FTA among countries is

most likely when trade between them is su¢ ciently �balanced�. In showing

this result, Grossman and Helpman consider a special case with particular

functional forms. I will work on that example below when I apply it to the

three-country setting instead.

3.4 Equilibrium FTAs With Exceptions

Grossman and Helpman (1995) shows how industry exclusions may make

an otherwise unfeasible FTA politically viable. First, consider the game

6These stances depend on whether the government chooses a position directly a¤ectedby domestic lobbies or it just makes independent decisions.

14

between lobbies and the government in which a unilateral stance for an

FTA with exclusions is determined.

Denote as �(T ) the set of politically optimal exclusions, where T is as-

sumed to be the maximum number of excluded industries.7 If there was no

such a restriction T , then all industries having the joint welfare of lobby

and government higher in status quo than under FTA belong to that set.

Denote the change in joint welfare of sector i as

gi = ��i + a�Wi

where ��i = �iF ��iN and �Wi =WiF �WiN . This way, the set �(T )

comprises all industries for which gi < 0.

Now, consider the case where there is in fact a constraint T on the

number of exclusions. One might intuitively think that when this constraint

binds, then the T excluded industries would be those with the most negative

joint welfare change. Grossman and Helpman (1995) shows that a unilateral

stance�s FTA exists if and only ifXi2"(�iN + aWiN ) +

Xi=2"(�iF + aWiF ) �

Xi

(�iN + aWiN ) (2)

The main conclusion one can extract from this result is that while (1)

may fail to hold, inequality (2) can be satis�ed. That is, where a unilateral

stance favoring and FTA without exclusions may fail, a unilateral stance for

an FTA in which the most sensitive industries can be excluded may favor

the agreement. Following Gawande et. al. (2005), one can rewrite (1) asPi(��i + a�Wi) � 0, and (2) as

Pi=2�(��i + a�Wi) � 0. Since the latter

excludes the sectors with most negative joint welfare changes, then it is clear

that inequality (2) is easier to be satis�ed.

In this stance, import-competing industries are the most politically sen-

sitive sectors in the economy, thus they are the �rst candidates to push the

government for being excluded from the agreement. In contrast, industries

7This restriction, which in a sense is obeying WTO rules, can be more general. Forinstance, it could be a limit on the fraction of excluded trade.

15

that can eventually export to the partner country at the expense of the rest

of the world are the most favored by the agreement, and then they make

contributions to the government in order to be included.

Consider now the next stance where both governments negotiate for the

agreement. Assuming that unilateral stances in each country favored an

agreement, in this stage both governments bring their lists of exclusions to

the bargaining table in order to get a favorable FTA with their sensitive

sectors excluded. In this sense, lobbies anticipate each possible outcome in

making their bids.

Now, the question is which sectors will be granted exclusions in an equi-

librium agreement? To answer this question, Grossman and Helpman treat

the negotiation process as a cooperative bargaining game, and apply a simple

Nash bargaining solution. This way, the equilibrium agreement is designed

to maximize a weighted average of the surpluses of the two negotiating gov-

ernments.

Since the governments always have the option to reject the agreement,

their surpluses are calculated with respect to the status quo. Technically, the

�threat point�in the Nash problem will be the joint welfare of government

and domestic lobbies under the pre-FTA scenario.

The equilibrium FTA can be characterized by a set of indicator variables

�i, where �i = 1 implies that sector i is excluded from the agreement, and

�i = 0 implies that sector i is included. The Nash bargaining solution solves

the following problem,

maxf�ig

Yj=A;B

"Xi

[(1� �i)(CjiF + aWjiF ) + �i(C

jiN + aW

jiN )]�G

j

#�j(3)

s:t:X

i�i � T

where �j is the Nash weight attached to the surplus of government j,

and Gj=Pi(C

jiN + aW

jiN ) is the political welfare of government j in the

status quo (the Nash�s threat point). One can �nd a solution to this problem

16

treating �i as continuous between 0 and 1, and rewriting (3) in log terms.

Then, the �rst order conditions are

0,1for

0for])()([

i

i

,

≥

=−≤

=−≥+−+

−∑=

λ

αλ

αλβ

and

aWCaWCGGBAj

jiN

jiN

jiF

jiFjj

j

where Gj is the equilibrium political welfare of government j in the bar-

gaining problem - the �rst term of the summation in problem (3), evaluated

at the equilibrium -, and � is the Lagrange multiplier associated with the

restriction over exclusions.

It follows that exclusions are granted to industries for which the weighted

di¤erence between the political bene�ts in the exporting country and the

political costs in the importing country is most negative. It would be even

more appealing if one can identify the sector excluded through an order-

ing that depends only on aggregate conditions and on supply and demand

characteristics. The following result takes into account this advantage: let

the industries be ordered so that !AgAi + !BgBi is increasing with i, where

!j = �j=(Gj�Gj). Then the FTA that solves the Nash bargaining problemexcludes all industries i 2 [0; i�], for some i� � 0 that depends, among otherthings, on T .

This result reveals that the same factors that determine the politically

optimal set of exclusions in each country at a unilateral stance also enter

into the con�guration of the set of exclusions in the negotiation between

partners. The following section compares these results with the results one

can �nd in a three-country model.

4 The Multilateral Trade Agreement

In the previous section, where we revisited Grossman and Helpman (1995),

we considered a world comprised by two small countries and the rest of the

world. Now, instead of working with the rest-of-the-world assumption, ex-

17

plicitly consider a world of three countries, A, B, and C. In what follows, I

will study the viability of a trade agreement among these countries, assum-

ing that countries A and B are already members of an FTA. My goal here

is to analyze the consequences of using Grossman-Helpman�s framework in

studying the dynamic path issue of trading blocs. One can see throughout

this section that the formal step taken from a two-country agreement toward

an agreement with three countries simply implies slight changes. Nonethe-

less, I will show in the next section an illustration in which the e¤ects of

including a third partner are not trivial.

As in the preceding section, I begin this section with the analysis of a

trade agreement without exceptions. In particular, consider the game be-

tween lobbies and government. At this stage, a unilateral stance equilibrium

without exclusions in each country results, as before, from the maximiza-

tion problem of the joint surplus of government and domestic lobbies. That

is, governments A and B will unilaterally favor a multilateral agreement

(outcomes are denoted with subscript M) if and only if,Xi�jiM + aW j

M �X

i�jiF + aW

jF (4)

for j = A;B

i.e., if and only if the joint welfare of government and lobbies in the

two countries is greater under the multilateral liberalization than under the

FTA. As for country C, this country will unilaterally favor a multilateral

agreement if and only if,Xi�CiM + aWC

M �X

i�CiN + aW

CN (5)

i.e., country C will favor an agreement if and only if the joint welfare of

government and lobbies is greater under the multilateral liberalization than

under the status quo.

The analysis of the viability of a multilateral agreement with no excep-

tions turns out to be straightforward when we already have an exogenous

FTA between two partners. Now, consider the equilibrium multilateral trade

18

agreement with industry exclusions. In particular, take �rst the game be-

tween lobbies and government in a unilateral stance for the grand accord.

Denote by E(T ) the set of politically optimal exclusions to the multi-

lateral agreement. This set may di¤er with respect to the previous set of

exclusions from the FTA - the set �(T ).8 Continue denoting the change in

joint welfare as gi = ��i + a�Wi, where the changes are expressed with

respect to the FTA in countries A and B, and with respect to the status quo

in C.

It is straightforward to replicate inequality (2) �which shows the con-

ditions for a unilateral stance�s bilateral agreement with exclusions - to the

multilateral case. This way, we have that the multilateral liberalization is

favored in countries A and B if and only ifXi2E(�j

i eN +aW j

i eN )+Xi=2E(�jiM +aW

jiM ) �

Xi2�(�jiN +aW

jiN )

Xi=2�(�jiF +aW

jiF );

(6)

for j = A;B.

As for country C, this country favors the agreement if and only ifXi2E(�CiN + aW

CiN ) +

Xi=2E(�CiM + aWC

iM ) �Xi

(�CiN + aWCiN ) (7)

The subscript � eN�in expression (6) is used to di¤erentiate between theoutcomes of the industries excluded from the multilateral agreement and

the outcomes of the industries already excluded from the FTA �which are

expressed with the subscript �N�. Notice that a particular sector in coun-

tries A or B can contribute in two ways for an exception to the multilateral

agreement. It can contribute to continue behaving as in the status quo pre-

vailing before the FTA was approved �as an industry already excluded from

the FTA -, or it can contribute to stay �inside�the FTA �which still is a

possible scenario if the multilateral agreement is rejected. Once again, one

can see that while inequality (5) for country C may fail to hold, inequality

8Assume, without loss of generality, that the number of restrictions T is the same asthe one for the FTA.

19

(7) can be satis�ed if we allow for the most sensitive industries to be ex-

cluded. Notice that this may not be the case for countries A and B, since

inequality (6) introduces two sets of exclusions, one arising from the FTA

and the other arising from the multilateral agreement.

Consider now the next stage where all the governments bargain for the

multilateral agreement. As in Grossman and Helpman (1995), each gov-

ernment exposes their lists of exclusions at the bargaining table, and the

question is about the sectors that will be granted exclusions. As in the pre-

vious section, I will analyze the negotiation process through the implemen-

tation of a Nash bargaining problem. It has been proved that the properties

of a Nash bargaining solution with two players continue to apply in an n-

player bargaining game.9 Then, the equilibrium agreement will be designed

to maximize a weighted average of the �surpluses�of the three negotiating

governments.

Since the governments always have the option to reject the agreement,

their surpluses are calculated with respect to the FTA, in the cases of coun-

tries A and B, and with respect to the status quo in the case of country

C. In other words, the �threat points�in the Nash problem will be just the

joint welfares of government and domestic lobbies before the multilateral

liberalization is negotiated.

Once again as in Grossman and Helpman (1995), the equilibrium agree-

ment can be characterized by a set of indicator variables �i, where �i = 1

implies that sector i is excluded, and �i = 0 implies that sector i is included.

Now, the Nash bargaining problem becomes one of solving,

maxf�ig

Yj=A;B;C

hGj(�i)�G

ji�j

(8)

s:t:X

i�i � T

9For a discussion on this topic, see Krishna and Serrano (1996).

20

where,

Gj(�i) =Pi[(1� �i)(C

jiM + aW j

iM ) + �i(Cj

i eN + aW j

i eN )] for j = A;B

Gj(�i) =Pi[(1� �i)(C

jiM + aW j

iM ) + �i(CjiN + aW

jiN )] for j = C

Gj=Pi2�(C

jiN + aW

jiN ) +

Pi=2�(C

jiF + aW

jiF ); are the threat points for

j = A;B

Gj=Pi(C

jiN + aW

jiN ); is the threat point for j = C:

Then, the �rst order conditions are

Pj=A;B

�j

Gj�Gj[(CjiM + aW j

iM )� (Cj

i eN + aW j

i eN )]+�C

GC�GC[(CCiM + aWC

iM )� (CjiN + aW

jiN )] � ��

for �i = 0, Pj=A;B

�j

Gj�Gj[(CjiM + aW j

iM )� (Cj

i eN + aW j

i eN )]+�C

GC�GC[(CCiM + aWC

iM )� (CjiN + aW

jiN )] � ��

for �i = 1, and � � 0; with � being the Lagrange multiplier associatedwith the restriction over exclusions.

Once again, one can see that exclusions are granted to industries for

which the weighted di¤erence between the political bene�ts in the exporting

countries and the political costs in the importing countries is most negative.

I further look for identifying the industry exclusions by performing an order-

ing that depends only on aggregate conditions and on supply and demand

characteristics. Consider the following result extended from Grossman and

Helpman (1995): let the industries be ordered so that !AgAi +!BgBi +!

CgCi is

increasing with i, where !j = �j=(Gj�Gj): Then the multilateral agreementthat solves the Nash bargaining problem excludes all industries i 2 [0; i�],for some i� � 0.

As in the two-country case, one can see that the same factors that de-

termine the politically optimal set of exclusions at a unilateral stance also

determine the set of exclusions in the multilateral negotiation process.

21

4.1 E¤ects from FTA and Multilateralism

In this sub-section, I will brie�y study the e¤ects of both an FTA and

multilateral liberalization on international prices and quantities. Consider a

particular industry i whose goods are imported by countries A and B from

country C. Figure 2 revisits Figure 1. Now, denote the world price as p,

and as before �Ai and �Bi represent 1 plus the tari¤ rate in each of these

countries. Without loss of generality, assume again that �Ai > �Bi � 1, then

the consumer and producer prices in A, p�Ai , exceed those of B, p�Bi .

In the �rst place, suppose that country B�s supply is XBi [3] in Figure

2. I will examine now how the joint excess demand of countries A and B

changes as a result of the FTA between them. One can see that a case of

reduced protection would lead to an increase in A�s imports demand, which

is satis�ed only by B. In turn, B�s demand now is satis�ed by products from

country C (trade creation e¤ect). However, consumer prices do not change

in B, and then B�s net import demand does not change at all.

Figure 3 depicts the impact both of the formation of an FTA and of

multilateral agreement on international prices and equilibrium quantities,

under the reduced protection case. At the original international price p, the

joint import demand of A and B rises after the formation of the FTA, which

sequentially leads to a higher international price pF . Then, the FTA implies

a positive externality on country C, given that it improves this country�s

terms of trade.

Consider now the e¤ects of multilateral liberalization between the three

countries. Given the assumption that country C is the exporter of good i,

the expected e¤ect of the grand accord is that the imports demand curve

from countries A and B (DA+Bi �XA+Bi ) shifts to the right, increasing the

international price toward pM . This again favors country C.

Suppose now that country B�s supply isXBi [1]. This is a case of enhanced

protection that leads to trade diversion within the bilateral agreement mem-

bers. The FTA between A and B does not change their consumer prices, and

does not change country A�s producer price. However, it does change the

producer price in B. This higher price su¢ ces to reduce the total demand

22

for imports of the two countries A and B. As a result, p falls toward pF ,

thus worsening C�s terms of trade and reducing this country�s welfare (see

Figure 4, which depicts the impact both of the formation of an FTA and of

multilateral agreement, under the enhanced protection case)

Consider now the e¤ect of multilateral liberalization between the three

economies. Given the assumption that country C is the exporter of good i,

in this case the multilateral agreement will cause the joint import demand

of A and B to increase �remember that multilateral liberalization implies

the reduction of domestic prices toward world prices. Since this increase

in demand generates an import demand expansion that will ultimately be

greater than the fall generated by the trade diverting FTA, then the inter-

national price will rise towards pM , which is found above the international

price of the status quo.

5 An Illustration

Consider a special case, originally introduced in Grossman and Helpman

(1995), where particular functional forms are speci�ed. Under a three-

country setting, I will analyze the determinants of the viability of multi-

lateral liberalization, along with those determinants I have studied before

on the viability of the formation of FTAs.

Suppose that households in the three countries share identical utility

functions, where all the non-numeraire goods enter symmetrically and each

ui(�) is quadratic. Then, the aggregate demand for any good i in country jhas the linear form

Dji (qji ) = D � bq

ji

for i = 1; : : : ; n and j = A;B;C, where D and b are both positive

parameters.

Assume that supply is inelastic in countries A and B, and symmetrically

23

described as

∗∗−

=

∗−∗

=

industries)­(1inindustriesin)1(

industries)­(1in)1(industriesin

nsXnsX

X

andnsX

nsXX

Bi

Ai

θθ

θθ

where s indicates the fraction of a type of industries in each country,

and � is a parameter that measures output imbalance in any one sector,

i.e., the relative size of output in any sector, each country. Without loss of

generality, take � > 1=2 and s � 1=2. Assume that country C is a large

exporter of all goods, and that it can supply any quantity at given world

prices standardized at 1. This assumption is introduced into the illustration

in order to avoid terms of trade e¤ects created by externalities that may

emerge from the formation of any agreement.

The illustration consists of two stages. In the �rst stage, I will analyze

the economic e¤ects of an FTA between A and B. As one expects, this does

not a¤ect country C�s welfare since the international price is unchanged.

Later, I will focus on the viability of multilateral agreement between A, B,

and C. It is clear that, once again, this will not a¤ect outcomes in C, thus this

country will be indi¤erent toward the agreement. Roughly, the idea here is

to evaluate what are the chances that countries A and B favor a multilateral

liberalization. Finally, I will also contemplate the case in which a set of

industries are allowed to be excluded from this multilateral agreement.

The viability of any agreement, whether bilateral or multilateral, de-

pends on the initial structure of MFN tari¤s. In this sense, assume that

tari¤s initially protecting the import-competing industries in each economy

result from a lobbying game as derived in Grossman and Helpman (1994),

and assume that politicians in each country place the same weight a on

aggregate welfare. Then, given the assumptions above one gets that the

domestic prices in A and B, will be such that

� ji = 1 +Xji

ab(9)

24

for j = A;B, and for all sectors in which imports are positive in the

initial equilibrium. As for country C, remember that being the exporter in

all goods implies that �Ci = 1 for all i.

Grossman and Helpman (1995) emphasizes that di¤erent types of out-

comes may emerge in this example depending on the parameter values. Ac-

cording to this idea, they examine three di¤erent sets of parameter restric-

tions illustrating some relevant cases that may appear.

We will focus on one of these sets of parameters in order to study the

case in which an FTA certainly emerges in the �rst step. This is the case

in which all sectors experience enhanced protection. With respect to the

remaining sets of parameters, we have that all sectors experience reduced

protection. Under this e¤ect, Grossman and Helpman show that an FTA is

not feasible, even when one allows for exclusions. Having this in mind, it is

clear that none multilateral liberalization is possible in those two remaining

cases, as they will also imply reduced protection in all industries, for both

countries A and B. Therefore, in what follows I will work on the enhanced

protection case.

5.1 Viability of an FTA

Suppose that A and B start negotiating on an FTA, and consider the fol-

lowing restriction on parameters10

(1� �)X < D � b�1 +

�X

ab

�(10)

Under this condition, all n goods are imported in both countries in the

initial equilibrium. Country A has the higher import tari¤ in the fraction s

of industries where its supply is �X, while country B has the higher import

tari¤ in the remaining sectors. Under an FTA, each country will import

from its partner all goods on which their MFN tari¤s are higher. Then, A

would import from B the fraction s of goods, while B would import from

A the remaining goods. Also, because of restriction (10), the output in the

10See Grossman and Helpman (1993, 1995) for more details.

25

low-tari¤ country would not su¢ ce to satisfy import demand in the partner

country at its pre-FTA domestic price. As a result, all sectors will experience

enhanced protection.

We can calculate the contribution of both types of industry to the change

in the joint welfare of government and lobbies from the agreement. These

are

a�WA +Xi

��Ai = �sn��(1� �)X2

b

�+

(1� s)n��(1� �)X2

b+(2� � 1)(1� �)X2

ab

�

a�WB +Xi

��Bi = �sn��(1� �)X2

b+(2� � 1)(1� �)X2

ab

�+

(1� s)n��(1� �)X2

b

�Given these expressions, one �nds that an FTA is supported in country

B for every s � 1=2, while country A will favor the agreement if and only if

s � 1

2+

� � 1=22� � 1 + 2a� < 1 (11)

Notice that inequality (11) will only be satis�ed for s su¢ ciently close

to 1=2.

Consider now the possibility that governments grant exclusions to a num-

ber of sensitive industries. The Grossman-Helpman�s approach provides,

through this illustration, the idea that exceptions can save an otherwise

unfeasible FTA. Consider the proposal to exclude a number EA of sectors

in which A would import from B, and a number EB of sectors in which B

would import from A, and let EA + EB � E be the WTO rule.

Recalculating the joint welfare change of government and lobbies in coun-

try A under the possibility of exclusions, we get the following range of values

26

of s for which an FTA would be feasible,

s � 1

2+

� � 1=22� � 1 + 2a� +

a�EA � (2� � 1 + a�)EBn(2� � 1 + 2a�) (12)

It can be shown that if EA > (2� � 1 + a�)EB=a�, then inequality (12)implies an expansion of the range of s for which the agreement is accepted

by A. If also EA < [EBa� + (2s � 1)na� + sn(2� � 1)]=(2� � 1 + a�), thenthe FTA with exclusions is supported by government B. This is roughly

illustrating he fact that there are values of s for which the FTA would be

rejected without sectors exclusions, but favored with them. In particular,

one can show that if (2s � 1) < E=n, then the FTA with EA = (2s � 1)nand EB = 0 satis�es WTO rules and politically dominates the status quo.

However, this is not the only set of exclusions that allows for an equilibrium

agreement.

So far, I have revisited an example taken by Grossman and Helpman

(1995). In this illustration, there are cases in which an FTA would fail

if all trade has to be included, while it can be preferred if exclusions are

allowed in both countries. In cases like this, one observes that an FTA is

saved if the agreement allows for exclusions, but still it is necessary that

both governments engage in an e¢ cient negotiation process. Grossman and

Helpman (1993) suggests that this bargaining process can be modeled as

an alternating bargaining game a-la Rubinstein (1982) or Binmore et. al.

(1986). But as I mentioned above, Grossman and Helpman (1995) solves this

bargaining problem using a relatively simpler Nash bargaining solution. In

any way, one can be safe from this particular issue, just noticing that there

are di¤erent alternatives assuring that a certain negotiation process will lead

to an equilibrium set of exclusions, ultimately guiding the acceptance of the

agreement in each country.

5.2 Viability of the Multilateral Agreement

Now, I will study the viability of the multilateral agreement between A, B,

and C, taking as given that an FTA between A and B has been formed

27

already. For this purpose, I will continue assuming that condition (10)

holds, which implies that the bilateral agreement arises from (and creates)

enhanced protection.

Notice that all industries in countries A and B will experience reduced

protection after the multilateral agreement, since all goods are assumed

to be exported by the low-price country C. As a result of the multilateral

liberalization without exclusions, the contribution of both types of sectors

to the change in the joint welfare of country A is

a�WA +Xi

��Ai = (sn� EA)�X2

�2a(1� �)� �

2ab

�+ (13)

[(1� s)n� EB](1� �)X2

�1� �(3 + 2a)

2ab

��

EA�(�X)2

2ab

�� EB

�[(1� �)X]2

2ab

�where EA and EB are the sets of exclusions from the �rst stage-FTA.11

With respect to country B�s outcome, it will simply be the mirror image

of (13). I should remark a few points from this expression. Notice that both

sets of previously excluded industries now provide a negative contribution

to the country�s joint welfare change (last two terms of [13]). The second

term in (13) is always negative, since � > 1=2. The �rst term in (13) can

be positive for a high enough a, that is for a su¢ ciently welfare maximizer

government A. Finally, I will show below that all these terms put together

result in a negative joint welfare change not only for country A but also for

country B, for any previously con�gured set of FTA exceptions.

Suppose that the equilibrium FTA that results from the �rst stage bar-

gaining process between A and B gives the sets of exclusions aforementioned:

11This general formulation also allows to consider the without-exclusions case from theFTA.

28

EA = (2s� 1)n and EB = 0. Then, expression (13) becomes

a�WA +Xi

��Ai =nX2

2abf(1� s)[�(2a(1� �)� �) +

(1� �)(1� �(3 + 2a))]� (2s� 1)�2g

which implies that an agreement will be favored by government A if and

only if

s[�2 + (1� �)(1� 3�)] < (1� �)(1� 3�) (14)

However, inequality (14) does not hold for any positive value of s. Gross-

man and Helpman (1993) shows that in general a range of equilibrium sets of

exclusions on the bargaining between governments A and B can be found be-

tween EAmin = (2s�1)n and EAmax = (2s�1)n+[(1�s)n(2��1)]=(a�+2��1),and EB = 0. Looking at expression (13), it is straightforward to note that

with the least number of exclusions, EA = (2s�1)n, one obtains the higher �although negative �joint welfare change, which implies that the multilateral

agreement is not desired by government A for any fraction s of industries,

and for any con�guration of sets of exclusions carried from the FTA.

Moreover, neither country B will support the agreement under any al-

ternative of exclusions from the FTA stage. Gathering all the terms accom-

panying EAon the mirror image of (13), one gets the joint welfare change in

country B expressed as

�EA(1� �)X2[2� �(4 + 2a)] < 0, for all EA

where, remember, I continue assuming that � > 1=2 and a � 0. This bringsthe �rst result from this illustration.

Result 1: The multilateral trade agreement without exceptions is not

favored either in A or B, for any set of exceptions that may have arisen in

the equilibrium FTA.

Consider now the possibility of having exclusions to the multilateral

29

agreement. In this case, the WTO rule would admit EA + EB + EC �E. However, country C will be indi¤erent upon the agreement as it is the

exporter country, thus one can set EC = 0.

According to the previous discussion, one can assume without loss of

generality that the �rst-step negotiation for the FTA provided the sets of

exclusions studied above: EA = (2s � 1)n and EB = 0. Now, consider

expression (13) modi�ed to allow some industries to be excluded from the

multilateral agreement. In this case, the joint welfare of country A becomes

abXEns

abaXEns

abaXEnsWa

E

B

A

i

Ai

A

2)(])12[(

2)23(1)1]()1[(

2])1(2[(])1[(

2

2

2

θ

θθ

θθθ

−−

−

+−

−−−

+

−−−−=∆Π+∆ ∑

(15)

where, with a slight abuse of notation, EE is the set of new exclusions to

the multilateral agreement out of the set of exclusions from the FTA stage.

One expects that the industries excluded from the original FTA continue to

be excluded from the multilateral liberalization, given their relatively high

cost to the welfare of both A and B. Thus, set EE = (2s � 1)n. Now, itis straightforward to check that there is no any EA and EB such that both

expression (15) for country A and its mirror image for country B (not written

here) are positive at the same time. This can be shown by rearranging (15)

and writing its symmetric expression for country B. This way, one can �nd a

combination of sets of exclusions such that both A and B favor the agreement

if and only if the following two conditions are satis�ed

EA�[2a(1� �)� �] + EB(1� �)[1� �(3 + 2a)] > 0

and

30

EA(1� �)[1� �(3 + 2a)] + EB�[2a(1� �)� �] > 0

Now, divide both inequalities by EB and rearrange to get

z[1� 2(2� �)�] + 1� 2(2� �)� < 0

for all � > 1=2, and z = EA=EB > 0, which leads to the second result from

this illustration.

Result 2: The multilateral trade agreement with industry exclusions is

not favored either in A or B, for any set of exceptions that may have arisen

in the equilibrium FTA.

This result comes up for the reason that the sum of the �rst and second

terms in (15) is negative if one takes away EA and EB, thus any increase in

the set of exclusions that may favor the agreement in A leads to a rejection

by country B, and vice versa. Intuitively, this result comes from the fact

that both countries experience reduced protection in all industries.

Now the question is, was the multilateral liberalization feasible before

countries A and B engage in an agreement? In other words, did the FTA

between these two countries impede the multilateral agreement? In what

follows I will analyze the viability of the multilateral agreement starting

from the status quo.

Beginning from a position where no country has a trade agreement, the

multilateral liberalization without exceptions leads to the following changes

in the joint welfare of both countries

a�WA +Xi

��Ai = �sn�X2

2ab� (1� s)n [(1� �)X]

2

2ab< 0

a�WB +Xi

��Bi = �sn[(1� �)X]2

2ab� (1� s)n�X

2

2ab< 0 (16)

31

As the expressions in (16) show, the multilateral agreement is not pre-

ferred to the status quo in any of the two countries, while country C is

still indi¤erent. Once more, the rejection to the agreement comes from

the fact that countries A and B experience reduced protection in all their

industries. In this sense, even though aggregate welfare increases in both

countries product of the fall in prices, the decrease in pro�ts of all import-

competing industries more than su¢ ces to contribute with negative joint

welfare changes. Indeed, notice that if one allows for exclusions to the grand

accord, rejection results do not change. This brings the third result from

this illustration.

Result 3: The multilateral trade agreement, with or without industry ex-

clusions, is not feasible starting from the status quo. In this sense, and given

results 1 and 2, there are political conditions for the existence of a partial

building bloc, i.e., an FTA can be feasible when the multilateral agreement

is not.

Summarizing, I have shown an example in which an FTA with enhanced

protection in all industries is possible between A and B, for certain values

of the fraction s of industries. Moreover, we have seen that this range of

s expands when we allow for a number of exclusions in sensitive import-

competing sectors.

Afterwards, I have analyzed the viability of a multilateral liberalization.

It is clear that this exercise does not require country C being active in the

negotiations as it is indi¤erent upon any agreement. However, this setting

allows one to think on the possibilities that arise for countries A and B in

coordinating a reduction of their tari¤s. As we have seen, the conditions

here do not provide a good reception for the agreement in any of both A

and B (with or without exclusions), and the main reason is the presence of

a reduced protection e¤ect at all sectors, in both economies. In addition, we

have seen that global free trade neither was feasible from the status quo.

In this sense, I can argue that under a Grossman-Helpman framework there

exist conditions for which a partial building bloc emerges.

32

6 Concluding Remarks

Throughout this paper there has been a broad issue that I had in mind

from the beginning. This issue is related to the sources that have led to the

proliferation of numerous preferential trade agreements nowadays.

In looking to address this subject, I have initially departed from Gross-

man and Helpman�s (1995) model of FTAs. This model was designed to

explain the viability of a free trade agreement between two small countries

under a framework in which governments not only maximize aggregate wel-

fare, but also take into account contributions from domestic special-interest

groups. Moreover, they study the conditions that emerge for this agree-

ment to be politically viable in each country considering a certain number

of exceptions on sensitive industries.

I have taken this approach built for explaining FTAs, and have extended

it by introducing a third party into the analysis so as to study the viabil-

ity of a grand accord. In doing so, one may notice that an extension that

merely includes a third country into Grossman-Helpman�s framework pro-

vides straightforward results in the sense that it is too general to extract

much on our main question. However, one might guess that �nding an agree-

ment that be accepted with exclusions in all the three countries is far more

less likely than for the two-country case, and thus the set of outcomes in

which an agreement is favored on a multilateral bargaining must shrink.

These theoretical conjectures are not proved here and are left for further

research. However, I have studied in Section V an illustration with speci�c

functional forms which follows directly from a setting analyzed by Gross-

man and Helpman (1995). The idea was to extend that setting, originally

designed to �nd conditions for the viability of a bilateral agreement, to a

setting in which the focus is put on the viability of multilateral liberalization.

For a set of restrictions on the parameters, reduced protection e¤ects

arise in all industries once the FTA is formed. In these cases, results are

not appealing in the sense that no a bilateral agreement emerges in the �rst

stage, nor multilateral agreement is feasible from the status quo. This is

happening because of the symmetry of the two countries negotiating for an

33

FTA in the �rst stage, and because reduced protection leads to a situation in

which industries successfully pressure the government in favor of the status

quo.

For another set of restrictions, enhanced protection arises under the FTA.

In these cases, there exist alternatives in which an FTA is possible, and

furthermore these alternatives expand when exclusions are permitted. How-

ever, I have shown that at a second-stage multilateral agreement is not viable

under any circumstance, neither without exclusions nor with them. Indeed,

multilateral liberalization is not even feasible starting from the status quo.

Finally, I have identi�ed a case where an FTA acts as a partial building bloc,

that is, when the grand accord cannot be feasible under any scenario, the

FTA can be feasible.

Concluding, it is true that more illustrations and a deeper analysis of

this Grossman-Helpman�s framework for the multilateral trade agreement

are needed to study the dynamic path question of trading blocs. Neverthe-

less, I believe I have made a step forward in providing an illustration so

that we have a �avor on how Grossman-Helpman�s approach would work in

answering this dynamic path question.

34

References

[1] Aghion, P., P. Antras, and E. Helpman, Negotiating Free Trade, Journal

of International Economics, 73(1) (2007), 1-30.

[2] Bernheim, B., B. Peleg, and M. Whinston, Coalition-Proof Nash Equi-

libria I: Concepts, Journal of Economic Theory 42(1) (1987), 1-12.

[3] Bhagwati J., The World Trading System at Risk. Princeton, NJ: Prince-

ton University Press (1991).

[4] Bhagwati, J., Regionalism and multilateralism: an overview. In: J. de

Melo and A. Panagariya, Editors, New dimensions in regional integra-

tion, Cambridge University Press, Cambridge (1993).

[5] Bhagwati J., US Trade Policy: The Infatuation with Free-Trade Areas,

in The Dangerous Drift to Preferential Trade Agreements. J. Bhawati

and A. Kruger, eds. Washington, DC: American Enterprise Institute

for Public Policy Research (1995).

[6] Binmore, K., A. Rubinstein, and A. Wolinsky, The Nash Bargaining

Solution in Economic Modeling, Rand Journal of Economics 17, (1986)

176-188.

[7] Bohara, A., K. Gawande and P. Sanguinetti, Exclusions for sale: Ev-

idence on the Grossman-Helpman Model of Free Trade Agreements,

mimeo (2005).

[8] Burbidge, J., J. DePater, G. Myers, and A. Sengupta, A coalition-

formation approach to equilibrium federations and trading blocs, Amer-

cian Economic Review 87(5) (1997), 940-956.

[9] Goldberg, P. and G. Maggi, Protection for Sale: an Empirical Investi-

gation, American Economic Review 89 (1999), 1135-1155.

[10] Grossman, G. and E. Helpman, The Politics of Free-Trade Agreements,

NBER Working Paper No. W4597 (1993).

35

[11] � � � , Protection for Sale, American Economic Review, 84 (1994),

833-850.

[12] � � � , The Politics of Free-Trade Agreements, American Economic

Review, 85 (1995), 667-690.

[13] Krishna, P., Regionalism and multilateralism: A political economy ap-

proach, The Quarterly Journal of Economics 113 (1998), pp. 227�252.

[14] Krishna, V. and R. Serrano, Multilateral Bargaining, Review of Eco-

nomic Studies 63 (1996), 61-80.

[15] Levy, P., A political�economic analysis of free-trade agreements, Amer-

ican Economic Review 87 (1997), 506�519.

[16] Osborne, M. and A. Rubinstein, Bargaining and markets, Academic

Press (1990), San Diego, CA.

[17] Osborne, M. and A. Rubinstein, A course in game theory, MIT Press

(1994), Cambridge, MA.

[18] Ornelas, E., Rent Destruction and the Political Viability of Free Trade

Agreements, Quarterly Journal of Economics 120(4) (2005), 1475-1506.

[19] Panagariya, A., Preferential Trade Liberalization: The Traditional The-

ory and New Developments, Journal of Economic Literature, Vol.

XXXVIII (2000), 287-331.

[20] Riezman, R., Customs Unions and the Core, Journal of International

Economics 19 (1985), 355-365.

[21] Riezman, R., Can bilateral trade agreements help to induce free-trade?,

Canadian Journal of Economics 32 (1999), 751�766.

[22] Saggi, K., and H.M. Yildiz, Bilateral Trade Agreements and the Feasi-

bility of Multilateral Free Trade (2006), Mimeo.

36

[23] Summers, L., Regionalism and the World Trading System, In Policy

Implications of Trade and Currency Zones, Federal Reserve Bank of

Kansas City, 295-301 (1991).

[24] Viner, J., The customs union issue, Carnegie Endowment for Interna-

tional Peace, New York (1950).

37

Appendix

jip

Bi

Ai

Ai XXD ,−

]1[BiX

]2[BiX

]3[BiX

Aiτ

Biτ

Figure 1: The FTA

Ai

Ai XD −

jip

Bi

Ai

Ai XXD ,−

]1[BiX

]2[BiX

]3[BiX

Aiτ

Biτ

Figure 1: The FTA

Ai

Ai XD −

jipτ

Bi

Ai

Ai XXD ,−

]1[BiX

]3[BiX

Aipτ

Bipτ

Figure 2: The FTA Revisited

Ai

Ai XD −

jipτ

Bi

Ai

Ai XXD ,−

]1[BiX

]3[BiX

Aipτ

Bipτ

Figure 2: The FTA Revisited

Ai

Ai XD −

38

p Ci

Ci DX −

Mp

Fp

Figure 3: Effects on the World Market(Reduced Protection)

p

Ci

Ci

BAi

BAi DXXD −− ++ ,

BAi

BAi XD ++ −

FTA effect

Multilateraleffect

p Ci

Ci DX −

Mp

Fp

Figure 3: Effects on the World Market(Reduced Protection)

p

Ci

Ci

BAi

BAi DXXD −− ++ ,

BAi

BAi XD ++ −

FTA effect

Multilateraleffect

p Ci

Ci DX −

Mp

Fp

Figure 4: Effects on the World Market(Enhanced Protection)

p

Ci

Ci

BAi

BAi DXXD −− ++ ,

BAi

BAi XD ++ −

FTA effect

Multilateraleffect

p Ci

Ci DX −

Mp

Fp

Figure 4: Effects on the World Market(Enhanced Protection)

p

Ci

Ci

BAi

BAi DXXD −− ++ ,

BAi

BAi XD ++ −

FTA effect

Multilateraleffect

39