CHAPTER 6 - International Trade Agreements

CHAPTER

International Trade Agreements

∗

Giovanni Maggi

Ya l e U nive r s i t y, FG V / E P G E a n d N B E R

Abstract

In this chapter I survey the recent theoretical and empirical advances in the economics of inter-

national trade agreements, focusing on three main topics: the motives for trade agreements; the

design of rules and institutions for trade policy; and regional trade agreements.

Keywords

International trade agreements

JEL classiﬁcation codes

F13, F53, K33

1. INTRODUCTION

The starting point for this survey is represented by the two chapters on trade agree-

ments in the previous volume of the Handbook of Inter n ational Economics (1995), namely

Robert Staiger’s chapter “International Rules and Institutions for Trade Policy” and

Richard Baldwin and AnthonyVenables’ chapter “Regional Economic Integration.” For

the most part I will focus on the post-1995 advances in the literature on trade agreements;

I refer the reader to the previous volume of this handbook for the pre-1995 literature.

Before I plunge into the literature, however, it i s useful to start with a quick review

of the main developments that have occurred in the real world of international trade

agreements since 1995. My aim is not to embark in a comprehensive discussion of these

developments, b ut simply to pr ovide a stylized historical context for the literature that I

will survey.

The ﬁrst major development is that, after the completion of the Uruguay Round

in 1995, the General Agreement on Tar iffs and Trade (GATT) has been replaced by

the World Trade Organization (WTO). T he WTO i s a considerably more developed

international institution than its predecessor, with a broader set of functions that include

∗

I am grateful to Bob Staiger and Rachel McCulloch for being my discussants at the handbook conference and providing

very detailed and useful comments on an earlier draft,to co-editor Elhanan Helpman for providing an additional set of

helpful suggestions, and to Kyle Bagwell, Matthew Grant,Andrés Rodríguez-Clare and par ticipants in the handbook

conference for helpful comments and discussions.

317

318 Giovanni Maggi

not only the implementation of the commitments made by member countries in the

Urugua y Round, but also a relatively sophisticated judicial system, known as the “Dispute

Settlement Procedure.”

A second important fact has been the growing role of developing and newly industri-

alized countries in the WTO and in regional trade agreements. An important milestone

in this respect was the 2001 accession of China to theWTO. Another signiﬁcant manifes-

tation of this trend has been the growing invo lvement of newly industrialized countries

(especially Brazil, India and, after 2001, China) in the WTO dispute settlement system.

The third major development has been a tremendous acceleration in the formation

of regional trade agreements. For example, during the ﬁr st 10 years of the WTO (1995–

2005) the number of regional trade agreements in force more than tripled, from 58 to

188. Currently the total number of regional trade agreements in force is about 380, and

several potential new ones are currently being negotiated, including one between the US

and the EU, which would constitute the largest regional trade agreement in the wo rld.

The fourth and ﬁfth developments are perhaps better described as non-developments.

After the completion of the Uruguay Round, the WTO member countries embarked in

a new major round of multilateral negotiations, the so-called Doha Round. These nego-

tiations started in 2001 but have stalled, and at this point it is safe to say that the Doha

Round is effectively dead. Negotiations have stumbled mostly over divisions between

developing and developed countries, but there has also been considerable contention

between the EU and the US over agricultural subsidies. This lack of progress in multi-

lateral trade liberalization is probably related to both of the developments I mentioned

above—the proliferation of regional trade agreements and the growing role of developing

countries—but the deep reasons for the failure of the Doha Round are an open question.

The second non-development is—contrary to the one discussed just above—a positive

one: the existing rules and institutions have held up rather well, even in the face of

challenges such as the 2008 Great Recession,which led to a dramatic (though temporary)

collapse of trade ﬂows, and the accession of China to the WTO. In particular , the WTO’s

dispute settlement system has been quite effective. Indeed, many scholars have argued

that the enforcement and judicial aspects of the WTO are stronger now than during the

GATT years. Furthermore, if judged by the standards of international organizations, it

is safe to say that the WTO has established itself as one of the most, if not the most,

successf ul international organization in terms of enforcement and dispute settlement.

This concludes my stylized portrait of the recent developments in trade agreements,

and against this backdrop I now turn to my discussion of the recent advances in the

academic literature on trade agreements.

Section 2 focuses on t heoretical and empirical research investigating the purpose of

trade agreements. In the theoretical part, I focus mostly on three theories: the “terms of

trade” theory, according to which the purpose of a trade agreement is to prevent govern-

ments from manipulating terms of trade; the “domestic commitment” theory, according

to which a trade agreement can provide a government with a means to tie its own hands

International Trade Agreements 319

vis-à-vis domestic agents; and the “New Trade” theory, which emphasizes the role that a

trade agreement can play in the presence of imperfect competition. The empirical par t

of the discussion focuses on two sets of contributions: those aimed at testing the predic-

tions of the main theories, and those that study the impacts of trade agreements on trade

barriers and trade ﬂows in a more descriptive way.

Section 3 discusses research aimed at understanding the design of trade agreements,

with a particular emphasis on the role of transaction costs. In particular, I will focus

on two distinct types of transaction cost: contracting fr ictions and enforcement frictions.I

will argue that taking these transaction costs into account is essential for understanding

the design of substantive policy rules (such as tariff ceilings, non-discrimination rules,

etc.), enforcement rules (which regulate punishment/retaliation), and dispute settlement

procedures.

Section 4 discusses recent research on regional trade agreements. In particular, I will

focus on the economic and political determinants of regional trade agreements; on the

impacts of such ag reements on its members’ external trade barriers and on multilateral

trade liberalization; and on the design of rules for trade negotiations, that is, on whether

the formation of regional trade agreements should be subject to constraints or even

prohibited.

Finally, Section 5 offers some concluding remarks and some thoughts about possible

avenues for future research.

2. THE MOTIVES FOR TRADE AGREEMENTS

The most basic question regarding trade agreements (TAs) is wh y countries sign them in

the ﬁrst place. In spite of this being the most fundamental of questions, it is only in the last

15 years or so that the academic literature has made substantial progress in answering it.

Krugman (1997, pp. 113–120) made a famous pessimistic statement:“Anyone who has

tr ied to make sense of inter n ational trade negotiations eventually realizes that they can only be

understood by realizing that they are a game scored according to mercantilist rules. (...) The implicit

mercantilist theory that underlies trade negotiations does not make sense on any level, indeed is

inconsistent with simple adding-up constraints; but it nonetheless governs actual policy (...) the

economic theor y underlying trade negotiations is nonsense.”The last 15 years of research onTAs

are in some way an attempt to prove Krugman wrong, and argue instead that the logic

of economics (in a broad sense that includes also the logic of political economy) can to a

large extent make sense of real-world TAs.

In this section I will offer my critical survey of the main theories for why countries

sign TAs and of the small but growing empir ical literature on this subject.

Before I proceed, I need to make clear what is the organizing principle of this section.

The distinction between analyzing the motives for a TA (which is the subject of this

section) and analyzing the design of a TA (which is the subject of the next section) can

320 Giovanni Maggi

sometimes be blur red, since the two aspects are obviously inter-related. But I think it

is important to keep these two aspects conceptually distinct. In the present section, I

abstract from issues of institutional/contract design (which, as I will argue below, are

interesting only in the presence of transaction costs) by assuming that governments can

negotiate directly and costlessly over all policies (and, in the presence of uncertainty, over

all contingencies).

There are two broad stor ies for why governments sign TAs. The ﬁr st one is that a TA

can provide governments with an escape from an international Prisoners’Dilemma,which

is in turn caused by international externalities from trade policy: these include the classic

terms-of-trade externalities and the “New Trade” externalities that arise with imperfect

competition (such as “delocation” and “proﬁt-shifting” externalities). The second broad

story for why governments sign TAs is that these may provide governments with a com-

mitment device vis-à-vis domestic actors, such as industrial lobbies or individuals making

investment decisions. I will start by focusing on the classic terms-of-trade story.

2.1. The Terms-of-Trade Theory

The type of international exter nality from trade policy that has received by far the most

attention in the literature (dating back at least to Johnson, 1953)istheterms-of-trade

(TOT) externality that arises in a perfectly competitive environment. As it will become

clear in Section 2.3,TOT exter nalities play an important role also with imperfect com-

petition, but their role is clearest in the case of perfect competition, so I will focus on this

case here.

I will proceed in two steps. I will ﬁrst illustrate the TOT theory using a simple

“workhorse” model that is very str uctured and delivers simple formulas and a number

of speciﬁc predictions. T his model is convenient also because it can be used as a basis

to illustrate a simple version of the domestic-commitment theory, as I will show i n

Section 2.2. I will then present a more general version of the TOT theory, which has

been developed mostly by Bagwell and Staiger.

2.1.1. A Simple Workhorse Model

My workhorse model is essentially a simpliﬁed version of Grossman and Helpman’s

(1995a) “Trade Wars and Trade Talks” model, where governments may have two motives

for unilateral trade policy intervention, namely a terms-of-trade motive and a political-

economy motive.

Consider a world with two countries (H and F) and with three sectors, a numeraire

sector (0) and tw o non-numeraire sectors ( j = 1, 2). All citizens have the same utility

function, which takes the form U = c



j=1

), where each u

is increasing and

concave.The numeraire good is produced one-for-one from labor, so the wage is pinned

down to one, while good j is produced from labor and capital according to the constant-

returns production function y

= F

, l

International Trade Agreements 321

Labor is perfectly mobile across sectors. Capital for the moment is assumed to be

immobile across sectors (so that it can be effectively viewed as a speciﬁc factor); in

Section 2.2, where I consider the domestic-commitment theory, I will assume that capital

is immobile in the short run b ut can move across sectors in the long run.

The owners of capital represent a negligible fraction of the total population; this will

simplify the political-economy envir onment. The size of the population is equal to one

in each country. Country H is the natural importer of good 1 and country F of good 2.

Assume the numeraire good is freely traded. Each government can choose speciﬁc

trade taxes/subsidies in the non-numeraire sectors. Later I will consider domestic policies

as well.

Next I describe the political-economy environment. In each country, the owners of

capital in a given sector may or may not be organized into a lobb y. The government’s

objective is 

= aW + C,whereW is aggregate welfare and C denotes total contribu-

tions received from lobbies. If sector j is organized, the lobby’s objective is 

= 

−C

where 

denotes returns to capital in sector j. Analogous notation (but with asterisks)

will apply to the F country.

Given t he quasi-linear preferences, the speciﬁc-f actor structure, and the wage pinned

down at one, this is essentially a partial-equilibrium setting. Deﬁning welfare as aggregate

indirect utility, Home welfare can be written (using standard techniques) as W = Y +

S + R,whereY is total factor income,S is total consumer surplus (summed over the two

goods) and R is revenue from trade policy (positive or negative). An analogous expression

holds for Foreign welfare.

For each sector j, the demand functions in the two countries ar e denoted b y d

)

and d

∗

), the supply functions b y y

) and y

∗

), and the import demand functions

by m

) and m

∗

Let τ

(τ

∗

) denote the speciﬁc trade tax/subsidy chosen by Home (For eign) in sector

j. If good j is imported by Home, τ

is interpreted as an import tax, and τ

∗

as an export

subsidy, and vice-versa if the good is exported by Home. The price arbitrage condition

= p

∗

+ τ

− τ

∗

) and the market clearing condition (m

) + m

∗

) = 0) together

determine equilibrium prices as functions of trade policies: p

(τ

− τ

∗

) and p

∗

(τ

− τ

∗

Finally, deﬁne the “world price” of good j as p

= p

− τ

= p

∗

− τ

∗

I start by focusing on the noncooperative scenario. In each country, assume that

government and lobbies choose trade policy (and contributions) by Nash bargaining,

taking foreign trade policy as given. Given that bargaining is efﬁcient, trade policy in

each country maximizes the joint surplus of government and lobbies given foreign trade

policy. Thus Home trade policy maximizes  ≡ 



j=1



= aW +



j=1



I note here that very little research has been done on the role of informational lobbying (as opposed to quid pro quo

lobbying) in inﬂuencing trade negotiations. The only paper of this kind that I am aware of is Milner and Rosendorff

(1996). Other papers on this general theme are Tovar (2011) and Ludema et al. (2010), but these papers focus only on

unilateral trade policy choices, not trade agreements.

322 Giovanni Maggi

where I

= 1 if sector j is organized and zero otherwise. Similarly, Foreign trade policy

maximizes 

∗

≡ a

∗



j=1

∗



∗

As mentioned abo ve,this model can be viewed as a simpliﬁed version of Grossman and

Helpman (1995a).

Note also that this model is equivalent to one where each government

maximizes a politically-adjusted welfare function that attaches an extra weight to t he

organized sectors’ rents, as for example in Baldwin (1987).

Maximizing  with respect to τ

and 

∗

with respect to τ

∗

yields the following

formulas for τ

and τ

∗

a · η

, (1)

∗

· η

∗

, (2)

where η

∗

≡

∗

∗

and η

≡



. The ﬁrst term in each formula is the well-known

Johnson’s optimum trade tax, which captures the terms of trade motive for trade policy; the

second part of each formula captures the political motive for trade policy. Note that the

two motives go in the same direction for organized import-competing industries (both

call for a tariff), but are in conﬂict for organized export sectors (TOT considerations call

for an export tax, political considerations call for an export subsidy).

We are now ready to examine the trade policies that are selected if the two countries

negotiate aTA. Assuming that the TA maximizes the joint surplus of all governments and

lobbies, that is  + 

∗

we obtain:

− τ

∗

a · η

−

∗

· η

∗

. (3)

Note that for each sector j the optimal agreement pins down only the net trade tax

− τ

∗

, not the exact levels of τ

and τ

∗

. To understand intuitively why, consider the

case in which governments maximize welfare: a = a

∗

=∞. Then the formula above

yields τ

= τ

∗

. A tariff of $1 and an export subsidy of $1 on the same good neutralize

One simpliﬁcation relative to Grossman and Helpman (1995a) is that the share of the population represented by lobbies

(the α

parameter) is negligible. Another simpliﬁcation is that the interaction between government and lobbies is

modeled as a Nash bargaining game, whereas in Grossman and Helpman it is modeled as a common-agency game, but

in both cases a country’s trade taxes maximize the joint sur plus of government and lobbies.

This will be the case, for example, if trade negotiations take the for m of a Nash bargain that involves the governments

and all lobbies and international transfers are available. I note that Grossman and Helpman (1995a) assume a two-stage

game where lobbies ﬁrst offer contribution schedules to their respective governments, and then gover nments bargain

in Rubinstein fashion, but the policy outcome in their model is the same as in the simpliﬁed model I consider here.

Also note that, even if intern ational transfers are not available, such transfers can be indirectly effected by adjusting

import and export tax levels, as I explain below.

International Trade Agreements 323

each other’s effect on domestic prices, and the only effect that remains is a revenue trans-

fer from the exporting country to the importing country; the condition τ

= τ

∗

is the

partial-equilibrium analog of the well-known “Mayer curve.”

If political motivations are present in the governments’ objectives (a and a

∗

are less

than inﬁnity), the efﬁcient policies will reﬂect only these motivations, and not the TOT

motivations: an efﬁcient agreement simply removesTOT considerations from trade policies.This in

turn suggests that the only source of inefﬁciency in the noncooperative equilibrium is the

gov ernments’ temptation to manipulate TOT. Or put another way, even if governments

are politically motivated, the reason for signing aTA is inherently economic, not political.

As I discuss next, this basic message is further developed and generalized by Bagwell and

Staiger.

2.1.2. The Bagwell-Staiger Approach

In an inﬂuential series of papers, Bagwell and Staiger hav e argued that the TOT theory—

and in particular the conclusion that the only motive for a TA can be traced to the

governments’ temptation to manipulate TOT—is considerably more general than previ-

ously thought. I refer the reader to Bagwell and Staiger’s (2002) book and Bagwell and

Staiger’s (2010a) sur vey for more detailed expositions of their work. In this section I will

offer a succinct exposition of Bagwell and Staiger’s theory under perfect competition; I

will discuss their models with imperfect competition in the next section.

Consider a two-country, two-good, perfectly competitive world, with the good

exported by Home taken as the numeraire.The production possibility frontier is assumed

to be concave ,and the goods are assumed to be normal in consumption.The model allows

for domestic distortions, such as consumption or production externalities, that call for

corrective domestic policies (but monopoly distortions are not allow ed).

Each govern-

ment has access to a complete set of (trade and domestic) policy instruments. It is useful

to partition policies in two categories: (i) Tax instruments that create wedges between

local prices and the world price, or more speciﬁcally, trade taxes, production taxes, and

consumption taxes. I will refer to these as“wedge policies”

;(ii) Other policies (e.g. labor

standards) that may affect market conditions but cannot affect price wedges. I will refer

to these as “non-wedge policies.”

Key to Bagwell and Staiger’s approach is the way government objectives are repre-

sented. These are represented in reduced form as (ν,p, q, p

) and 

∗

(ν

∗

, q

∗

, p

where p ( p

∗

) is the Home (Foreign) consumer price, q (q

∗

) the Home (Foreign) producer

See Mayer (1981), who characterizes the locus of efﬁcient tar iff combinations in a general equilibrium model with

two goods and two countries.

The framework I present here can be seen as combining three variants of the Bagwell-Staiger model: Bagwell and

Staiger (1999a),which considers only trade taxes; Bagwell and Staiger (2001),which considers trade taxes and domestic

standards; and Bagwell and Staiger (2006), which considers trade taxes and production subsidies.

Of course there is a degree of redundancy in these three taxation instruments, since a trade tax is equivalent to a

combination of a production subsidy and a consumption tax.

324 Giovanni Maggi

price, ν (ν

∗

) the v ector of Home (For eign) non-wedge policies, and p

the world pr ice.

Since wedge policies can be wr i tten as price wedges, t hey need not be included as argu-

ments in the objective functions. Of course, the equilibrium price levels in general will

depend on all policies.

A central feature of this setting is that a government’s policies affect the other govern-

ment’s payoff through a unique channel: the world price. In other words, the only inter-

national externality is the TOT externality. An implicit assumption in this setting is that

there are no non-pecuniary international externalities (such as cross-border pollution).

But aside from this restriction, the above representation of governments’objectives is gen-

eral enough to capture the presence of political-economy motives for protection; indeed,

Bagwell and Staiger argue that many of the existing political-economy models (including

the Grossman-Helpman model presented above) can be represented in this fashion.

The only structure Bagwell and Staiger impose on government preferences is that, for

given domestic prices,a government dislikes a worsening of terms of trade:

∂

∂p

< 0 <

∂

∗

∂p

This seems like a r easonable restriction, since domestic interest groups care about domestic

prices, not directly about world prices. If p

increases while domestic prices are kept

constant,there is simply a transfer of revenue from Home to Foreign (given the assumption

of normal goods), so this amounts to assuming that, all else equal, a government values

rev enue. Finally, the Lerner paradox and the Metzler paradox are assumed away.

As usual, the analysis starts with the noncooperative scenario, that is the Nash equilib-

rium of the game where governments simultaneously choose policies. Three basic points

canbemadeinthissetting.

The ﬁrst point is that any Nash equilibr ium is Pareto-inefﬁcient from t he point of

view of the governments’ objectives. This result is intuitive, since a country’s policies exert

externalities on the other country through TOT, and hence unilateral policy choices will

generically not be efﬁcient.

The second point is that the gov ernments’ temptation to manipulate T OT is the only

source of inefﬁciency in the Nash equilibrium, and hence it is the only motive for a TA.

Bagwell and Staiger establish this point by considering a diagnostic test to ascertain if

T OT manipulation is the only cause of the “disease” in the noncooperative equilibrium.

The test is based on the following question: if each government did not value the pure

terms-of-trade effects of its policies, would governments make efﬁcient choices? If the

test is positive, the diagnosis is that TOT manipulation is the only cause of the disease.

More speciﬁcally, Bagwell and Staiger deﬁne the politically optimal (PO) policies as

those that w ould result if governments did not v alue changes in p

. If the PO policies

are efﬁcient, then TOT manipulation is deemed to be the only cause of inefﬁciency in

the noncooperative equilibrium. In the setting under consideration, Bagwell and Staiger

The Lerner paradox occurs when an increase in a country’s tariff leads to an increase in the world relative price of the

imported good; the Metzler paradox occurs when an increase in a country’s tariff leads to a decrease in the domestic

relative price of the imported good.

International Trade Agreements 325

show that the PO policies are indeed efﬁcient. A simple intuition for this result can be

gained by considering a setting where utility is transferrable, with changes in p

acting as

pure transfers, so that efﬁcient policies must maximize the global payoff  + 

∗

.Inthis

case, at a political opt imum, Home’s policies maximize  given p

and Foreign policies

maximize 

∗

given p

, therefore the global payoff must be maximum because p

is a

pure transfer.

The third point is that trade volume at the Nash equilibrium is inefﬁciently low, and

a mutually beneﬁcial TA must entail a reciprocal expansion of market access relative to

the Nash equilibrium.

I will provide a simple local intuition for this result, abstracting

from domestic policies for simplicity. Starting from the Nash equilibrium,a small increase

in a trade tax has a negative externality on the trading partner. This is not self-evident,

since an increase in a country’s trade tax could in principle ha ve a positive effect on the

trading partner thr ough the latter’s local price. But this cannot be the case locally at the

Nash equilibrium: the optimality of a country’s unilateral policies implies that, at the

Nash equilibrium, any effect through the local price cannot outweigh the adverse effect

through the world price, so the externality from an increase in the foreign trade tax is

negative.

Given the negative international externalities from trade taxes, it is intuitive

that, starting from the Nash equilibrium, the only way to achieve a Pareto improvement

is to decrease both trade taxes, which in turn will expand trade.

In light of the three points highlighted above, Bagwell and Staiger conclude that

under the assumptions of their model the only purpose of a TA is to preclude countries

from manipulating TOT, and this in turn entails a reciprocal expansion of market access

relative to the noncooperative equilibrium. This prediction of the model resonates with

the emphasis placed by the GATT-WTO on the exchange of market access between

countries.

Next I highlight a prediction that presents a special challenge for the TOT theory,

because it is at odds with observed TAs. According to the TOT theory, a TA should tend

to increase export subsidies relative to the noncooperative equilibrium, whereas in reality

export subsidies are typically restricted byTAs.This counterfactual prediction of theTOT

Bagwell and Staiger make a distinction between an expansion of “market access” and an expansion of trade volume

(see Bagwell and Staiger, 2001, pp. 537–538).The for mer is a weaker condition than the latter: a policy change is said

to expand market access to country A if it shifts out country A’s import demand curve for at least some world price. As

Bagwell and Staiger show, a mutually beneﬁcial trade agreement must entail a reciprocal expansion of market access,

but in general it need not entail an expansion of import volumes; it will entail an expansion of import volumes under

the additional assumption that any policy change shifts a country’s impor t demand cur ve in the same direction for all

world-price levels. In my intuitive discussion in the text I abstract from domestic policies, so this distinction is not

necessary, and a mutually beneﬁcial agreement always entails an expansion of trade volumes.

To see this formally, note ﬁrst that if only trade taxes are available, we can write Home’s payoff as (p, p

).Letting

τ and τ

∗

denote trade taxes, we can write the externality of τ

∗

on Home (with a slight abuse of notation) as



∗

= 

∗

+ 

∗

= (

+ 

∗

(where I used p

∗

= p

∗

). At a Nash equilibrium, the ﬁrst order

condition (FOC) is 

= 

+ 

= 0, which using p

= p

+ 1yields

=−



, which is negative

by the no-Lerner-paradox and no-Metzler-paradox assumptions. This in turn implies 

∗

< 0.

326 Giovanni Maggi

model can be illustrated by focusing on a model with two goods.The well-known Lerner

symmetry theorem states that an import tax is equivalent to an export tax, thus w e can

suppose without loss of generality that each government uses only an export tax (or if

negative,an export subsidy). In this case,if the noncooperative equilibrium entails export

subsidies (which is possible if export interests are politically strong), it is easy to show

that a mutually beneﬁcial TA must increase their le vels. Intuitively, increasing a country’s

export subsidy has a positive TOT externality on the other country, so governments

“under-subsidize” exports in equilibrium. I will refer to this feature as the export subsidy

puzzle in the TOT theory. As I discuss below, possible ways to resolve this puzzle include

considering domestic-commitment motives and “New Trade” motives for TAs.

Thus far I have focused on a two-country world. Extending the analysis to a multi-

country world introduces new considerations. As Bagwell and Staiger (1999a) make clear,

when trade policies can discriminate across trading partners, there is no longer a single

world price but a whole v ector of bilateral world prices, and importantly, international

externalities can no longer be viewed as traveling solely through world prices. As a

consequence, if trade policies can be discriminatory, the PO policies are inefﬁcient. To

understand this point, focus on the impact of foreign trade policies on the Home country.

Deﬁne the multilateral TOT as an import-weighted average of bilateral world prices.

Since the import weights depend on foreign local pr ices, now international externalities

travel not only through world prices but also through foreign local prices. It is then

intuitive that the PO policies are not efﬁcient. On the other hand, if governments ar e

constrained by a Most Favored Nation (MFN) rule to choose non-discriminatory trade

policies, then Bagwell and Staiger show that all international externalities are channeled

through a single world price, and again the PO policies are efﬁcient. To summarize, the

PO policies are efﬁcient if and only if trade policies are constrained by the MFN rule.

The result I just highlighted can be interpreted in more than one w ay. Bagwell and

Staiger argue that the result conﬁrms the general point that in a perfectly competitive

environment the only purpose of a TA is to prevent the manipulation of TOT. But

one could argue that the appropriate thought experiment should diagnose the cause of

the disease in a scenario where no institutional constraints are in place, not e ven the

MFN rule , in which case PO policies are inefﬁcient and one should conclude that TOT

manipulation is not the only motivation for aTA. Thus there is a legitimate question as to

which of the two diagnostic tests (with unconstrained policies or with MFN-constrained

policies) is more informative about the deep motivation for a TA.

Bagwell and Staiger build on the model outlined above to argue that it can explain

some key rules of the GATT-WTO, such as reciprocity, MFN, and t he so-called “non

violation”rule. I will come back to these themes in the next section,where I focus on the

design of TAs, but here I wish to re-iterate a point alr eady mentioned abo ve: in a world

without transaction costs, the theory would not be able to explain any such rules, because

then go vernments could simply negotiate directly on the policy levels,and there would be

International Trade Agreements 327

no need for additional rules, so this second part of Bagwell and Staiger’s theory implicitly

relies on the presence of some kind of transaction costs. As already mentioned, in the

present section I am assuming that there are no transaction costs, and hence governments

can negotiate directly and costlessly over all policies, so I post pone issues of rules design

to the next section.

2.2. The Domestic-Commitment Theory

The TOT theory is by far the one with the deepest roots in the literature, but it is not

clear that TOT considerations are the whole story behind TAs, for at least two reasons.

First, casual empiricism suggests that small countries (which have negligible inﬂuence

on world prices) often agree to signiﬁcant cuts in their trade barriers when they join

a TA, an observation that is not easy to reconcile with the TOT theory.

And second,

as I mentioned above, the TOT theory implies that TAs should tend to increase export

subsidies relative to their noncooperative levels, which is a counterfactual prediction. An

alternative theory that can explain these observations is based on the idea that a TA can

help a government tie its own hands vis-à-vis domestic actors.

There are several models in the literature that fall within the broadly deﬁned domestic-

commitment theory. Some are of a purely economic nature, for example Staiger and

Tabellini (1987),Tornell (1991),and Lapan (1988), and some are of a political-economy

nature, in par ticular Maggi and Rodriguez-Clare (1998, 2007), Mitra (2002), Brou and

Ruta (2009), Limão andTovar (2011),andLiu and Or nelas (2012). Since the former type

of domestic-commitment models was covered by Staiger’s (1995a) chapter, I will focus

on the latter type, and in particular on the version due to Maggi and Rodriguez-Clare

(1998, 2007).

The general idea proposed by Maggi and Rodriguez-Clare is that a TA can serve as

a commitment dev ice for a government t o close the door to domestic lobbies. It has

been argued by a number of scholars and commentators that this type of motivation

was central to Mexico’s negotiations of the North American Free Trade Agreement

(NAFTA). For example, Whalley (1998) argued that Mexican negotiators of NAFTA

“were less concerned to secure an exchange of concessions between them and their

negotiating partners, and were more concerned to make unilateral concessions to larger

negotiating par tners with whom they had little negotiating leverage...The idea was clearly

The reason I use the expression “casual empiricism” is the following. There is little doubt that at least in some cases

countries with negligible monopsony power on given goods have agreed to signiﬁcant tariff cuts on those goods, but

I am not aware of any empirical study that investigates whether this is the case more systematically.

Interestingly, in the same 1997 essay where Krugman declared it impossible to understand trade negotiations from a

rational perspective, he left a small opening for the domestic-commitment theory of trade ag reements, although still

with some degree of skepticism. He summarizes this theory as maintaining that “the true purpose of international

negotiations is arguably not to protect us from unf air foreign competition,but to protect us from ourselves,” then states

that “one cannot dismiss such political-economy arguments as foolish,” but questions whether in reality international

agreements are truly effective in achieving this purpose.

328 Giovanni Maggi

to help lock in domestic policy reform.” Similarly, Bajona and Chu (2010) view China’s

accession to WTO as a way to “ ... lock-in the agenda for fundamental domestic reforms,

which has been difﬁcult to implement by domestic measures alone.”

Notice however that, if one considers the typical models of lobbying that ha ve been

proposed in the literature, in particular those in the tradition of Grossman and Helpman’s

(1994) “Protection for Sale,” it is not clear why a government would ever want to tie its

own hands, since it derives positive rents from the political process.

Magg i and Rodriguez-Clare (1998) provide a theoretical justiﬁcation for the domestic-

commitment argument based on a simple dynamic model.The idea is that a government

can deri ve rents from the interaction with lobbies in the short run, but in the long run

this will distort the allocation of resources, because investor s will overinvest in the sectors

that are expected to get trade protection, and the government is not compensated for this

long-run distortion. As a consequence, the government may be better off committing to

free trade ex-ante, thereby shutting down the lobbying process.

The basic points can be illustrated within the workhorse model of Section 2.1. Con-

sider the same economic and political structure as in that model, but now suppose that

H is a small country, while F is a large “rest of the world.”Also assume for simplicity that

both sectors 1 and 2 are politically organized.

Consider the following timing: (0) the small-country gov ernment chooses whether to

commit to f ree trade; (1) capital i s allocated;and (2) given the capital allocation,trade pol-

icy and cont ributions are determined by Nash bargaining between the government and

the lobbies (with σ denoting the government’s bargaining power). This timing captures

the idea that capital is mobile in the long run but not in the short run.

Suppose ﬁrst that the government does not commit to free trade. Let us proceed

b y backward induction and ﬁnd the second-stage equilibrium payoffs gi ven the capital

allocation. For the Home country, let K denote the vector of capital allocations and τ

the vector of trade policies. Also, let W (τ, K) and (τ, K ) denote respectively the levels

of general welfare and the aggregate returns to capital in sectors 1 and 2 as functions

of trade policies and capital allocations. Given that the government and the lobbies

engage in Nash bargaining over policies and contributions, the ﬁrst step is to derive the

status-quo (disagreement) pa yoffs. In the status quo, lobbies give no contributions and

the government chooses the welfare-maximizing policy, which is free trade, hence the

government’s status-quo payoff is aW (0, K ), and the lobbies’ total status-quo pa yoff is

(0, K). The next step is to write down the joint surplus of the government and the

lobbies:

J(K ) = max

[aW (τ, K) + (τ, K )]−[aW (0, K) + (0, K )].

I note that, while Maggi and Rodr iguez-Clare focus on a setting where a government can be pressured only by its

domestic lobbies, similar beneﬁts from committing to free trade may arise if a government can be inﬂuenced also by

foreign lobbies. For a paper that documents the empirical importance of foreign lobbying, see Gawande et al. (2006).

International Trade Agreements 329

The government walks away with a share σ of this joint surplus, therefore its payoff in

thesecondstageisgivenbyaW (0, K) + σ J(K ).

The next step is to derive the equilibrium allocation in the ﬁrst stage, which I denote

K . The key point is that, if σ < 1, this will generically be different from the free trade

allocation (

K = K

), and hence inefﬁcient, while

K = K

if σ = 1. This is intuitive,

because as long as lobbies have any bargaining power (σ < 1), the presence of lobbying

distorts the net returns to capital relative to free trade. If, on the other hand, lobbies have

no bargaining power (σ = 1), they will walk away from the bargain with no surplus, and

hence the lobbying process does not affect the returns to capital net of contributions,so

the equilibrium allocation is efﬁcient. With this in mind, we can write the government’s

equilibrium payoff in the no-commitment scenario as G

= aW (0,

K ) + σ J (

K).

Now suppose the government commits to free trade. In this case, expecting free

trade, capital owners will make efﬁcient allocation decisions: K = K

, and hence the

government’s payoff in this case is G

COMM

= aW (0, K

The government will commit to free trade if and only if G

COMM

.Now

observe that: (i) if σ = 0, then G

COMM

, because W (0, K

) >W(0,

K);

and (ii) if σ = 1, as I noted above we have

K = K

, and since J (K

) > 0then

COMM

. We can then conclude that if σ is sufﬁciently low the government

will commit to free trade, and if σ is sufﬁciently high it will not. Moreover, under some

conditions G

will be increasing in σ , in which case there will be a critical level of σ

below which the government commits to free trade and above which it does not. Thus

the model yields an interesting prediction: countries whose governments have a weak er

bargaining position vis-à-vis domestic lobbies should be more likely to join a TA.

Another prediction generated by the model concerns the impact of the parameter a,

the government’s valuation of welfare relative to contributions. Provided σ is sufﬁciently

small, the value of commitment (V = G

COMM

− G

) is non-monotonic in a:itstarts

negative, then it turns positive, and eventually it approaches zero as a →∞.

This in

turn implies that, if there is a small cost of joining the agreement, the government will

choose to join if a f alls in some intermediate range.

Importantly,note that if export interests are organized the noncooperative equilibrium

will entail export subsidies, so in Maggi and Rodriguez-Clare (1998) the government

may want to commit to the elimination of export subsidies. Thus the model suggests

a possible solution to the “export subsidy puzzle” highlighted abo ve in the context of

the TOT theor y : if TAs are motivated by domestic-commitment issues, they will reduce

export subsidies r elative to their noncooperative levels.

To see this, note that (i) if a = 0, the government does not care about welfare, so clearly V<0; (ii) if a =∞,the

gover nment only cares about welf are, so tariffs are zero in the political equilibrium, hence V = 0; (iii) that V must be

positive for a range of a if σ is sufﬁciently small follows from the observation made above that, for ﬁxed a>0, if σ is

sufﬁciently small then V>0.

330 Giovanni Maggi

Next I make a point that will be useful to k eep in mind when I focus on the

implications of incomplete contracting for TAs (Section 3). Recall that in Maggi and

Rodr iguez-Clare (1998) the inefﬁciency in the noncooperative equilibrium stems from

the government’s lack of c ommitment vis-à-vis domestic agents, and the core of the

problem is that the government does not get compensated for the long-run distortions

fr om trade protection. But note that the same problem can be viewed also as a problem

of incomplete contracting between the government and domestic agents: if the government

could sign a long-term contract with all the future beneﬁciaries of protection, in which

it commits to future trade policies and gets compensated for them, t he problem would

disappear. Of course,if capital is mobile in the long run, this long-term contracting w ould

have to inv olve all capital owners in the economy, not only those that are currently in the

organized sectors, thus it seems reasonable to assume that such long-term contracting is

not feasible.

Magg i and Rodriguez-Clare (2007) extends the previous model in four directions.

First, it allows for tw o large countries; thus the model nests two motives for a TA: a

domestic-commitment motive and a TOT motive. Second, governments can commit

to arbitrary tariff levels (as opposed to free trade or nothing); mor eover, they can do

so through exact tariff commitments (a complete contract) or through tariff caps (an

incomplete contract). Third, speciﬁc-factor owners can lobby ex-ante to inﬂuence the

shape of the agreement, not only ex-post. And four th, the model allows for different

deg rees of capital mobility across sectors.

The model considers the following dynamic scenario. The world is sitting at the non-

cooperative equilibrium—with its associated allocation distortions—when the opportu-

nity to negotiate an agreement ar rives.

The agreement maximizes the joint surplus of

governments and lobbies. After the agreement is signed, each investor gets a chance to

move his or her capital with an (exogenous) probability z. The parameter z thus captures

the degree of mobility of capital. After the reallocation of capital has taken place, tariffs

are chosen in each country by the government and the lobb y subject to the constraints

set by the agreement. Of course, this ex-post lobbying process is relevant only if the

agreement leaves some discretion, that is, if the TA takes the form of tariff ceilings.

The key results of the model are four. First, the extent of trade liberalization (the tariff

cuts enacted by the TA) is increasing in the degree of capital mobility (z). Intuitively, if

z is higher, current lobby member s care less about future protection, and hence they are

less resistant to tariff cuts. This in turn suggests a further pr ediction, beyond those high-

lighted above in the context of the small-country model: tariff cuts should be deeper in

sectors where capital is more mobile.This prediction seems consistent with the anecdotal

In the basic version of the model the opportunity to sign a trade agreement is a surprise to investors, but Maggi and

Rodriguez-Clare (2007) also consider a version of the model in which the trade agreement is perfectly anticipated b y

investors.

International Trade Agreements 331

observation that in reality trade liberalization has been hard to come by in the agricultural

sector , but it would be interesting to test this prediction in a mor e systematic way.

The second result concerns the impact of “politics”—captured inversely by the gov-

ernments’ valuation of welfare (a)—on the extent of trade liberalization: tariff cuts are

deeper when politics are more important, provided the domestic-commitment motive

is strong enough (z sufﬁciently high). This result stands in interesting contrast with the

prediction of the pureTOT model, where tariff cuts if anything tend to be less deep when

a is lower: the reason is that a lower a implies higher noncooperative tariffs, hence a lower

trade volume and a weaker TOT externality, and this calls for smaller tariff cuts. Also in

Magg i and Rodriguez-Clare (2007),alower a implies higher noncooperative tariffs, but

this in turn implies a bigger allocation distortion, and hence bigger tariff cuts are called

for. If z is high, this consideration dominates the previous one.

At a more fundamental level,the divergence in results highlighted above is a manifesta-

tion of a key difference between the domestic-commitment theory and theTOT theory.

In the domestic-commitment theory, the motive for a TA is inherently political, since

the TA is directly aimed at blunting domestic lobbying pressures, thus the TA is directly

affected by political parameters such as the governments’ valuation of welfare; whereas in

the TOT theory, the motive for a TA is inherently economic, and hence political forces

affect a TA only indirectly through economic variables (e.g. outputs and trade volumes).

The third insight is that the presence of a domestic-commitment motive can explain

why trade liberalization typically occurs in a gradual manner. In particular, the reduction

in tariffs happens in two phases: ﬁrst, there is an instantaneous drop in tariffs, which

reﬂects the T OT motive for the TA, and subsequently there is a gradual tariff reduction,

which reﬂects the domestic-commitment motive. Intuitively, the allocation distortions

caused b y protection are more severe in the long run than in the short run, and hence

the domestic-commitment motive calls for bigger tariff reductions in the long run than

in the short run. Furthermore, the speed of liberalization is increasing in z . The reason

is that, if z is lower, the expected length of time for which capital owners are “stuck” in

a sector is longer, so the lobby will insist on keeping a high protection level for a longer

period of time.

Finally, Maggi and Rodriguez-Clare (2007) show that tariff ceilings are prefer red to

exact tariff commitments.The intuition is in two steps. Fir st, if one focuses on complete

TAs, the optimal exact tariff commitments in general are positive, though low er than the

noncooperative levels,and hence induce allocation distortions. Second,consider replacing

an optimal exact tariff commitment with a tariff ceiling at the same height: the former

shuts down ex-post lobbying and contributions, while the latter leaves some discretion

(governments have the option of setting tariffs below the ceilings) and hence induces

ex-post lobb ying and contributions; the latter is preferable because the anticipation of

ex-post contributions reduces the expected net returns to capital in organized sectors, and

hence mitigates the investment distortion. I will come back to the topic of tariff ceilings

332 Giovanni Maggi

and the incompleteness of TAs in Section 3, where I focus on the design of TAs, but here

I note that Maggi and Rodriguez-Clare’s model can explain why TAs are incomplete

contracts without relying on the presence of contracting frictions between gov ernments

(although, as I highlighted abov e, contracting frictions between a government and its

domestic actors are key).

Next I brieﬂy discuss other papers that have highlighted domestic-commitment

motives for TAs in the presence of lobbying. Mitra (2002) shows that a similar domestic-

commitment motive as i n Maggi and Rodriguez-Clare (1998) arises also in a setting

where there is no long-run distortion in the capital allocation, but there is a resource

cost of lobb y formation: in this case, if the government does not commit to free trade,

the long-run inefﬁciency generated by the prospect of trade protection (that the gov-

ernment does not get compensated for) is given by the cost of lobb y formation. More

broadly,Mitra’s paper suggests that there may be a domestic-commitment motive for aTA

any time the prospect of trade protection leads to a long-run misallocation of resources,

whether it is in the form of misallocation of resources between pr oductive activities or

waste of resources in unproductive activities.

Brou and Ruta (2009) extend Maggi

and Rodriguez-Clare’s (1998) model by allowing governments to use trade policies and

domestic subsidies, and argue that the domestic-commitment theory of TAs can provide

a rationale for the WTO’s r estrictions on the use of production subsidies.

Limão and Tovar (2011) propose a different version of the domestic-commitment

argument forTAs. They consider a setting in which a small-country government bargains

with a domestic lobby over two policy instruments, a tariff and a non-tariff barrier, where

the latter is the less efﬁcient redistribut ive instrument. In this setting they show that the

government may beneﬁt from committing to a tariff reduction because this may improve

its bargaining position, and this beneﬁt may outweigh the cost of constraining the more

efﬁcient redistributi ve tool. Finally, Liu and Ornelas (2012) argue that a TA can serve

as a commitment device for the purpose of stabilizing a democratic regime. The key

idea of this paper is that an incumbent go vernment may value a TA because it leads to

the destruction of rents, which in turn reduces the likelihood of a coup by rent-seeking

autocratic groups, thereby helping consolidate unstable democracies.

I conclude this section by mentioning another model where a government’s lack

of commitment vis-à-vis domestic agents has important implications for TAs. McLaren

(1997) considers a two-period Ricardian model where a small country (S) negotiates a

TA with a large country (L). In the ﬁrst period, domestic agents commit their resources

Krishna and Mitra (2005) explore an interesting consequence of Mitra’s argument: if a country liberalizes unilaterally

because of a commitment issue as in Mitra (2002), this will have effects on the trade policies chosen by its trading

partners. They show that it may induce a trading partner to reduce its own tar iffs, because it increases the incentives

for the export lobby in the partner country to form and lobby against the import-competing lobby there for lower

protection.

This paper focuses on the rationale for joining free trade agreements, but the basic argument applies equally well to

the case of a multilateral TA.

International Trade Agreements 333

to a sector; in the second period, the governments negotiate over a tariff and a transfer

through Nash bargaining. Given the resource allocation, the equilibriumTA involves free

trade and a transfer from S to L. Ex-ante, anticipating free trade, agents commit resources

to the sector where S has a comparative adv antage. But this leads L to choose a higher

tariff in the Nash equilibrium, which in turn worsens the outside option of country S in

the trade negotiation. McLaren shows that this adverse effect of the anticipation of a TA

on the welfare of the small country may outweigh the standard gains from trade, so this

country may be better off by committing ex-ante not to sign a TA.

McLaren’s point relates in an interesting way to the domestic-commitment theory of

TAs. In McLaren’s model, the TA can be interpreted as a short-term contract, because it

occurs after investment decisions are made. But if theTA were a long-term contract,in the

sense of occurring before investment decisions are made, then the hold-up problem high-

lighted by McLaren would not arise.Thus McLaren’s model suggests thatTAs can help only

if they are effective long-run commitments (consistently with the domestic-commitment

theory), while they can ha ve perv erse effects if they are only short-term commitments.

2.3. New Trade Theories of Trade Agreements

A new and important line of research has emerged recently that explores the implica-

tions of imperfect competition for TAs. A central theme in this new area of research

is that, in the presence of imperfect competition, TOT externalities are not the only

international externalities from trade policy. In par ticular, three new externalities have

been identiﬁed and examined: (i) “ﬁrm-delocation” externalities in the presence of free

entry ( Venables, 1985, 1987; Ossa, 2011; Bagwell and Staiger, 2009, 2012b), (ii) “proﬁt-

shifting” externalities (Mrazova, 2011; Ossa, 2012; Bagwell and Staiger, 2012a), and (iii)

trade-volume externalities when prices are determined by bilateral bargaining (Antras

and Staiger, 2012a,b). As I discuss below, these non-TOT externalities may be a separate

cause of inefﬁciency in noncooperative policies, hence giving rise to new rationales for

TAs, and can have important implications for the design of TAs.

2.3.1. Firm-Delocation and Proﬁt-Shifting Externalities

In this section I focus on the implications of ﬁrm-delocation and proﬁt-shifting exter-

nalities from trade policy, starting with the former type.

Venables (1985,1987) was the ﬁrst to identify the possibility of ﬁrm-delocation exter-

nalities from trade policies. This type of externality can arise whenever markets are

imperfectly competitive and there is free entry. T he basic idea is the following: if a

In a recent paper,Sovey (2012) develops a model whereTAs are motivated by“political hold-up”problems. In her model,

if a gover nment makes a public investment in its comparative-advantage sector and hence makes itself more“dependent”

on trade , it gives its trading partner an increased ability to extract political concessions in the future. As in McLaren

(1997), a political hold-up problem calls for a long-ter m TA. Sovey then argues that a long-term TA is harder to self-

enforce than a short-term one,because of the additional political uncertainty over the longer time horizon,and for this

reason a multilateral institution like theWTO, by increasing the sever ity of punishments,can facilitate self-enforcement.

334 Giovanni Maggi

country imposes a tariff on imports, this will tilt the balance of competition in favor of

domestic ﬁrms, and this in turn will induce exit of foreign ﬁr ms and entry of domestic

ﬁrms. In the presence of transport costs, this effect tends to beneﬁt the country imposing

the tariff and hurt the exporting country.

Ossa (2011) has explored the implications of ﬁrm-delocation externalities for the

purpose and design of TAs. In particular, Ossa considers a Krugman-type model with

monopolistic competition, CES preferences ov er varieties and iceberg transport costs.

Governments maximize welfare and can only choose ad-valorem impor t tariffs. In this

model, import tariffs have no TOT effects at all. Intuitively, ﬁrms apply a constant mark-

up over marginal cost,so the incidence of an ad-valorem tariff falls entirely on the importing

country, and hence ex-factory prices are unaffected.

The feature that import tariffs have

no TOT effects of course depends on the special model structure, but it serves to isolate

the delocation externality, which operates in the following way: an increase in the Home

tariff leaves the total number of domestic and foreign ﬁrms unchanged, but modiﬁes its

composition in fav or of domestic ﬁrms; because of transport costs, this lowers the Home

price index and i ncreases the Foreign price index, thus leading to an increase in Home

welf are and a decrease in Foreign welfare.

As a consequence of the negative delocation externality that a tariff exerts on the

exporting country, the noncooperative equilibrium entails inefﬁciently high tariffs, and

so there is scope for a TA to reduce tariff levels. Ossa argues that this rationale for TAs

resonates with the often-heard informal argument that import protection leads to a loss

of manufacturing ﬁrms and “good jobs” in the exporting country, and the role of a TA is

to prevent governments from engaging in this beggar-thy-neighbor behavior.

Mrazova (2011) and Ossa (2012) focus on the implications of a different type of

policy externality that may arise under imperfect competition, namely the proﬁt-shifting

externality. The proﬁt-shifting effect of trade policies was ﬁrst studied by Brander and

Spencer (1984, 1985) in the context of a Cournot oligopoly with a ﬁxed number of

ﬁrms. Mrazova (2011) focuses on a setting similar to Brander and Spencer’s, while Ossa

(2012) focuses on a monopolistic competition model with a ﬁx ed number of ﬁrms. In

Ter ms of trade in this setting can be deﬁned in two different ways: as the ratio between the ex-factory price of a

foreign variety and that of a domestic variety, or as the ratio between the price index for exported varieties and that for

imported varieties. Ossa shows that with the ﬁrst deﬁnition tariffs do not affect TOT, and with the second deﬁnition

atariffworsens the country’s TOT. In the text I am implicitly adopting the ﬁrst of these two deﬁnitions. It is also

important to point out that, while import tariffs have no TOT effects (according to the ﬁrst deﬁnition above), export

taxes would have dollar-for-dollar effects on TOT, as emphasized by Bagwell and Staiger (2009).

There is also a counteracting effect, because the tar iff makes foreign products more expensive for consumers, but this

effect is shown to be dominated.

Ossa also argues that his model can provide a rationale for GATT’s rules of reciprocity and MFN, much in the same

way as a TOT model; I will come back to this aspect in Section 3.

International Trade Agreements 335

both models,governments can only use import tariffs.

Just like theTOT externality, the

proﬁt-shifting exter nality from a tariff is negative (holding TOT ﬁxed, an increase in the

tariff hurts the exporting country), and as a consequence, a mutually beneﬁcial TA must

reduce tariff levels relative to the noncooperative equilibrium. Mrazova (2011) in addition

shows that t he Bagwell-Staiger “test” fails in her setting (PO tariffs are inefﬁcient), thus

the purpose of a TA indeed goes beyond the correction of TOT externalities.

Enter Bagwell and Staiger. In two companion papers (Bagwell and Staiger, 2009,

2012a) they argue that, if import instruments and export instruments are available, even

in the presence of delocation or pr oﬁt-shifting externalities the only purpose of a TA

remains the correction of T OT externalities. They consider a number of possible market

structures, including monopoly, monopolistic competition, and Cournot oligopoly (with

or without fr ee entry, and with or without integrated markets), and show that, if countries

can use both import and export taxes and there are no income effects (quasi-linear pref-

erences), then PO policies are efﬁcient. On this basis, Bagwell and Staiger conclude that

neither delocation nor proﬁt-shifting externalities constitute a “fundamental” rationale

for TAs.

I will next try to illuminate the logic of Bagwell and Staiger’s argument by considering

a slightly more general setting. Focus on a two-country world with any number of goods,

and suppose gov ernments can choose speciﬁc trade taxes.

With a slight abuse of notation,

(τ, τ

∗

, p, p

∗

, p

) will now denote the vectors of trade taxes and prices.

Government objectives can always be expressed in reduced form as functions of trade

taxes ((τ,τ

∗

) and 

∗

(τ, τ

∗

)), and trade taxes in turn can be written as price wedges

(τ = p − p

and τ

∗

= p

∗

− p

), so government objectives can always be expressed as

functions of local, foreign, and world prices (

(p, p

∗

, p

) ≡ (p − p

, p

∗

− p

) and



∗

(p, p

∗

, p

) ≡ 

∗

(p − p

, p

∗

− p

)), regar d less of the nature of the international policy

externalities. Importantly, note that this setting allows for virtually any underlying market

structure.

Note also that a g overnment objective may depend on local prices in both

Mrazova justiﬁes the assumption that only tariffs are available by proposing a complementary theory that explains why

export subsidies have been banned by the GATT-WTO. I will be more speciﬁc on this part of her theory below,where

I focus on possible resolutions of the export subsidy puzzle.

Ossa has replied to this criticism by observing that in reality the use of expor t instruments is severely restricted: (i)

export subsidies have been banned by GATT a long time ago, and the subsequent rounds of negotiations have focused

mostly on import barriers,and (ii) the US has banned export taxes by constitution. My personal opinion is that Bagwell

and Staiger are correct in pointing out that a complete theor y should in principle explain, not assume, the pre-existing

restrictions on export instruments. However, it is not hard to imagine a model where there are transaction costs or

political frictions such that trade negotiations do not address import barriers and export instruments simultaneously

in a single round but rather in a sequential manner, or such that the unilateral use of expor t instruments is subject to

fr ictions. In such a richer model, delocation (or proﬁt-shifting) externalities would indeed be a distinct motive forTAs.

The argument can be easily extended to allow for production and consumption taxes.

Note that I have implicitly made two assumptions for simplicity. The ﬁrst is that a gover nment can apply different

trade taxes for different goods. If goods are differentiated and a gover nment must apply the same trade tax on all the

336 Giovanni Maggi

countries, and this is the new feature relative to the perfect-competition setting described

in Section 2.1. This i s a key point to keep in mind as we proceed: non-TOT externalities

can always be seen as local-price externalities.

Assume that there are no income effects. Together with the assumption that there are

only two countries, this implies that the local price of each good depends only on the

total trade tax on that good. This feature is the key to Bagwell and Staiger’s argument:

under the assumptions I just stated, the import tax and the export tax on a given good

are perfectly substitutable in affecting local prices.

The PO policies are deﬁned as the ones that would result if governments did not value

changes in p

. Since local prices depend only on total trade taxes (denoted ¯τ ≡ τ + τ

∗

this means that at the political optimum governments are effectively choosing the same

variables, ¯τ . Since the common choice of ¯τ must maximize both the Home objective

and t he Foreign objective, and since changes in p

act as pure transfers, it follows that

PO policies are efﬁcient. Formally, PO policies are deﬁned by the ﬁr st order conditions



¯τ



∗

¯τ

= 0and



∗

¯τ



∗

¯τ

= 0 (where the notation has the intuitive meaning).

Deﬁning global welfare as

 +



∗

, and noting that a change in p

does not affect global

welfare (it is a pure transfer), the FOC for global efﬁciency is (



¯τ

+ (



∗



∗

¯τ

= 0. Clearly, the PO policies satisfy the FOC for global efﬁciency.

I summarize this discussion with the following:

Efﬁciency of Political Optimum (EPO). Assume:(i) there are only two countries; (ii) there

are no income effects; and (iii) governments choose only trade taxes. Then PO policies

are efﬁcient, even in the presence of local-price externalities.

This result is in a way very general and in a way very special. The sense in which

it is v ery general is that it holds regardless of the nature of international policy exter-

nalities. Indeed, i t holds even in scenarios where intuition might suggest that there are

other motives for a TA be yond the corr ection of TOT externalities. For example, sup-

pose there ar e non-pecuniary international externalities, for example because of cross-

border pollution: the argument above is still valid, and hence PO policies are efﬁcient.

To be more concrete, consider a perfectly-competitive, par tial-equilibrium model with

a single good, where governments maximize welf are and there i s a cross-border pollu-

tion externality generated by production in the Foreign country. Let ϒ(x

∗

) denote the

environmental damage caused in the Home country by foreign production and x

∗

)

varieties of that good, the argument must be slightly adapted, but it still goes through. The second assumption is that

markets are integrated, so arbitrage implies that there is a single world (offshore) price for each good. If markets are

segmented there may be two-way trade in identical commodities, and so there may be two distinct offshore prices for

the same good, one for each direction of trade; but again, the argument is easily extended to cover this case.

Intuitively, note ﬁrst that the wedge between local prices is given by the total trade tax (p − p

∗

= τ + τ

∗

). Next

note that changing τ and τ

∗

in a way that leaves the total trade tax constant causes a transfer of revenue between the

countries. If there are no income effects, this transfer of revenue will not affect demand functions, thus only the total

trade tax matters for equilibrium local prices, not its composition.

International Trade Agreements 337

the foreign supply function. If Home welfare net of environmental damage is given by

W (p, p

) − ϒ(x

∗

)) ≡ (p, p

∗

, p

), and Foreign welfare is deﬁned analogously, then

the EPO result abov e immediately applies.

At the same time, the three assumptions (i)–(iii) are very restrictive, and even though

the EPO r esult only states a sufﬁcient condition, each of the three assumptions plays a

key role in delivering the efﬁciency of the PO policies: when local-price externalities are

present, if any of the three assumptions is violated then import-side policies and export -

side policies in general are not perfectly substitutable in affecting local prices, and hence

PO policies will typically be inefﬁcient .

Before concluding this subsection,I return one last time to the“export subsidy puzzle.”

In Section 2.2, I discussed how the domestic-commitment theory can provide a possible

resolution to this puzzle. As I discuss next, also models with ﬁrm-delocation and proﬁt-

shifting externalities offer avenues to address the puzzle.

Bagwell and Staiger (2012b) consider a linear Cournot delocation model á laVenables

(1985), where two governments choose trade taxes to maximize welfare. First they show

that, starting from free trade, a country acting unilaterally can increase its welfare with

a small export subsidy (because its beneﬁcial delocation effect outweighs its adverse

TOT effect) and with a small import tariff. This suggests that imposing a cap on export

subsidies may be jointly beneﬁcial for the two countries. However, it turns out that the

Nash equilibrium involves both import taxes and export taxes; what is responsible for this

sur prising result is the fact that import and export taxes are complementar y (increasing

the tariff makes an export tax more attractive). Thus, if governments negotiate over

import and export instruments starting from the Nash equilibrium, the model is not

able to explain why a TA would cap export subsidies. But if negotiations initially focus

only on import tariffs, bringing them close enough to zero, in a subsequent phase of

negotiation ther e will be scope for imposing a cap on export subsidies. In a similar vein,

DeRemer (2011) considers a monopolistic competition model where governments can

choose trade taxes and production subsidies. He argues that capping both export subsidies

The EPO result applies also if, in addition to trade taxes, governments can use pr oduction taxes,which are the ﬁrst-best

instrument to correct the environmental externality (see footnote 23).

To be clear, there may be special circumstances in which PO policies are efﬁcient even if some of the conditions (i)–(iii)

are not satisﬁed. For example, Bagwell and Staiger (2012a) show that PO policies are efﬁcient in a special three-country

setting with competing exporters. And in Antras and Staiger (2012a), as I will mention later, PO policies are efﬁcient

in the special case where governments maximize welfare, in spite of there being three countries in their model. But

these cases are rather special, and that is why I use the word “typically” in the text. Here I should also highlight the

relationship between my EPO result and a result shown by Bagwell and Staiger (2012a): they show that, if conditions

(i)–(iii) above are satisﬁed and (iv) gover nment preferences can be represented as functions of world and local pr ices,

then PO policies are efﬁcient. Importantly, they emphasize that condition (iv) can only be assessed given the speciﬁc

economic structure, and they check that this is the case in a number of imperfect-competition models.The value added

of my EPO result is to show that, if conditions (i)–(iii) are satisﬁed, there is no need to know anything else about the

economic structure or the nature of the international externalities to conclude that PO policies are efﬁcient.

338 Giovanni Maggi

and production subsidies may be desirable, but only if import tariffs are sufﬁciently close

to zero, not if they are close to their Nash equilibrium levels.

While the two models mentioned above are broadly suggestive of reasons wh y the

GATT-WTO has banned export subsidies, neither of them can explain an outcome

where export subsidies ar e present in the noncooperative equilibrium and a TA bans

export subsidies. Mrazova’s (2011) model can explain both of these features. Her model

allows for political economy considerations, so the Nash equilibrium may entail export

subsidies. The basic idea to explain the ban on export subsidies is the following. Recall

from Section 2.1.1 that an efﬁcient TA determines only net trade taxes, not the separate

import and export tax levels (see equation (3)) so that a given point on the efﬁciency

frontier can be achieved with import instruments alone or with export instruments alone.

Assuming a ﬁxed cost of administering each policy instrument, efﬁciency requires the use

of (at most) one policy instrument for each good. Mrazova then considers a repeated-

game model of TAs and argues that, due to proﬁt-shifting effects, a tariff-only agreement

is more easily self-enforced than a subsidy-only agreement, so an export subsidy ban is

desirable.

2.3.2. Trade-Volume Externalities Due to Bilateral Bargaining

When prices of international transactions are determined by bilateral bargaining rather

than by market clearing, the international externalities exerted by trade policy are of a dif-

ferent nature than the ones highlighted so far, and have nov el implications for the purpose

and design of TAs.This is the focus of two recent papers by Antras and Staiger (2012a,b).

It is convenient to start with Antras and Staiger (2012b), which focuses on a simple

matching model to highlight some key implications of bilateral bargaining forTAs. More

speciﬁcally, this paper considers a two-country, partial-equilibrium model where all pro-

ducers are located in the Foreign country. Each producer is matched with a consumer,

and within each match the quantity exchanged and the price are determined by bilateral

bargaining. Governments maximize welfare, with the For eign government choosing an

export tax and a domestic input subsidy, and the Home government choosing an import

tariff.

In this environment, international policy externalities cannot be viewed as traveling

simply through TOT. To understand the key difference between this environment and

a setting with market clearing, notice ﬁr st that in t he case of market clearing, Foreign

policies can affect the point of Home’s offer cur ve that is selected in equilibrium, but

cannot affect Home’s offer curve itself; so they can affect world price and trade volume

but cannot control them separately; in this sense, there is a single channel of international

policy externality, which can be viewed alternatively as a TOT externality or a trade-

volume externality. With bilateral bargaining, on the other hand, this link is broken, and

Foreign policies can affect TOT and trade volume separately, thus the rationale for a TA

is to jointly correct these two separate externalities. Indeed, with bilateral bargaining, PO

International Trade Agreements 339

policies can be shown to be inefﬁcient, so the purpose of aTA goes beyond the correction

of TOT externalities.

Pricing by bilateral bargaining is particularly relevant when ﬁrms offshore the pro-

duction of specialized inputs and there is incomplete contracting between downstream

and upstream producers. The implications of offshoring for TAs are the focus of Antras

and Staiger (2012a). This paper considers an environment with three countries: Home,

Foreign, and the rest of the wo rld (ROW). Home is the sole producer of a ﬁnal good that

requires a custom-made intermediate input; Foreign is the sole producer of the inter-

mediate input; and ROW specializes in a plain-vanilla numeraire good. Once a Foreign

upstream ﬁrm and a Home downstream ﬁrm are matched, the former must customize the

intermediate input for the latter, and then the price of the input is determined by bilateral

bargaining. The customization of the input cannot be contracted upon ex-ante, thus a

standard hold-up problem arises. Each government chooses trade taxes to maximize a

possibly politically adjusted social welfare function. In this setting, beyond standardTOT

externalities, trade policies exert trade-volume externalities of the kind described above,

with the additional feature that, by affecting trade volumes, a country’s trade policies can

affect the severity of the hold-up problem in the other country.

In this environment,Antras and Staiger (2012a) show that PO policies are in general

inefﬁcient, except in the special case where governments maximize welfare. Thus, as long

as there are political-economy considerations in the gov ernments’ objectives, the rationale

for TAs goes beyond the cor rection of TOT externalities.

It is clear t hat, in this setting, the international externalities from trade policies do

not simply travel through TOT. However, as the EPO result of Section 2.3.1 highlights,

the presence of non-T OT externalities is not sufﬁcient to conclude that PO policies are

inefﬁcient. So why does the EPO result not apply here? The key reason is that in this

setting there are more than two countries, and as a consequence, export taxes and import

taxes are not perfectly substitutable in affecting local prices.

To conﬁrm this point, I note

that if Antras and Staiger’s (2012a) model were played out in a two-country world, then

the PO policies would be always efﬁcient, regardless of the government objectives.Thus in

The implication that PO policies are inefﬁcient is not highlighted in the paper, but it is easy to show.The paper focuses

on a design question, namely whether a “shallow integration” approach can ach ieve global efﬁciency, but the paper

is relevant also for the question of the purpose of a TA. I will come back to the design question in Section 3.Also,

it is interesting to note that the EPO result presented in Section 2.3.1 does not apply in Antras and Staiger’s (2012b)

setting, because trade taxes are not the only policy instruments: Foreign can use also a non-wedge policy, namely an

input subsidy, and there is no Home policy that is a perfect substitute for it in affecting local prices.

A second question examined by Antras and Staiger (2012a) concerns the desirability of a“shallow integration”approach.

Again, I will focus on this design question in Section 3.

Intuitively, recall that Home is the sole producer of the ﬁnal good, so the local price of this good in the Home market

(say p

) depends on all three trade taxes on this good, but not through their sum, because Home’s export tax has

a larger impact on p

than each of the other countries’ import tax. Also note that the reason why the EPO result

does not apply in Antras and Staiger (2012a) is different than the one in Antras and Staiger (2012b). As highlighted in

footnote 28, in the latter model (which has only two countries) the reason the EPO result does not apply is that trade

taxes are not the only policy instruments.

340 Giovanni Maggi

some sense the mor e surprising aspect of the results inAntras and Staiger (2012a) is not that

the PO policies are inefﬁcient in the presence of political-economy considerations, but

rather that the PO policies are efﬁcient in the case of welfare-maximizing governments.

2.4. The Uncertainty-Managing Motive for a TA

The papers discussed so far abstract from the presence of uncertainty, and highlight the

gains that a TA can offer by changing the levels of trade policies relative to the non-

cooperative equilibrium. But if the political/economic environment is uncertain, one

can distinguish between an “uncertainty managing” motive and a “mean manag ing”

motive for a TA. Suppose that, because of shocks in the political/economic environment,

noncooperative trade policies are themselves subject to shocks. We can then ask the

following question: can go vernments achieve mutual gains by changing the degree of

uncertainty in trade policies relative to the noncooperative equilibrium, holding their mean

levels constant? If the answer is yes, we can say that there is an “ uncer tainty managing”

motive for a TA. Limão and Maggi (2013) examine under what conditions there exists

an uncertainty-managing motive for a TA, whether it calls for a reduction or an increase

in policy uncertainty, and what are the potential gains from a TA that regulates policy

uncertainty.

In a standard competitive trade model with risk neutrality, where trade policies exert

international externalities only through TOT, Limão and Maggi show that there tends

to be an uncertainty-increasing motive for a TA, due to the convexity of indirect utility

and revenue functions in prices. This model thus seems at odds with the often-heard

argument that TAs can provide welfare gains by r educing trade policy uncertainty. When

individuals are risk averse, on the other hand, the direction of the uncertainty motive

for a TA is determined by a trade-off between risk aversion and ﬂexibility: the degree of

risk aversion, in interaction with the degree of openness, favors an uncertainty-reducing

motive; while the degree of ﬂexibility of the economy, which is in turn determined

by the export supply elasticity and the degree of production diversiﬁcation, favors an

uncertainty-increasing motive. Empirically lower-income countries tend to have lower

export supply elasticities and a lo wer degree of diversiﬁcation, thus Limão and Maggi’s

model suggests that the uncertainty-reducing motive might be relatively more important

for lower-income countries. Another key result is that, as trade costs decline, the gains

from reducing trade policy uncertainty tend to become more import ant relative to the

gains from reducing average trade barriers. A broad implication of this ﬁnding is that

Policy makers and practitioners often argue that one of the main goals of TAs is to reduce uncer tainty in trade policies,

and various TAs including the WTO include “mission” statements that can be interpreted along similar lines. For

example, the WTO states in its website that “Just as important as freer trade—perhaps more important—are other

principles of the WTO system. For example: non-discrimination, and making sure the conditions for trade are stable,

predictable and transparent.”

International Trade Agreements 341

uncertainty-reducing motives for TAs are likely to emerge as the world becomes more

integrated, and are more likely to be present for countries within a region.

2.5. Empirical Evidence

The empirical literature on TAs is still in its inf ancy, but it has seen a considerable accel-

eration in the last decade or so. In this section I discuss some papers that attempt to

get at the underlying motives for TAs, and some that examine the impacts of TAs on

trade barriers and trade ﬂows in a more descriptive way. I postpone a discussion of the

empirical work on regional trade agreements to Section 4.

2.5.1. Tests of the TOT Theory

A number of recent papers have set out to test the predictions of the TOT theory.

Four papers stand out in this group. The ﬁrst one is Broda et al. (2008),who focus

on the prediction that, in a noncooperative scenario, tariffs should tend to be higher

for countries/goods where market pow er (the inverse of the export supply elasticity) is

higher. Broda et al. consider the tariffs set by 15 non-WTO countries,on the presumption

that these countries choose trade policies in a noncooperative manner. They estimate

export supply elasticities by country and good—a signiﬁcant contribution in itself—and

ﬁnd that these elasticities are related with t ariffs in the way predicted by the theor y,

particularly if one focuses on the variation across goods within a country. Next they

control for political-economy determinants of tariffs, using a parsimonious speciﬁcation

á la Grossman-Helpman (with the additional assumption that all industr ies are politically

organized), and ﬁnd that export supply elasticities retain explanatory power even in the

extended speciﬁcation.

Bagwell and Staiger (2011) test the predictions of the TOT model regarding the

tariff cuts that a country should make when acceding the WTO. Bagwell and Staiger

start by showing that, if demand and supply functions are linear, the model predicts that

the tariff cut should be deeper, other things equal, when the noncooperative volume

of imports is higher. They then test this pr ediction across six-digit HS level goods and

across 16 countries that acceded the WTO between 1995 and 2005, ﬁnding a strong

positive correlation. The corr elation survives also in the presence of country and good

Also Handley (2012) and Handley and Limão (2012, 2013) focus on the links between uncertainty and TAs, but from

a very different perspective. They examine (theoretically and empirically) the impact that TAs have on trade ﬂows by

removing the risk of future increases in protection,taking trade policy (before and after theTA) as exogenous. Handley

and Limão do not consider the key questions addressed in Limão and Maggi (2013), namely whether there is “too

much” or “too little” uncertainty in the noncooperative trade policies and what are the gains from “correcting” the

degree of policy uncertainty through a TA. I will come back to the papers by Handley and Limão in the next section.

However, there is one ﬁnding in Broda et al. (2008) that is not easy to reconcile with the TOT theory. According

to the theor y, a country acting noncooperatively should set discriminatory tar iffs, because export supply elasticities

vary across exporters, but this almost never happens in the data. The authors argue that the presence of administrative

costs can reconcile this observation with the theory, but these costs would have to be very high, because export supply

elasticities vary widely across exporters and therefore the potential gains from discrimination are high.

342 Giovanni Maggi

ﬁxed effects,and importantly, it passes the “placebo”test that it should hold only for tariffs

on imports from other WTO members, not on imports from non-WTO countries.

The papers discussed above focus on non-WTO countries (Broda et al.) or countries

that recently joined the WTO (Bagwell and Staiger), so they leave out the vast majority

of current WTO countries. Ludema and Mayda (2010) focus instead on the MFN tariffs

of all WTO members. Their test of the TOT theory is based on the following idea: the

MFN rule causes a well-known free-rider problem in multilateral negotiations, and for

this r eason negotiations are only partially successful in removingTOT considerations from

tariff levels, therefore the negotiated MFN tariffs should still partially re ﬂect the market

power of importing countries. Moreover,the correlation between MFN tariffs and market

power should be stronger when exporter concentration (as measured for example by the

Herﬁndahl index) is lower, because in this case the free-rider problem is more severe,

thus MFN tariffs should be negatively related to the product of exporter concentration

and importer market power. Ludema and Mayda test this prediction on a cross-section

of MFN tariffs set by WTO members in the Uruguay Round, ﬁnding supportive results.

Finally, Bown and Crowley (2013a) test the predictions of a repeated-game version

of the TOT model (namely, Bagwell and Staiger’s (1990) model of “managed trade”)

using data on US temporary tariffs imposed under the US’s antidumping and safeguard

laws over 1997–2006. The key idea of the model is that, if a TA is to be self-enforcing,

it needs to provide for “escape clauses” that allow countries to raise tariffs in periods

when the temptation to defect from the agreement is stronger, that is when the incentive

to manipulate TOT is stronger, which in turn tends to happen when trade volumes are

higher.

Thus a key prediction of the model is that temporary tariffs should be observed

with higher lik elihood when import volumes are higher. Bown and Crowley ﬁnd strong

support for this prediction in the data.

Finally, I should mention that there are a number of empirical studies documenting

that a countr y’s tariffs can signiﬁcantly affect its TOT, which of course is a pre-requisite

for the empirical relevance of the T OT theory. P apers in this group include Kreinin

(1961),Winters and Chang (2000, 2002),and Bown and Crowley (2006).

2.5.2. Tests of the Domestic-Commitment Theory

As a whole , the studies discussed abo ve are quite supportive of the T OT theory. At the

same time, I do not think this body of research has established that the TOT motive is

the only empirically signiﬁcant motive forTAs. This leads me to the next question, which

is whether other motives for TA s are empirically important. The short answer to this

Bagwell and Staiger also consider a more general relationship predicted by TOT theory between tar iff cuts, trade

elasticities and import volumes, and test this prediction using Broda et al.’s (2008) trade elasticities for the ﬁve

countries with in their sample for which such elasticities are available. Also the results of this test are consistent with

the predictions of the theory.

See Section 3.2.1 for fur ther discussion of repeated-game versions of the TOT model.

International Trade Agreements 343

question is that we do not know yet: domestic-commitment theories and New Trade

theor i es of TAs have thus far received less empirical attention than the TOT theory, and

the jury is still out. I will start by focusing on the empirical research on the domestic-

commitment theory.

The ﬁrst paper in this area is by Staiger and Tabellini (1999), who test a prediction

of their theoretical model (Staiger and Tabellini, 1987), in which the government is

subject to a time-inconsistency problem due to the fact that it chooses trade policy

after domestic agents have made their allocation decisions. This model suggests that, if

the government commits to a TA to address this time-inconsistency problem, the TA

should lead to deeper trade liberalization in sectors where the potential for production

distortions from protection is larger. Staiger and Tabellini test this prediction by focusing

on the sectoral exclusions chosen by the US government in the Tokyo Round of GATT,

using as “control” gr oup the US tariff decisions made under the GATT’s escape clause,

which arguably were not effectively constrained by GATT. Their ﬁndings are broadly

supportive of the model’s prediction.

Limão and Tovar (2011) test their theoretical model (see Section 2.2) using data

on tariffs and non-tariff barriers (NTBs) in Turkey. One key prediction of their model

is that a government is more likely to commit to a tariff cap i n industries where its

bargaining power relative to the lobby is lower, and conditional on committing, the

tariff cap should be tighter when the government’s bargaining power is lower . A key

ingredient in testing this predi ction is measuring the government’s relative bargaining

power industry by industry. To do so, Limão and Tovar posit that the relative bargaining

power of the government in a given industry is lower when the rate of ﬁrm exit in

that industry is lower. The i dea is that, if the exit rate is lower, the ﬁrms (and the lobby

that represents them) discount the future less, while the government’s discount rate does

not v ary across industries, and noncooperative bargaining theory suggests that a player’s

relative bargaining power is higher when his or her relative patience is higher. Using their

estimates of relative bargaining powers, Limão andTovar ﬁnd that in the Uruguay Round

the Turkish government indeed committed to less stringent tariff bindings in industries

where its relative bargaining pow er was stronger, and did not commit at all if the latter was

strong enough.

It is interesting to note that this ﬁnding is broadly consistent also with

the predictions of Maggi and Rodriguez-Clare’s (1998, 2007) version of the domestic-

commitment theory.

Liu and Ornelas (2012) test their t heory that aTA may serve as a commitment device

for a fragile democracy to destro y protectionistic rents and hence reduce the lik elihood of

Another ﬁnding in Limão and Tovar (2011) is that Turkey used NTBs more heavily after the tariff bindings were

imposed. This ﬁnding as well is consistent with their model, but I note that this kind of policy substitution between

tariffs and NTBs is consistent also with a variety of models that do not feature domestic-commitment motives for aTA.

In Maggi and Rodr iguez-Clare (1998) the government is more likely to commit to free trade when its bargaining

power is lower, and in Maggi and Rodriguez-Clare (2007) tar iff caps tend to be tighter when governments have lower

bargaining power (see section I.D of that paper).

344 Giovanni Maggi

coups by rent-seeking authoritarian groups (see Section 2.2 ),by using data on preferential

trade agreements for 116 countries over the period 1960–2007. In line with their model’s

predictions, Liu and Ornelas ﬁnd that more fragile democracies are i ndeed more likely

to sign preferential TAs, and that signing a preferential TA in turn lowers the likelihood

of democratic failure.

I would summarize the thin empirical literature on the domestic-commitment theory

of TAs by saying that it has found support for some predictions of some versions of the

theory, but a broad and systematic empiri cal investigation of this theor y is still missing.

Ultimately, the hope is to be able to quantify the relative importance of TOT motives

and domestic-commitment motives for TAs, but this is certainly no easy task.

2.5.3. Empirical Work on the New Trade Theory

Empirical research focused on New Trade theories of TAs is at the very beginning. I

am not aware of any attempts to test these theories with econometric approaches, but a

recent paper by Ossa (2013) takes the theory to the data using a calibration approach.

Ossa develops a multi-country model that allows for three drivers of trade protection:

T OT effects, proﬁt-shifting effects, and political-economy considerations.

The model,

which combines elements from Krugman (1980) and Grossman and Helpman (1995a),is

calibrated to match observed trade ﬂows and tar iffs at the industry level in 2005. Using a

technique introduced by Dekle et al. (2007), Ossa performs counterfactual analysis using

only estimates of the elasticities of substitution (taken from Broda and Weinstein, 2006),

estimates of political-economy weights (taken from Goldberg and Maggi, 1999), and

f actual levels of trade ﬂows and tariffs. Several interesting ﬁndings arise. First,TOT and

proﬁt-shifting drivers of protection quantitatively dominate political-economy drivers.

Second, a global trade war would lead to t ariffs averaging about 60% across industries and

countries, and would reduce welfare by about 3.5% relative to the cooperative outcome.

Finally, relative to where we are today, the potential gains f rom further multilateral trade

negotiations are negligible.

Whether these are “numbers we can believe in” is not obvious, given the very stylized

nature of the model, but this is a thought-provoking paper that points to a promising way

forward for addressing important questions such as quantifying t he relative importance

of different motives for trade protection, the gains achieved by pastTAs, and the potential

gains from future TAs.

2.5.4. Impacts of the GATT-WTO

In this subsection I brieﬂy discuss a recent wave of papers that have examined the impact

of the GATT-WTO on trade barri ers and trade ﬂows. This literature was triggered by

An earlier attempt at quantifying New Trade motives for protection can be found in Ossa (2011), in the context of a

model featuring only ﬁrm-delocation effects. In that calibration exercise, Ossa allows each country to set only a single

tariff on all imports, and only focuses on noncooperative tar iffs.

International Trade Agreements 345

Rose (2004a), who sent shockwaves through the trade policy community (academic and

not) b y arguing that the WTO had virtually no impact on trade ﬂows, based on a simple

reduced-form regression analysis. In a similar vein, Rose (2004b) argued that the WTO

had a negligible effect on the trade policies actually applied by countries.

These papers spawned a number of follow-up studies, most of which qualiﬁed Rose’s

results in signiﬁcant ways. Subramanian andWei (2007) show that the impact of WTO has

been very uneven across countries and sectors, for example because developing countries

enjoyed exemptions from trade liberalization in speciﬁc sectors (such as textiles); once

these exceptions are accounted for,theWTO is found to signiﬁcantly promote trade.To m z

et al. (2007) argue that many countries were mistakenly classiﬁed as non-members of the

GATT, while in reality they were de facto members with similar rights and obligations as

formal members, and show that this misclassiﬁcation leads to underestimating the effect

of GATT on trade ﬂows. Liu (2009) shows the importance of “zeroes” in bilateral trade

ﬂows: if one takes into account that the WTO has lead new country pairs to initiate

bilateral trade—the “extensive partner-level margin” of trade—then the WT O is found

to have a signiﬁcant positive effect on trade. Dutt et al. (2011) ﬁnd that the impact of

WTO membership is signiﬁcant on the extensive product margin of trade, that is,WTO

membership leads to an increase in the number of goods traded, but the impact of the

WTO is negligible on the intensive margin (the trade volume of already-traded goods).

Next I discuss some recent papers that also examine the effects of TAs on trade ﬂows,

but use more structural approaches, and provide some evidence about the mechanisms

thr ough which a TA affects trade.

Eicher and Henn (2011) examine the effects of WTO and regional trade agreements

on trade ﬂows by considering a panel of 177 countries over 50 years. They start with

a reduced-form gravity approach that encompasses t he speciﬁcations by Rose (2004a),

Tomz et al. (2007),and Subramanian and Wei (2007), and ﬁnd that only regional trade

agreements have a signiﬁcant impact on trade, not the WTO. Then they consider an

augmented gravity equation that incorporates a key effect suggested by the TOT theory,

namely that countries with more marke t power should agree to bigger tariff cuts as they

join the WTO, and hence their trade volumes should increase by more. When a measure

of market power (pre-accession import volumes) is incorporated in the reg ressions, the

WT O is found to have a signiﬁcant effect for countries with import volumes above the

85th percentile. This ﬁnding contributes to reconcile the seemingly contradictory results

of reduced-form studies á la Rose (2004a) and theory-dri ven studies á la Bagwell and

Staiger (2010a).

Finally, Handley (2012) and Handley and Limão (2012, 2013) show that the mecha-

nisms through which TAs affect trade ﬂows may be more subtle than just a decrease of

tariff levels. These papers argue that, when trade policies are subject to shocks, export-

Another paper worthy of mention is Tang and Wei (2010), which examines the impact of WTO on GDP g rowth,

ﬁnding that WTO membership tends to be associated with higher GDP growth rates for developing countries.

346 Giovanni Maggi

ing ﬁrms respond not just to changes in the applied levels of trade barriers, but also

to changes in the probability that trade barriers might be raised in the future. Thus, by

reducing the risk of future protectionist spikes, a TA may encourage investment in export

markets and boost trade volumes even if no change in applied policy levels is observed.

Handley (2012) focuses onAustralia’s accession toWTO, ﬁnding evidence that this caused

an increase in exports to Australia more because committing to WTO bindings removed

the risk of future “bad news” for exporters, than because of actual reductions in Aus-

tralia’s applied tar iffs. Handley and Limão (2012) ﬁnd evidence that Portugal’s accession

to the EC boosted Portuguese exports to other EC countries in spite of the fact that

Portugal already enjoyed free access to those countries before accession, thanks to pre-

existing preferences, and estimate that a signiﬁcant fraction of this effect was due to the

f act that accession to the EC eliminated the risk faced by Port uguese exporters of losing

pre-existing preferences. Finally, Handley and Limão (2013) estimate that a signiﬁcant

portion of China’s rapid increase in exports to the US starting in 2001 is explained by the

permanent MFN status gained by China as a consequence of its WTO accession, which

removed t he US threat of imposing “column 2” tariffs on impor t s from China.

3. THE DESIGN OF TRADE AGREEMENTS

In a world without transaction costs, the issue of how to design a TA would be rather

uninteresting. Imagine for a moment that governments could costlessly write a complete

contract that covers all relevant policies and contingencies and can be perfectly enforced.

In such a world (whi ch is the world I effectively considered in the previous section) there

would be no need to think hard about how to design a TA: go vernments would be able

to achieve a fully efﬁcient outcome by writing a complete contract. Such contract would

contain a large amount of detail, but would be conceptually straightforward.

In this section I will discuss the literature on the design of TAs as viewed from

the perspective of transaction costs. To be a bit more speciﬁc, I will use the expression

“transaction costs” as an umbrella term that encompasses two categor ies of frictions: (1)

contracting frictions, which include costs of negotiating and writing contracts, imperfect

veriﬁability, and private information, and (2) imperfect enforcement, by which I mean t he

lack of external enforcement power, so that TAs must be self-enforcing contracts.

Note that, while contracting frictions naturally lead to contract incompleteness,

meaning that rele vant contingencies and/or policies are missing from the contract,

enforcement frictions typically do not generate contract incompleteness: the presence

of self-enforcement constraints per se is not a reason for removing contingencies or poli-

cies from the contract. In fact, it can have the opposite effect, in the sense of inducing

governments to intr oduce contingencies that otherwise would not be present in the

contract. For example, self-enforcement constraints may require the agreement to be

contingent on import shocks (escape clauses), whereas a perfectly enforceable ag reement

International Trade Agreements 347

would not need to be contingent.

For this reason, below I organize my discussion in

two parts: ﬁrst I will focus on the implications of contract incompleteness for TA design,

and then I will focus on the implications of imperfect enforcement.

What would we miss if we ignored the presence of transaction costs? Why not stop

at the world considered in the previous section, wher e the only questions concern the

motives for a TA and how the efﬁcient policies differ from noncooperative policies? If

theory stopped there, it would not be able to explain a number of important features of

real-worldTAs. For example, it would be hard to explain why the GATT-WTO speciﬁes

tariff caps instead of exact tariff commitments; or why many domestic policies such as

standards or domestic tax es are left to the gov ernments’ discretion; and so on.

Also, if w e ignored transaction costs it w ould be difﬁcult to explain the nature of

trade disputes in theWTO and the role of the WTO court, the Dispute Settlement Body

(DSB): absent transaction costs the only possible role for a court w ould be to enforce

the obligations speciﬁed in the agreement, but in reality trade disputes are more often

about the interpretation of vague obligations, or instances for which the agreement is

silent, than about the enforcement of clear obligations. Thus a potentially important role

of the DSB is to “complete” various dimensions of an incomplete contract, and therefore

designing the role of the DSB becomes of key impor t ance.

Before proceeding, it is useful to distinguish between three dimensions of TA design:

(1) the design of substantive policy rules, that is constraints on governments’ policy choices

(e.g. tariff ceilings, non-discrimination rules); (2) the design of enforcement rules (how

should governments behave after a violation of the agreement?); and (3) the design of

procedures,such as dispute settlement procedures.

I will argue that introducing transaction

costs in our conceptual frame can take us a long way toward understanding the design of

TAs along these three dimensions.

3.1. Contract Incompleteness and Trade Agreements

The overarching theme of this section is that viewingTAs as incomplete contracts can help

understand the way TAs are designed. In the models I will discuss below, the incomplete-

ness of the TA is sometimes derived endogenously from contracting frictions, sometimes

assumed exogenously, and sometimes left implicit. I would argue that modeling con-

tracting fr ictions explicitly has the advantage of forcing us to think in a disciplined way

about the rationale for the rules and procedures we observe in real-world TAs. I am not

advocating a dogmatic approach where everything must be explained from “ﬁrst princi-

ples,” but I think that too often a theoretical “story” proposed to explain a certain rule

This is the case for example in Bagwell and Staiger’s (1990) model of “managed trade” (see Section 3.2.1).

Another important type of procedure is the bargaining protocol for trade negotiations. In spite of the obvious impor tance

of negotiation protocols, however, I am not aware of any formal literature addressing this question. On a distinct note,

a paper that focuses on questions of procedure but does not ﬁt easily in the taxonomy above is Conconi et al. (2012),

which examines how domestic fast-track procedures for cong ressional approval of trade agreements affect the outcome

of trade negotiations.

348 Giovanni Maggi

X has some intuitive appeal, but does not stand up to a more rigorous test, which is the

following question: can r ule X be part of an optimally designed contract, at least for some

plausible contracting environment?

I will organize the discussion below as follows. I will ﬁrst lay out a simple model of

TAs where two distinct forms of contract incompleteness—rigidity and discretion—arise

endogenously from contracting frictions, and argue that this model can help explain a

number of design features of the observedTAs. I will then discuss other models that have

been proposed in the literature to explain speciﬁc rules such as tariff caps, reciprocity,

MFN, market-access rules, and “liability” v s “property” rules. Finally, I will focus on the

design of dispute settlement procedures.

3.1.1. Contracting costs, rigidity, and discretion

To lay out some basic concepts I will start by outlining a model by Horn et al. (2010),

where the incompleteness of a TA emerges endogenously from the costs of writing a

contract. In spite of its simplicity, this form of contracting friction can go surprisingly far

in explaining the way TAs are designed.

At the basis of this model is the observation that there are two important sources of

complexity in writing a TA: (a) uncertainty about future economic/political conditions,

which calls for agreements that are highly contingent, and (b) the wide array of policies

(domestic and border measures) that can affect trade, which in turn calls for agreements

that are very comprehensive in their policy coverage. For these reasons,writing a complete

contract would be very costly, since all contingencies and policies would need to be

described ex-ante and veriﬁed ex-post.

In this context, one can think of two ways to

save on writing costs: lea ving contingencies out of the contract, which leads to rigidity,or

leaving policies out of the contract, which amounts to intr oducing discretion.

Real-world TAs exhibit an interesting combination of rigidity and discretion. For

example, the GATT-WTO binds trade instruments, but domestic instruments are largely

left to discretion,except that (i) theWTO has introduced regulation of domestic subsidies,

and (ii) all domestic policies must satisfy the National Treatment rule. Also, constraints

on tariffs take the form of ceilings (so governments have downward discretion), and such

ceilings are largely rigid, but the contract also provides for “escape clauses” under some

contingencies. Horn et al. argue that the presence of contracting costs can help explain

these design features.

Writing costs can be interpreted more broadly as captur ing also the costs of negotiating a TA and the costs of verifying

contingencies and policies. I also note here that this approach to modeling endogenous contract incompleteness was

ﬁrst developed, though in a different setting, by Battigalli and Maggi (2002).

Are contracting costs empirically important for an agreement such as theWTO?This is a legitimate question,but given

the huge number of products, countries, policy instruments, and contingencies that are involved in the WTO, and the

fact that this agreement took eight years of negotiations to complete, contracting costs seem quantitatively important.

Many trade-law scholars agree with this view. For example, Schwar tz and Sykes (2002) write: “...Many contracts are

negotiated under conditions of considerable complexity and uncertainty, and it is not economical for the par ties to

specify in advance how they ought to behave under ever y conceivable contingency.” (pp. 181–182)

International Trade Agreements 349

Horn et al. consider a two-country, partial-equilibrium model where markets are

perfectly competitive but there may be production and consumption (localized) exter-

nalities, so that there is an efﬁciency rationale for multiple policies, in particular tariffs

and production subsidies.There can be uncertainty both in the underlying trade volume

and in t he externality levels, so that a ﬁrst-best agreement would need to specify policies

in a state-contingent way. At the core of the model is the assumption that the cost of the

agreement is increasing in the number of policy instruments and contingencies it speciﬁes.

The analysis focuses on four questions of TA design. The ﬁrst one is whether it is

desirable to leave domestic subsidies to the governments’ discretion. Horn et al. ﬁnd

that, for a given level of contracting costs, this is more lik ely if: (a) trade volumes are

low, so that countries gain little fr o m manipulating TOT; or (b) countries have little

monopoly power in trade, so that they have little capacity to manipulate TOT; or (c)

subsi dies are not a good substitute for tariffs as a means of manipulatingTOT. The trade-

volume effect at point (a) suggests an explanation for why domestic subsidies hav e been

constrained byWTO while they were largely left to discretion under GATT, namely, that

a general expansion of trade volumes over time has made i t more costly to leave subsidies

to discretion.Moreover, the “monopoly power” and “instrument substitutability” effects

at points (b) and (c) together suggest a possible explanation for why developing countries

have been largely exempted from constraints on subsidies through“special and differential

treatment” clauses: developing countries typically do not hav e str ong market power and

do not have a broad range of domestic policy instruments that can easily substitute for

tariffs.

The second design question is whether and how the TA should be state-contingent.

An interesting result in this regard is that, conditional on leaving domestic subsidies to dis-

cretion, it can be optimal to specify an escape clause that allows a government to increase

its tariff when the level of imports is high, as a way to mitigate the stronger incentives

to distort domestic subsidies in periods of high import v olume. Thus the model suggests

a novel explanation for the desirability of escape clauses in TAs: these can be attrac-

tive because they provide an indirect means of managing the distortions associated with

leaving domestic policies to discretion.

The third point made by Horn et al. is that the presence of contracting costs can

explain why tariffs are constrained by ceilings rather than by exact levels. More speciﬁcally,

the optimal agreement may include rigid tariff ceilings. Intuitively, conditional on some

contingencies being missing from the agreement, leaving downward discretion in tariffs

can only be beneﬁcial, since a government is always t empted to distort tariffs upwards,

and there may be states of the world where the unilaterally optimal tariff lies below the

Note that this explanation for escape clauses is very different from others that have been proposed in the theoretical

literature, and in particular those that are based on self-enforcement considerations (e.g. Bagwell and Staiger, 1990,

discussed in Section 3.2.1).

350 Giovanni Maggi

ceiling. In Section 3.1.2, I will discuss other possible explanations for the use of tariff

ceilings.

Finally, the model by Horn et al. can provide a novel rationale for the National

Treatment (NT) rule, showing that such a rule can serve to save on contracting costs.

To mak e this point, Horn et al. extend the basic model outlined abo ve by allowing for

consumption taxes on domestically-pro duced and imported goods,and formalize the NT

rule as a constraint that these taxes be equalized.A preliminary observation is that aTA that

imposes the NT rule but leaves discretion over the (non-discriminatory) consumption

tax can achieve a new form of discretion that cannot be achieved without the NT rule,

namely, discretion over the consumer price w edge;indeed, a non-NT agreement can only

leave discretion over the producer price wedge.

Horn et al. then show that under some

conditions the NT rule can indeed be part of an optimal TA. The key observation here

is that leaving discretion over the (non-discriminatory) consumption tax may involve a

subtle beneﬁt, namely that this tax will be responsive to contingencies, thus this form of

discretion is an indirect way to make theTA state-contingent. If specifying contingencies

is quite costly and the “indirect state-contingency” effect just highlighted is important,

then introducing the NT rule in the TA may be optimal.

Before proceeding, I emphasize that this model focuses on a setting where TAs are

motivated by TOT externalities, but an interesting and still unexplored question is the

extent to which an incomplete-contracting approach of this kind but applied in the con-

text of the domestic-commitment theory might generate new insights and help interpret

features of real-world TAs.

3.1.2. Tariﬀ Caps

As I pointed out in the previous section, tariff restrictions in the GATT-WTO take the

form of tariff caps. The rationale for the use of tariff caps has been the subject of several

To understand this slightly cryptic statement,focus on the importer country. In the absence of NT, the wedge between

consumer price and world price is given by p − p

= τ + t

,whereτ is the tariff and t

is the consumption tax on

the imported good, while the wedge between producer price and world price is q − p

= τ + t

+ s − t

,wheres

is the production subsidy and t

the consumption tax on the domestically-produced good. Next focus on the price

wedges under the NT rule, which imposes the constraint t

= t

.Lettingt denote the common level of the internal

tax, the NT rule transforms the set of policy instruments from (τ, s, t

, t

) to (τ, s, t), and the price wedges become

p − p

= τ + t and q − p

= τ + s. Now notice that with an NT-based ag reement that constrains τ and s and leaves

t to discretion, it is possible to tie down the producer price wedge q − p

while leaving discretion over the consumer

price wedge p − p

, whereas this is not possible with a non-NT agreement.

The role of the NT rule for domestic taxes has been examined also by Hor n (2006). He considers a model where TAs

are exogenously incomplete contracts that can include tar iff bindings and an NT rule for consumption taxes. Domestic

and imported goods are differentiated, so efﬁciency may call for discriminatory consumption taxes, but Horn shows

that,if tariff bindings are set at appropriate levels, including an NT rule always improves global welfare. Saggi and Sara

(2008) extend Horn’s (2006) analysis by allowing for heterogeneity in product quality and countr y size. There is also

a small literature that examines the role of the NT rule for product standards: see in particular Battigalli and Maggi

(2003), Costinot (2008),and Staiger and Sykes (2011).

International Trade Agreements 351

papers in the literature (in addition to Horn et al., 2010 and Magg i and Rodriguez-Clare,

2007, already discussed above).

Bagwell and Staiger (2005a) consider a model where governments have private infor-

mation about domestic political-econom y shocks, and show that in such a setting tariff

caps tend to be preferable to exact tariff commitments.The intuition is similar as in Horn

et al. (2010): since the TA cannot be contingent on the political-econom y shock, and

since a government’s temptation is to distort the tariff upward, leaving downward ﬂexibil-

ity cannot hurt, and it is strictly preferred if the support of the shock is sufﬁciently wide.

However, just as in Horn et al. (2010), this model can explain only why we do not observe

exact tariff commitments, and it stops short of characterizing the optimal tariff rule.

Amador and Bagwell (2013) is the ﬁrst paper that provides a full theoretical explanation

of tariff caps, by showing that under some conditions a tariff cap is not only preferable

to an exact tariff commitment, but is also the optimal tariff rule. Speciﬁcally, Amador

and Bagwell consider a partial-equilibrium model where an import tariff is the only

available policy instrument, and the importing government has private or non-veriﬁable

information about domestic political pressures. Contingent transfers are not available,

thus aTA can only specify a set of permissible tariffs that the importing government may

apply.The governments’objective functions are speciﬁed in reduced form,say (τ,γ) for

the Home government (where τ is the tariff and γ a political-economy parameter) and



∗

(τ ) for the Foreign government.

A tariff cap is shown to be optimal if the convexity

of the Foreign objective function (

∗

ττ

) is not too pr onounced relative to the concavity

of the Home objective function (

ττ

) and the density of γ does not decrease too fast.

Amador and Bagwell then examine when these conditions are satisﬁed in the context

of two speciﬁc market structures, a perfectly competitive setting and a monopolistic

competition setting.

In reality, some governments do exercise the downward discretion afforded by tariff

caps and apply tariffs strictly below the cap levels (the so-called “binding overhang”).

Empirically there is considerable variation in the extent of binding o verhang, as well as

in the levels of tariff ceilings, acr oss countries and sectors (see for example Bacchetta

and Piermartini, 2011). Beshkar et al. (2011) propose a model that sheds light on one

important dimension of this variation. They consider a model where governments ha ve

private information about domestic political pressures, and examine how the levels of

tariff ceilings and the expected size of the binding overhang depend on countries’ market

power.Their main result is that,when a country has more market power,the optimal tariff

ceiling is lower and the expected binding overhang is smaller. This is a consequence of

the fact that when a country has more market power its tariffs exert stronger TOT exter-

nalities, so providing ﬂexibility through higher tariff bindings causes greater efﬁciency

I note here that Amador and Bagwell (2013) consider a slightly more general setting than the one described in the

text, where the importing government is allowed to “bur n” money.

352 Giovanni Maggi

loss. Beshkar et al. then present econometric evidence in support of this prediction, using

a dataset on applied and bound tariffs for WTO member countries.

It is useful at this juncture to recall from Section 2.2 that there is another possible

explanation for the use of tariff caps, which is not based on frictions in government-

to-government contracting, but rather on domestic-commitment issues, as pointed out

by Maggi and Rodriguez-Clare (2007). One interesting difference between these two

explanations of tariff ceilings—the one based on international contracting frictions and

the one based on domestic-commitment issues—is that the former predicts binding

overhang with positive probability in equilibrium, whereas the latter predicts no binding

overhang in equilibrium. For this reason, while domestic-commitment considerations

may be part of the explanation for tariff ceilings, they are probably not the whole story

behind them.

One of the challenges of this body of theor y is t o explain not just why the WTO

imposes rule X or why it imposes rule Y, but also why it imposes rules X and Y. This is

a non-trivial question, since different rules can interact in complex ways, and often i t is

not easy to rationalize the WTO’s joint use of disparate approaches to regulating trade

policies. Beshkar and Bond (2012) take a step in this direction by developing a model that

can explain why the GATT-WTO combines two different for ms of ﬂexibility on tar iffs,

namely tariff caps and escape clauses. More speciﬁcally, the model has a similar structure

as Amador and Bagwell (2013), except that a government can produce evidence (at a cost)

about the state of its domestic political pressures. Beshkar and Bond show that it may be

desirable to combine a tar iff cap with an escape clause that allows a government to raise

the tariff above the cap if it produces ev idence that doing so is politically efﬁcient. They

also show that these two forms of ﬂexibility are substitutes, in the sense that the optimal

tariff ceiling is lower—and tariff overhang may even disappear—in the presence of the

escape clause than in its absence.

3.1.3. “Shallow” vs “Deep” Integration

The GATT ag reement was largely based on a “shallow integration” approach, in the

sense that the agreement placed direct constraints on border measures (such as trade

taxes and quotas), while domestic policies were largely left to discretion, except for the

requirement that they not be used to erode the market-access levels previously nego-

Amador and Bagwell (2012) also presents an interesting result regarding binding overhang. They consider a var iant of

Amador and Bagwell (2013) where governments have pr ivate infor mation about the value of tar iff revenue,a setting that

is arguably relevant for developing countries. In a linear-quadratic speciﬁcation with a uniform type distribution, they

show that the optimal tariff cap and the probability of binding overhang are higher when there is greater uncer tainty

in the type distribution and when the upper bound of the support of the distribution is higher.

It is also relevant to note however that,if tariff ceilings were due solely to the presence of non-contractible contingencies,

we would expect that, for a given product in a given country, the applied tariff is sometimes at the bound level and

sometimes below it, but it is not clear from existing evidence that this is actually the case in reality.

International Trade Agreements 353

tiated by governments.

Bagwell and Staiger have argued in several papers that the

GATT’s “non-violation” clause (Art. XXIII.1b) can be interpr eted as imposing a kind of

“market-access-preservation” constraint on governments. The WTO, on the other hand,

has gone beyond a shallow integration appr oach, b y introducing direct restrictions on

some domestic policies, notably production subsidies, so in this sense it has moved closer

to a “deep integration” approach. What are the relative merits of these t wo approaches,

and why has the approach changed in going from GATT to WTO?

The benchmark paper on this topic is Bagwell and Staiger (2001), which focuses

on a standard two-good model with perfect competition and no uncertainty. I will

brieﬂy discuss this paper using the notation introduced in Section 2.1.2. The basic idea

can be illustrated by supposing that the Foreign go vernment is passive, so its objective

can be written simply as 

∗

), whereas the Home government can use a tariff τ

and a non-wedge policy ν (e.g. a labor standard), so that its objective can be written

as (ν,p, p

). Supposing that Home imports the non-numeraire good, let X

∗

) be

Foreign’s export supply for this good and let p

(τ, ν) be the equilibrium world-price

level. In its bare-bone form, a “shallow integration” agreement speciﬁes a reference tariff

level τ

and allows Home to choose its policies (including the tariff) subject to the

only constraint that the resulting trade volume be the same as that implied by τ

and

the noncooperative domestic policy level, sa y ν

. Such a market-access-preservation

constraint thus takes the form: X

∗

(τ, ν)) = X

∗

(τ

,ν

)).

It is immediate to show that, under perfect competition and in the absence of uncer-

tainty, a shallow-integration agreement is sufﬁcient to achieve global efﬁciency.The intu-

ition is straightforward: since there is a one-to-one link between trade volume and TOT,

a trade-volume-preservation constraint effectively prevents the Home g overnment from

manipulating TOT; and since TOT manipulation is the only cause of inefﬁciency in the

Nash equilibrium, this constraint is sufﬁcient to achieve efﬁciency.

This argument can

be extended to allow the Foreign government to be policy-active as well.

In addition, GATT imposed that certain domestic instruments such as consumption taxes and product standards must

satisfy the National Treatment r ule, as I discussed in Section 3.1.1.

Note that a simpler but equivalent contract would be one that speciﬁes a trade volume

∗

that Home is required to

deliver, so Home chooses (τ, ν) subject to the constraint X

∗

( p

(τ, ν)) =

∗

. Clearly this contract is equivalent to

the one mentioned in the text, because choosing the baseline tariff level τ

effectively amounts to choosing the target

trade level

∗

See also Bagwell and Staiger (2006),who extend the analysis to a setting where governments can use trade policies and

production subsidies. Another related paper is the one by Bajona and Ederington (2012), who argue that, if the TA

must be self-enforcing and domestic policies are not observable, the optimal TA m ay include not only market-access

constraints but also tariff bindings.

There is an important caveat to this argument: in GATT-WTO the negotiated tariff levels are not simply reference

levels, but tariff ceilings.InBagwell and Staiger’s (2001) model, if the contract includes also a tariff ceiling together

with the market-access constraint, the inability of a government to raise the tar iff above the ceiling may under some

conditions lead to inefﬁcient outcomes.This observation leads Bagwell and Staiger to argue that WTO r ules could be

made more efﬁcient while at the same time affording governments more sovereignty.

354 Giovanni Maggi

Bagwell and Staiger’s (2001) result represents an important benchmark, but a number

of subsequent papers have highlighted reasons why a shallow integration approach may

not be sufﬁcient to achieve globally efﬁcient outcomes. One reason may be the presence

of non-T OT externalities, such as the ones emphasized by the New Trade theories; a

second reason may be the presence of domestic-commitment motives for TAs; and a

third reason may be the presence of contracting frictions.

A setting with non-TOT externalities where shallow integration is not sufﬁcient to

achieve efﬁciency is the one considered by Antras and Staiger (2012a,b).What is responsi-

ble for the insufﬁciency of shallow integration in that setting is that prices are determined

by bilateral bargaining rather than market clearing. This is intuitive in light of the fact

that, as discussed in Section 2.3 above, the purpose of a TA in these models goes beyond

the correction of TOT externalities.

Another example is provided by DeRemer (2011),

who argues that shallow integration is not sufﬁcient to achieve efﬁciency in a monopo-

listic competition model with delocation externalities, when governments can use t rade

tax es and production subsidies. Moreover, a natural conjecture is that the same would be

true in a Cournot setting with proﬁt-shifting externalities, such as the one in Mrazova

(2011) or Bagwell and Staiger (2012a), if domestic policies were allowed in these models.

It should be clear from the discussion above that a shallow integration approach can

achieve efﬁciency only if the international policy externalities travel solely through the

TOT channel. It should not come as a surprise, therefore, that if the TA is motivated

by domestic-commitment motives then a deep integration approach will be required to

achieve efﬁciency. This point is made in the paper by Brou and Ruta (2009),which I

already mentioned in S ection 2.2. One important point that emerges clearly from all the

papers discussed above is that the motives for a TA crucially affect the optimal design of a TA.

As I mentioned earlier , a ﬁnal reason why a shallow integration approach may not be

enough to induce globally efﬁcient outcomes is the presence of contracting frictions. To

understand why, let us go back to the Bagwell-Staiger setting considered at the beginning

of this section, with perfect competition and a single policy-active country (Home).

Suppose that the state of the world is uncertain at the time theTA is negotiated,but Home

observes it before choosing its policies. Then, in order for a market-access-preservation

rule to induce a globally efﬁcient outcome, such rule would need to be state-contingent.

But i f some contingencies are not veriﬁable (or if specifying them i n the contract is

costly), then a shallow integration approach may not achieve full efﬁciency. This point is

made by Lee (2007, 2011) in the context of a model where governments choose trade

tax es and production taxes, and have private information about the level of a domestic

externality.

However it is important to note that the two questions,“Are the PO policies efﬁcient?” and “Is shallow integration

sufﬁcient to achieve efﬁciency?” do not necessarily yield the same answer. It is possible that PO policies are efﬁcient

but shallow integration does not achieve efﬁciency: in Antras and Staiger (2012a), for example, in the case of welfare-

maximizing governments PO policies are efﬁcient but shallow integration does not achieve efﬁciency.

International Trade Agreements 355

I now feel the need to take a step back and reﬂect on the way this design question

has been addressed in the literature, and on the way I think it should ideally be framed.

The typical approach in the literature (at least implicitly) has been to ask whether a

shallow-integration agreement can achieve global efﬁciency: if the answer is yes, then a

shallow-integration agreement is considered preferable to a deep-integration agreement,

based on an (implicit or explicit) assumption that“more sovereignty is always better;” and

if the answer is no, the conclusion is that a deep-integration agreement is preferable. I

ﬁnd this conceptual approach useful but not entirely satisfactory, for the r easons I explain

next.

The argument that it is preferable,other things equal,to give countries sovereignty over

domestic policies is presumab ly based on a notion that relinquishing control over domestic

policies entails some cost to a government. This idea has some intuitive appeal, but it is

not obvious ho w it can be given a rational-choice justiﬁcation. Why would governments

not feel similarly about relinquishing control over border measures? Perhaps one way

to rationalize the “sovereignty” argument is that domestic policies are needed to address

domestic economic or political needs that vary over time, and taking away sovereignty

over domestic policies remo ves a government’s ﬂexibility in responding to these v arying

needs. But this implicitly assumes that the TA constrains domestic policies in a rigid way;

in the absence of contracting frictions, aTA would constrain these policies in a contingent

way, so there would be no shortcoming from a deep-integration approach. Thus one can

rationalize the idea t hat giv ing up sovereignty is costly, but at the core, this boils do wn to

an argument that contracting frictions cause rigidity in TAs. For this reason, I think that

introducing uncertainty and contracting frictions explicitly in our conceptual framework

is important to examine the tradeoff between shallow and deep integ ration.

The discussion above in turn suggests an alternative,and I think preferable,approach to

the comparison between shallow and deep integration. The key question in my opinion

should be:what are the transaction costs associated with each of these two approaches,and

which one is more efﬁcient when transaction costs are taken into account? The answ er

to this question is far from obvious: a deep-integration agreement requires specifying

all border and domestic policies, and this is likely to involve large transaction costs; a

shallow-integration agreement, on the other hand, may save on the costs of contracting

over domestic policies, but it will achieve less efﬁcient outcomes (if the environment is

uncertain); and if specifying contingencies in the TA is costly, the tradeoff becomes even

more complicated. If one takes this explicit transaction costs perspective, then the results

of the existing literature appear as a valuable step, but only a ﬁrst step.

3.1.4. Reciprocity and MFN

The principles of reciprocity and MFN are two major pillars of the GATT-WTO. The

role of these principles has been examined by Bagwell and Staiger in a number of writings,

356 Giovanni Maggi

and most notably in the context of the perfect-competition setting of Bagwell and Staiger

(1999a). I will start by focusing on the principle of reciprocity in the context of a two-

country model (where MFN of course pla ys no role).

Bagwell and Staiger deﬁne a reciprocal change in trade policies as one that leaves

world prices unchanged.

More speciﬁcally, they distinguish between two notions of

reciprocity. The ﬁrst one is a principle guiding tariff negotiations starting from the Nash

equilibrium: Bagwell and Staiger show that, if tariff negotiations satisfy reciprocity (that

is, if they leave world prices unchanged relative to the Nash equilibrium), they will lead

to a Pareto-improvement over the Nash equilibrium, although in general they will not

lead all the way to the Pareto-efﬁcient frontier.

An important note of interpretation

is the following: in the GATT-WTO the principle of reciprocity as it applies to tariff

negotiations is not a strict rule, but just an informal principle, so Bagwell and Staiger’s

(1999a) analysis of reciprocity-guided negotiations is best interpreted in normative terms,

as highlighting how the negotiation outcome would be affected if reciprocity were strictly

imposed as a rule.

The second notion of reciprocity considered by Bagwell and Staiger applies to tar-

iff renegotiations. Unlike the informal reciprocity principle that applies to negotiations,

GATT-WTO does impose a formal rule of reciprocity for tariff renegotiations,speciﬁcally

in Article XXVIII of GATT. Bagwell and Staiger formalize this rule through a two-stage

negotiation game, as follows: (i) in the ﬁrst stage, governments negotiate a pair of tariffs,

with the disagreement point given by the Nash equi librium tariffs;(ii) in the second stage,

governments can renegotiate the tariffs, and if the renegotiation fails, a government can

unilaterally change its tariff from the level that was agreed upon in the ﬁrst stage, but at

the condition that the trading partner get compensated through a reciprocal tariff change.

Before proceeding, I make an observation that will be useful later on. Notice the

nature of the disagreement point for the process of renegotiation under reciprocity: if the

renegotiation fails, a gov ernment can choose to unilaterally “breach” the contract and

compensate the trading partner, with the compensation taking the form of a reciprocal

tariff increase by the trading partner. This observation leads me to note that there is a

simpler way to intepret such recipr ocity rule, which I ﬁnd more illuminating. Rather

than imposing a constraint on renegotiation, this rule can equivalently be modeled as

changi ng the nature of the contract itself, from one that speciﬁes tariff commitments

without allowing for “breach” (in law and economics jargon, a “property” contract), to

one that allows a government to breach the contract in exchange for a certain amount of

“damages” (in law and economics jargon, a “liability” contract). Notice that, if reciprocity

is modeled in this way, simply as a liability rule incorporated in the initial contract, then

there is no scope for renegotiation in Bagwell and Staiger’s setting,and we can simply think

As Bagwell and Staiger show, an equivalent deﬁnition is that changes in trade policies are reciprocal if they bring about

equal-value changes in each country’s volumes of impor ts and exports when valued at the initial world pr ices.

The only case in which negotiations according to reciprocity lead to Pareto efﬁciency is the one where countries are

perfectly symmetric.

International Trade Agreements 357

of governments as negotiating a contract within this particular class (liability contracts

with breach damages g iven by reciprocal tariff changes), ignoring renegotiation.

Bagwell and Staiger’s main results regarding the rule of reciprocity are two. The ﬁrst

one is that the only efﬁcient tariff pair that can be implemented under reciprocity (or,

adopting my interpretation above,when the contract is restricted to be a liability contract

with breach damages given by reciprocal tariff changes) is given by the politically efﬁcient

(PO) tariffs.This result is a consequence of the fact that,when viewed in tariff space, each

gov ernment’s iso-payoff curv e is tangent to the iso-world-price curve at the PO point,

thus starting from this point there is no unilateral incentive for a government to move

along the iso-world-price curv e. This in turn implies that the reciprocity-constrained

Pareto frontier lies inside the unconstrained Pareto frontier, except at the PO point. The

second result is that, when go vernments bargain over the recipr ocity-constrained Pareto

fr ontier, the outcome is generically inefﬁcient (it is not the PO point), but it tends to be

closer to the PO point as compared with the unrestricted bargaining scenario.

Bagwell and Staiger interpret these results as suggesting that the reciprocity rule

induces a rebalancing of power across countries, since it moves the negotiation outcome

toward a point that is not affected by bargaining powers (the PO point). This conclusion

resonates with the GATT’s emphasis on “rules” vs “power,” but as remarked above, the

rebalancing of power induced by the reciprocity r ule in g eneral entails an “efﬁciency

penalty.” Bagwell and Staiger then argue informally that reciprocity may provide efﬁ-

ciency gains if it encourages weaker countries to participate in GATT. This idea is based

on McLaren’s (1997) model: recall from Section 2.2 that in this model a small country

trading with a large country may prefer to stay out of a TA to avoid being “held up,”

and for this reason the large country would like to commit not to exploit its bargaining

power, in order to encourage the small country to participate in the TA. Thus the broad

idea is that, if one thinks of GATT as initially including a set of large/powerful countries,

but there is also a set of smaller/weaker countries that may consider accessing GATT at

a later stage, the initial members may prefer to commit not to exploit their bargaining

power in future negotiations, in ord er to encourage other countries to seek participation.

I will make two further comments about Bagwell and Staiger’s analysis of the reci-

procity rule.The ﬁrst one is a“devil’s advocate” comment. Bagwell and Staiger’s interpre-

tation of the results outlined above is that reciprocity“ works well”whenTOT externalities

are the only motive for aTA. But one might argue that the model tells the opposite story:

reciprocity causes the negotiation outcome to be inside the P areto frontier, so the world

would be more efﬁcient without r eciprocity. A more cautious interpretation of Bagwell

and Staiger’s analysis w ould be that it pr ovides a positive (as opposed to normative)

I note that this way of thinking about reciprocity, as a r ule specifying breach remedies rather than as a constraint on

renegotiation, is close to the way Ossa (2011) formalizes the reciprocity rule in his model.

I say “tends to be closer to the PO point” because this is true only under some conditions, as explained in footnote 25

of Bagwell and Staiger (1999a).

358 Giovanni Maggi

evaluation of the reciprocity rule, highlighting that this rule has a distributional effect

(which might be desirable in a richer model that includes participation considerations)

and an efﬁciency cost.

The second comment is that, aside from the participation argument outlined above,

Bagwell and Staiger’s analysis is suggestive of another potential efﬁciency beneﬁt of reci-

procity. Since, as I argued above, imposing the reciprocity rule is akin to structuring the

TA as a (speciﬁc type of) “ liability” contract, this might be appealing as a way to inject

ﬂexibility in the TA when countries are subject to shocks and the TA cannot be fully

contingent. But this is just suggestive, since there is no uncertainty or contract incom-

pleteness in the model.

This brings me back to one of my overarching points, namely

that a better understanding of TA design requir es bringing transaction costs explicitly

into the picture.

After their analysis of reciprocity within a tw o-country setting, Bagwell and Staiger

(1999a) turn to an examination of the implications of the MFN r ule within a multi-

country setting. As I mentioned in Section 2.1.2, in a multilateral world where govern-

ments can set discr iminatory tariffs, there is a whole vector of bilateral offshore prices,

which generates a complicated pattern of international policy externalities,but the MFN

rule has the effect of channeling all these externalities into a single world-price external-

ity. Building on this observation, Bagw ell and Staiger show that the MFN rule , if used in

tandem with the reciprocity rule, guides countries toward the PO tariffs (which, recall,

are efﬁcient under MFN). The k ey point here is that the combination of MFN and

reciprocity has similar effects in a multi-country world as the reciprocity rule does in a

two-country world.

So far I have focused on the implications of reciprocity and MFN within a perfectly

competitive environment. It has been argued, however, t hat reciprocity and MFN can be

rationalized also within an imperfectly competitive environment, where the motives for

a TA go beyond the correction of TOT externalities. In particular, Ossa (2011) argues

that, in a monopolistic-competition setting, recipr ocity and MFN can be interpreted as

helping countries internalize the production-delocation externalities generated by trade

policies.

3.1.5. “Property” vs “Liability” Rules

In the previous section I mentioned the distinction between a “property” contract (one

that does not provide for the possibility of breach) and a “liability” contract (one that

As I will discuss in the next section, papers by Maggi and Staiger (2012) and Beshkar (2010a,b) have examined this

idea more formally in models with non-veriﬁable or pr ivately observed shocks.

It should be noted however that the implications of reciprocity and MFN in Ossa’s model are somewhat different than in

Bagwell and Staiger’s model. Ossa shows that reciprocity can help countries achieve an efﬁcient outcome if reciprocity

is applied to multilateral trade negotiations, and then argues that MFN can serve to multilateralize trade negotiations;

Bagwell and Staiger, on the other hand, show that reciprocity and MFN ensure that any bilateral negotiation will lead

to an efﬁcient outcome.

International Trade Agreements 359

gives a government the option to breach-and-compensate). In real-world TAs, there is

considerable variation between liability-type rules and property-type rules, both across

issues and over time. For example, Pauwelyn (2008) argues that property r ules are the

“default” approach in theWTO and NAFTA, but for certain issues such as tariff bindings

and production subsidies, a liability-rule approach has been taken.

Moreover, most legal

scholars take the view that the early GATT operated as a system of liability rules, while

in more recent times the GATT/WTO has evolved toward a proper ty-rule system (see

for example Jackson, 1997).

As mentioned above, a liability approach is appealing in the presence of uncertainty

because it can inject ﬂexibility in the TA without the need to specify contingencies

explicitly in the contract: intuitively, a liability rule can induce a country to internalize

the externalities that its trade policy exerts on its trading partners. However, in the

international trade arena, the liability approach has an important shortcoming, namely

that government-to-government compensation is inefﬁcient: cash transfers are typically

not available, and compensation typically takes the form of “self-help” through tariff

retaliation. For this reason,a liability rule can generate deadweight loss, and this gives r i se

to a non-trivial tradeoff between property and liability rules.

The t radeoff between proper ty and liability r ules becomes even more subtle if gov-

ernments can renegotiate theTA, because in this case a property rule does not necessarily

imply inﬂexible policy outcomes (since it can be renegotiated ex-post),and it can give rise

to inefﬁcient compensation in case of renegotiation. Taking renegotiation into account

seems important also in light of its empirical importance in the context of GATT-WTO,

where renegotiations have taken place in many instances over the years.

The choice between property and liability rules in the presence of renegotiation is

analyzed by Maggi and Staiger (2012). In this model, governments negotiate o ver a binary

trade policy (free trade or protection) under uncertainty about the future joint beneﬁts of

protection (which can be positive or negative, due to political economy considerations),

and can renegotiate theTA after the uncertainty is resolved. A key feature of the model,in

line with the discussion above, is that government-to-government compensation entails

a deadweight cost.

This includes for example the provisions for escape from negotiated tariff bindings (GATT Articles XIX and XXVIII,

respectively), the rules on “actionable” production subsidies in WTO, and the provisions to protect investors against

expropriation in NAFTA (and in many other bilateral investment treaties).

Here I note that, while the ter minology of proper ty and liability rules is more common in the law-and-economics liter-

ature,the choice between these two contractual for ms is an important topic in the economic literature on optimal con-

tract design,where a liability contract is often referred to as an“option contract,”and a proper ty contract is often referred

to as a “noncontingent contract,” or simply a “property-right” contract (see for example Segal and Whinston, 2002).

The inefﬁciency of government-to-government compensation is not the only possible shortcoming of a liability

approach. Another limitation—which I abstract from here—is that the damage inﬂicted by trade protection on trading

partners is typically non-veriﬁable, and this makes it impossible to induce a government to perfectly internalize trade

policy externalities.

360 Giovanni Maggi

Maggi and Staiger ﬁnd that a property rule is optimal if uncertainty about the joint

beneﬁts of protection is small enough, while a liability rule is optimal when this uncer-

tainty is large. If one interprets uncertainty about the joint beneﬁts of protection as due

to political-economy shocks, then this r esult suggests that liability rules should be more

prevalent in issue areas that are mor e politically sensitive.This prediction seems consistent

with the observation that the WTO has taken a liability approach in the areas of import

tariffs and production subsidies (which are arguably very sensitive to political-economy

shocks) and a property-rule approach in other areas. A further r esult is that, if a liability

rule is optimal,t he optimal level of damages falls short of fully compensating the exporter,

contrary to the well-known “efﬁcient breach” argument. This result is shown also in a

related model by Beshkar (2010a), though in a setting without renegotiation.

The model yields predictions also about the occurrence of renegotiation in equilib-

rium, and how it correlates with the optimal contract form. I report two ﬁndings here.

The ﬁrst is that renegotiation, when it occurs, always entails the exporter agreeing to

compensate the importer in exchange for trade liberalization.

The second is that an

optimal property rule is never renegotiated in equilibrium. This ﬁnding may seem coun-

terintuitive, since a pr operty rule is inherently rigid and renegotiation should be useful

to mitigate such rigidity, but it tur ns out that renegotiation can improve t he performance

of a property rule only if such rule is suboptimally adopted.

3.1.6. Dispute Settlement Procedures and Contract Incompleteness

As I discussed at the beginning of Section 3, the role of the WTO’s DSB seems to go well

beyond a pure enforcement role, at least judging from casual observations of real-world

trade disputes, where the DSB seems to often play an “activist” role by interpreting vague

clauses and ﬁlling gaps of the agreement. The potential role of the DSB in completing

an incomplete agreement is the focus of a paper by Maggi and Staiger (2011).

Beshkar (2010b) builds on Beshkar (2010a) by allowing theWTO court (DSB) to observe a noisy signal of the state of

the world,and shows that the performance of the contract can be improved by making the remedies for breach (that is,

the injured country’s retaliatory tariff) contingent on the DSB’s signal, which in turn can be accomplished by allowing

the injured country to ﬁle a complaint that triggers DSB intervention. See also Beshkar (2011), which extends the

previous models by allowing for a limited form of renegotiation between governments.

This asymmetry in the predicted direction of renegotiation is a consequence of two features: that the contractual

obligation is to liberalize trade, and that it is never optimal to induce renegotiation in states of the world where the

threat point is the contractual obligation itself.

In the model just discussed, the cour t/DSB does not play an active role in equilibr ium, so the model is silent about

what determines the occurrence of trade disputes and their outcomes. In a more recent working paper, Maggi and

Staiger (2013) consider a richer model in which trade disputes can occur in equilibrium and can result in a variety

of outcomes; in particular, governments may settle early or trigger a DSB ruling, and in the latter case, they may

implement the DSB ruling or renegotiate after the ruling. Two new features of the model are responsible for this

rich set of possibilities: ﬁrst, the DSB can observe a noisy signal of the joint beneﬁts of protection (interpreted as the

outcome of a DSB investigation), so governments are uncertain about the direction of the DSB ruling; and second,

governments have a further opportunity to renegotiate the TA after the DSB issues a ruling.

An early attempt to examine the potential role of the DSB in completing an incomplete contract can be found in

Battigalli and Maggi (2003), who focus on agreements on product standards. In that model, the TA cannot specify

International Trade Agreements 361

Magg i and Staiger consider three possible activ ist roles for the DSB:inter preting vaguely

stated obligations; ﬁlling gaps in the agreement; and modifying rigid obligations; and for

each of these roles, the DSB may or may not have authority to set precedent for future

rulings. Governments design the institution—that is the combination of contract and

DSB mandate—under uncertainty about the future state of the world. The relevant con-

tingencies are assumed to be too costly to describe in a crisp way, so the contract is

necessarily incomplete. The model allows for three possible forms of contractual i ncom-

pleteness: the ﬁrst two, rigidity and discretion, are familiar (see Section 3.1.1); the third one,

vagueness, is novel.

There is a natural pairing between these three forms of contractual

incompleteness and the three possible activist DSB roles described above: t he DSB can

interpret a vague contract; it can ﬁll gaps if the contract leaves discretion; and it can grant

exceptions if the contract is rigid. Or, the DSB can play a non-activist role and simply

enforce clearly stated obligations. Furthermore, for each of the activist roles,the DSB may

or may not have precedent-setting author ity. A key feature of the model is that invoking

the DSB entails two kinds of inefﬁciency: one is that the governments incur litigation

costs, and the other is that DSB rulings are imperfectly accurate.

Maggi and Staiger show that the optimal institutional form depends critically on the

degree of DSB accuracy. If the DSB is sufﬁciently accurate, it is optimal to leave gaps

in the contract and g ive the DSB a mandate to ﬁll those gaps. If the DSB is sufﬁciently

inaccurate, it is optimal to write a vague or rigid contract and give a non-activist mandate

to the DSB. And if the level of DSB accuracy is intermediate, it is optimal to write a

v ague contract and allow the DSB to interpr et the contract. The “modiﬁcation” role of

the DSB turns out to be always suboptimal.

The model delivers good news and bad news r egarding the potential of the DSB to

enhance efﬁciency. The good news is that, if the DSB is sufﬁciently accurate, the ﬁrst-

best outcome can be achieved, in spite of the incompleteness of the contract, the costs

of using the DSB, and the imperfection in DSB rulings. The r eason is that the threat of

inv o king the DSB and the expectation of a not-too-inaccurate DSB ruling are enough

to discourage opportunistic beha vior by gov ernments. This suggests that an activist DSB

can generate substantial efﬁciency gains even if its information is not perfect. The bad

news is that the outcome tends to be efﬁcient only when the DSB is not invoked in

equilibrium, and disputes are more frequent when the DSB is less accurate . Thus, the

standards for products that will appear in the future. If a new product appears, governments can negotiate ex-post

over the standards for that product, but international transfers are not available ex-post, and as a consequence the

ex-post negotiation outcome is ex-ante inefﬁcient.The DSB acts as an “arbitrator” that can be invoked if the ex-post

negotiation fails. Battigalli and Maggi show that the DSB can improve ex-ante efﬁciency by specifying ex-post—if

invoked—the standards that should apply to the new product, provided the DSB has sufﬁciently precise information.

Vagueness is modeled as a language whose meaning is only partially deﬁned. As an example, consider a contract stating

that trade protection is allowed only if “there is substantial injury to the domestic industry.”The idea is that there are

states of the world where the latter sentence is clearly true (e.g. if there is an import surge, th e domestic industry shuts

down, and the majority of workers in the industry are unemployed), others where it is clearly false (if none of the

above events has occurred), and there are “gray area” states where it is not deﬁned whether the sentence is true or false.

362 Giovanni Maggi

efﬁciency-enhancing effect of the DSB is associated with its off-equilibrium impacts. If

the DSB is invoked in equilibrium, it is always because one of the governments is being

opportunistic: either the importer is protecting when it should not, hoping that the DSB

ruling will get it wrong; or the exporter is tr ying to force free trade by ﬁling a dispute

when it should not. A corollary of these observations is that the frequency of DSB use is

not a good indicator of the performance of the institution.

The model also has interesting implications regarding the “bias” in observed DSB

rulings. Because of selection effects in the ﬁling of disputes, DSB rulings tend to have

a pro-trade bias if litigation costs fall more on the exporter government than on the

importer government. In reality, it is arguably the case that litigation costs fall more on

the exporter government, because the burden of proof falls on the complainant, and the

complainant is typically the exporter government. Thus the model suggests a possible

explanation for the fact that, both under the GATT and the WTO, complainants have

mostly w on their cases.

Finally, Maggi and Staiger extend the basic model to a two-per i od setting in order

to examine whether it is desirable to give the DSB precedent-setting authority. This is

an important issue of institutional design, and one that is receiving increasing attention

from legal scholars.

Introducing precedent in this setting is shown to have two opposite

effects on efﬁciency. The beneﬁcial effect of precedent is that it reduces the probability of

dispute occurrence tomor r ow, by clarifying the obligations that will apply should the same

state of the world occur again;this leads to a sav ings in litigation costs.The harmful effect

of precedent is that it increases the probability of dispute occur rence today; this in turn

implies more waste in litigation costs and a less efﬁcient policy outcome (because the

DSB ruling is subject to error). Maggi and Staiger show that,as a net result of these effects,

precedent is more likely to be beneﬁcial when the accuracy of DSB rulings is lower and

when governments care less about the future, or are less likely to interact repeatedly.

3.2. Imperfect Enforcement and Trade Agreements

In this section I will focus on how the presence of enforcement fri ctions can shape the

design of rules and procedures in aTA. More speciﬁcally,I will discuss how the presence of

self-enforcement constraints affects the design of substantive policy rules, of enforcement

rules (i.e. rules that regulate punishments), and of dispute settlement procedures.

In the literature, the dominant approach for modeling a self-enforcing TA has been

to consider a game where governments choose trade policies repeatedly and focus on

(constrained-)Pareto-efﬁcient equilibria of this game. The implicit assumption in this

approach is that governments bargain efﬁciently over the set of equilibria of the repeated

Even at the positive level, the extent to which the WTO-DSB currently operates on a precedent system seems subject

to debate. According to Jackson (2006, p. 177),“there is quite a powerful precedent effect in the jurisprudence of the

WTO, but ... it is not so powerful as to require panels or the Appellate Body considering new cases to follow prior

cases.” Jackson concludes that “the ‘ﬂavor’ of the precedent effect in the WTO is still somewhat ﬂuid.”

International Trade Agreements 363

game. This approach has become fairly standar d in the literature and is explained in

Staiger’s (1995a) chapter, so I will take it for granted and simply provide an informal

overview of the contributions in this area after 1995.

3.2.1. Policy Rules

The presence of self-enforcement constraints can have deep implications for the design

of policy rules. This point was ﬁrst made in a pathbreaking paper by Bagwell and Staiger

(1990), which showed that, in the presence of (publicly observed) i.i.d. shocks to trade

volume, the need to accommodate self-enforcement constraints makes it desirable to

include an escape clause in theTA.The basic idea is that, in periods of high import volume,

a country has a stronger incentive to deviate from its trade policy commitments, so in

such periods it may be a good idea to allow a country to escape from its commitments

in order to keep cooperation sustainable.

More recently, Bagwell and Staiger (2003) have extended their previous model by

allowing for persistent shocks to trade volume as well as an acyclic component of the

shock. In this setting, Bagwell and Staiger show that trade protection tends to be coun-

tercyclical. The reason is t hat a boom phase tends to be characterized by f ast growth in

trade volume, which helps countri es sustain lower tariffs than in a recession phase, while

acyclic increases in trade volume g ive rise to increases in tariffs, for a similar reason as in

Bagwell and Staiger (1990).

Another group of papers has highlighted that the presence of self-enforcement con-

straints can help explain why TAs typically take a gradual approach to trade liberalization.

The ﬁrst two papers in this group are Staiger (1995b) and Devereux (1997),theformer

focusing on the implications of sector-speciﬁc skills that depreciate when not in use,

and the latter on the implications of technological learning-by-doing. Furusawa and Lai

(1999) and Chisik (2003) propose two further mechanisms that can generate gradual trade

liberalization as part of an optimal self-enforcing TA: Furusawa and Lai focus on the role

of adjustment costs in worker reallocation across sectors, and Chisik focuses on the role

of irreversible investments in country-speciﬁc export capacity. The common theme in all

of these papers is the non-stationary nature of the repeated game between governments,

Before surveying the relevant literature, I mention brieﬂy four papers that do not focus on issues of r u le design, but on

a more classic question: what conditions facilitate the self-enforcement of TAs? Furusawa (1999) examines how the

sustainability of cooperation is affected by the governments’relative patience and the lag between detection of a violation

and retaliation; Park (2000) focuses on a trade agreement between a small country and a large countr y, and examines

how the set of sustainable payoffs is affected by the availability of direct transfers and the presence of sunk investments;

Conconi and Sahuguet (2009) explore the impact of policymakers’ horizons and alternative electoral regimes on the

sustainability of international cooperation; and Conconi and Perroni (2009) examine whether and how the ability of

gover nments to commit to policies in the domestic arena affects the sustainability of international cooperation.

See also Milner and Rosendorff (2001) for a related model of self-enforcing ag reements where the presence of

uncertainty makes it optimal to introduce escape clauses.

The question of whether or not trade barriers are countercyclical has been the subject of interesting empirical work

recently. See in particular Bown and Crowley (2013b) and Rose (2013).

364 Giovanni Maggi

whereby an initial reduction in tariffs leads to a change in some economy-wide state

variable (such as the level of physical or human capital allocated t o the export sector),

which in turn relaxes the self-enforcement constraint and allows governments to sustain

further tariff reductions.

One might think that gradual trade liberalization can only be explained in a non-

stationary trading environment, but Bond and Park (2004) mak e the surprising point that

gradualism can arise even in a stationary economic environment. Speciﬁcally, they show

that the optimal self-enforcing TA may entail gradual tariff reductions if countries are

asymmetric. In this case, it is possible that in the initial phase of the agreement the self-

enforcement constraint of only one country is binding, and given this initial asymmetry,

the most efﬁcient way to provide incentives to such country is to “backload” its payoff,

that is, to promise this country a rising payoff over time.

3.2.2. Enforcement Rules

When countries negotiate a TA, they negotiate not only substantive policy rules, but also

rules that regulate punishments for violations of the agreement. I will refer to these as

“enforcement rules.”

A key consideration for the design of enforcement rules is that they must them-

selves be self-enforcing, or in other words they must be credible: in the language of

repeated-game theory, this means that the punishment strategy must be an equilibrium

of the continuation game after the initial deviation. In this perspective, the question

“What are the optimal enforcement rules?” can be f ormally phrased as “What is the

optimal equilibrium punishment strategy?”It is important to keep in mind that in a

repeated game there is a vast multiplicity of equilibrium punishment strategies, just as

there is a vast multiplicity of overall equilibria. The implicit assumption in this modeling

approach is that, when governments negotiate a TA, they bargain efﬁciently over policy

rules and enforcement rules, subject to the constraint that all rules be part of an overall

equilibrium of the repeated game.

Maggi (1999) examines the optimal design of enforcement rules in the context of a

multilateral trading system. A key question in this context is whether and to what extent

punishments should be multilateral rather than bilateral.To be concrete, the question is: if

country A cheats on country B, should country C be involved in the punishment? Con-

sider ﬁrst a benchmark scenario where bilateral trading relationships are symmetric and

separable (in the sense that changing a bilateral trade barrier does not affect third coun-

tries):in such scenario,Maggi shows that there are no gains from multilateral punishments.

The intuition is that, while making punishments multilateral incr eases the future loss from

Another paper that highlights implications of self-enforcement constraints for the design of policy rules is Mrazova

(2011).This paper (which I already mentioned in Section 2.3.1) argues that the need to make theTA self-enforcing, in

conjunction with the presence of proﬁt-shifting externalities and costs of administer ing policy instruments, can help

explain the WTO ban on export subsidies.

International Trade Agreements 365

a deviation, it also increases the one-time gain from deviating, because if a country is to

deviate it will do so against all trading partners; when bilateral relationships are symmetric

and separable these two effects cancel each other out. Next consider a scenario charac-

terized by bilateral imbalances of power, in the sense that different governments stand to

lose different amounts from a trade war, with the more “powerful” governments standing

to lose less. In this scenario, multilateral punishments are desirable because they allow

for an exchange of enforcement po wer that bilateral punishments cannot achieve: more

speciﬁcally, if punishments are multilateral, each country can offer third-party enforce-

ment in bilateral relationships where it is “strong” in exchange for receiving third-party

enforcement from other countries in bilateral relationships where it is “weak.”

The next point made by Maggi (1999) is that, even though some third-country pun-

ishment i s in general desirable, there may be no need to make it very severe. More

speciﬁcally, increasing the severity of third-country punishments beyond a certain point

does not enhance cooperation, so there is no need to make these punishments “maxi-

mal.” Furthermore, the threat of third-country punishments is necessary only for certain

violations, namely those by stronger countries against weaker countries, which are hard

to deter with bilateral sanctions alone . The intuition is based on the result I mentioned

earlier: if there are no bilateral asymmetries, there is no need for any thir d-country pun-

ishments, so it is intuitive that if there is only a limited degree of bilateral asymmetries, a

limited amount of third-country punishment will be sufﬁcient.These results seem broadly

consistent with the fact that in the GATT-WTO the role of third-country punishments

is more subtle and selective than the role of bilateral punishments.

Also Bagwell et al. (2007) focus on the design of enforcement rules in a multilateral

world. In particular,this paper examines the desirability of “tradeable”retaliation rights,an

idea proposed a few years ago by Mexico in theWT O.The basic idea is that small countries

have a limited ability to retaliate against large countries, so allowing small countries to

sell their rights of retaliation to third countries might improve the performance of the

Gains from multilateral punishments can arise also from non-separabilities across bilateral relationships, which arise if

there are trade-diversion effects of bilateral trade barriers. In this case, the beneﬁts from multilateralizing punishments

arise fr om the aggregation of enforcement power across trading relationships (see Maggi, 1994). A simple intuition

for this effect can be gained by thinking about the effects of trade embargoes: a multilateral embargo inﬂicts a

proportionally more severe punishment than a bilateral embargo, since the latter is partially neutralized by substitution

across bilateral trade ﬂows.

There is a legitimate question as to whether third-country punishments play any role in the GATT-WTO . Maggi

(1999) argues that they do, although in subtle ways. One form of third-country punishment for example may be the

withdrawal of some “goodwill” by third countries toward the defecting country, resulting for example in a reluctance

to enter new agreements with that country. Also, one should not forget that, as Thomas Schelling made clear, the

effectiveness of an army sometimes must be judged by how little it is u sed: while it is tr ue that full-blown multilateral

retaliation has never been observed in the GATT-WTO, it is also true that there have been no cases of blatant and

repeated violations of key rules, even by strong countries against vulnerable trading par tners. It is reasonable to think

that strong countries may have been deterred from abusing weaker par tners not by the threat of bilateral retaliation,but

by the implicit threat that the whole trading system may unravel as a consequence, that is, by the threat of a multilateral

breakdown of cooperation.

366 Giovanni Maggi

system. Bagwell et al. consider a scenario where a violation of the agreement has already

occurred and the injured country has been granted the right to retaliate,and examine two

mechanisms for selling this right: a “basic” auction, in which the injured country is not

allowed to bid to retire the right, and an “extended” auction, where the injured country is

allowed to bid as well. Bagwell et al. ﬁnd that the basic auction may “fail”,in the sense that

no bids are made despite positive valuation by bidders, while the “extended” auction can

never fail, and in such auction the right of retaliation is always retired. The two auction

formats are then evaluated from a normative standpoint, and the ranking between them

is found to depend critically on the speciﬁc normative criterion that one adopts.

Another interesting question of enforcement-rule design arises when countries seek

to cooperate over multiple policy areas. In this case , one may ask whether there should

be issue linkage in the enforcement of the TA, that is, whether violations in one policy

area should be met with retaliation in other areas. Tw o papers that addr ess this question

are Ederington (2002) and Limão (2005).

Ederington (2002) considers a repeated-game model where each government can

choose a tariff and a domestic policy. In this model markets are competitive, but in each

country there is a localized externality which can be corrected by using the domestic

policy. First, Ederington shows that domestic policies are always set at their efﬁcient

(Pigouvian) levels in the optimal self-enforcing TA, while tariffs are set above their efﬁ-

cient levels to accommodate self-enforcement constraints. The intuition is related to the

targeting principle: since the only international externality is a TOT externality, the only

reason countries are tempted to deviate from the TA is to manipulate TOT, so raising tar-

iffs is the most efﬁcient way to neutralize this incentive to defect.

The second, related

result is that issue linkage need not enhance cooperation: in particular, if the punish-

ment strategy takes the form of a permanent reversion to the one-shot Nash equilibrium

(“grim-trigger” punishment), then the beneﬁts from issue linkage are nil.

Limão (2005) considers a setting characterized not only by T OT externalities but

also by cross-border pollution externalities. Governments choose tariffs and production

taxes in repeated-game fashion. In this setting, Limão shows that issue linkage always

allows governments to achieve a higher joint welfare relative to a non-linked agreement.

This in itself may not be surprising, but a subtle question concerns ho w linkage affects

the level of cooperation on a policy-by-policy basis. The key ﬁnding is that, i f policies

are independent in the governments’ objective functions, linkage promotes cooperation

Bagwell et al. (2007) are agnostic about the cr iterion according to which the mechanisms should be evaluated, but I

will not be agnostic here. In my view, the natural evaluation criterion is given by the efﬁciency of the equilibrium

of the repeated game that the enforcement mechanism allows to achieve (see my discussion at the beginning of the

section). The analysis of Bagwell et al. (2007) is silent about this criterion, simply because they do not model the

repeated-gam e explicitly, but take it as a given that a violation has already occur red and take the amount of permissible

retaliation as exogenous.

This result was ﬁrst derived in Eder ington (2001), and then extended by Ederington (2002) to allow for both linked

and non-linked agreements, as well as for different types of punishment strategies.

International Trade Agreements 367

in one policy area at the expense of the other, because linkage effects a reallocation of

enforcement power across issues; but if policies are strategic complements, t hen linkage

can lead to more cooperation in both policy areas.

My next theme of discussion is the role of retaliation in the presence of asymmetric

infor mation. When governments have private information, for example about domestic

political shocks, retaliation can play two distinct roles. First, retaliation can be a “punitive”

tool that serves to deter blatant violations of the agreement, along the lines discussed

earlier in this section. This type of retaliation is meant to be an off-equilibrium threat, so

it need not be observed in equilibrium. Second, retaliation can be used for “screening”

(or “truthtelling”) purposes, in the sense of inducing governments to increase protec-

tion only in states of t he world where it is (politically) efﬁcient to do so. Intuitively,

retaliation imposes a cost on a government that increases protection, so if the severity

of retaliation is appropr i ately calibrated, it may induce a government to increase pro-

tection when—and only when—the political gains from protection are high. Unlike

“punitive” retaliation, this latter type of retaliation is meant to occur in equilibrium

under some contingencies. Some scholars, for example Schwartz and Sykes (2002),have

argued that the reciprocity rule in Article XXVIII of GATT—which provides for the

“withdra wal of substantially equivalent concessions” in response to a breach of negoti-

ated tariff bindings—can be interpreted as serving a screening function along the lines I

just described. What makes this interpretation appealing is that the recipr ocity rule pro-

vides only for a limited amount of retaliation, which seems consistent with the screening

function, whereas the “punitive” function calls for more severe retaliation threats. I also

note that the screening role of retaliation is relevant even if the agreement is perfectly

enforceable, or if governments are so patient that self-enforcement constraints do not

bind. Indeed, some of the papers that consider this role of retaliation, such as Beshkar

(2010a,b) or (in a more reduced-form setting) Magg i and Staiger (2012, 2013), assume

perfectly enforceable TAs.

In the context of a repeated tariff game with private information, the distinction

between the punitive role and the screening role of retaliation can be understood in

the following way. Suppose the importing government chooses a tariff and pri vately

observes the value of a political-economy parameter. The agreement can be thought of

Here I will also mention a paper by Limão and Saggi (2008) that examines whether it may be desirable to use monetary

ﬁnes or bonds as part of the enforcement mechanism. Limão and Sagg i show that the use of monetary ﬁnes does not

help if ﬁnes must be self-enforcing and hence ultimately supported by the threat of retaliatory tar iffs. On the other

hand, bonds can enhance efﬁciency if they are posted with a third party prior to trading, and the third party uses the

bond to compensate the injured country in case a violation is committed: intuitively, this can relax self-enforcement

constraints because it indirectly injects some exter nal enforcement into the ag reement.

See Section 3.1.5 for a discussion of these papers. In Maggi and Staiger’s models, retaliation is not modeled explicitly,

but is captured in reduced form as a deadweight cost associated with government-to-government compensation. I

also note that in these models there is no private information, but political-economy shocks are not veriﬁable by the

court/DSB, and this for m of infor mation asymmetry has similar implications as private information for the design of

incentive contracts.

368 Giovanni Maggi

as specifying a tariff sc hedule that links the tariff to the realization of the political-economy

parameter . In order to be self-enforcing, the TA must discourage two types of deviations:

(i) “on-schedule” deviations, whereby the government applies a tariff that is meant for

another “type”; and (ii)“off-schedule” deviations,whereby the government applies a tariff

level that is not meant for any “type”. In this context, one can think of the punitive role of

retaliation as discouraging off-schedule deviations,while the screening role of retaliation is

to discourage on-schedule deviations. A paper that focuses on these themes is Martin and

Vergote (2008), which considers a setting where shocks are i.i.d. over time. An interesting

ﬁnding of this paper is that, even if theTA can specify a“reciprocity” mechanism whereby

a government’s tariff increase is met with a contemporaneous tariff response by the

other government, it is always desirable for the TA to include future retaliation, provided

governments are sufﬁciently patient. Relatedly, Bagwell (2009) considers a setting where

political-economy shocks can be persistent over time,and shows that persistence can make

enforcement more difﬁcult,because of a “ratchet”effect whereby some government types

are more reluctant to reveal themselves today for fear of being perceived as having weak

retaliatory power tomorrow. For a broader and informal discussion of these themes, see

also the piece by Bagwell (2008).

Finally, it is important to note that retaliation is not the only way to provide

“truthtelling” incentives to governments. For example, Bagwell and Staiger (2005a) argue

that this can be accomplished by imposing a dynamic constraint on the use of safeguard

actions. The nature of this constraint is that, if a government invokes the escape clause

toda y, it has to wait a certain amount of time before invo king it again. Intuitively, this

can mitigate governments’ temptation to misrepresent their information and over-use

the escape clause. A dynamic usage constraint of this kind is contained in the WTO’s

Safeguard Agreement.

3.2.3. The Enforcement Role of the DSB

An important question concerning the enforcement of TAs is whether and how a judicial

system such as the WTO’s DSB can assist with the enforcement of the TA. The answer

to this question is far fr om obvious, since in reality TAs must be self-enforcing and inter-

national courts have no direct enforcement power. One possible answer to this question

is suggested by Magg i’s (1999) model, which I brieﬂy discussed in the previous section:

the DSB can make multilateral enfor cement possible,by disseminating information about

violations of the agreement to the whole trading community. To illustrate, suppose that

country A commits a violation against country B, and that in the absence of a DSB this

violation would be observed only by country B . Then a potential role for the DSB is to

Another potential way to induce truthtelling is introducing cross-policy linkages,that is requiring a government to make

adjustments in other policies when raising trade protection.This idea plays a key role in the models by Lee (2007,2011).

International Trade Agreements 369

verify the violation and bring it to the attention of third countries, thus exposing country

A to punishments by third countries.

It is interesting to note that in the context of theWTO institution there exists another

procedure that helps disseminate information about trade policies, namely the “Trade

Policy Review Mechanism.” This mechanism is arguably complementary with the DSB

as a way to disseminate information and improve multilateral monitoring: the former

is a systematic, periodic, and wide-rang ing review, whereas the DSB conducts more

thor ough and targeted investigations when a country (or a group of countries) ﬁles a

complaint.

In practice, one important aspect of the WTO’s dispute settlement procedure is that it

encourages governments to renegotiate,even following clear violations of the agreement.

In Section 3.1.5, I pointed out that the renegotiation of trade policy commitments can

be beneﬁcial when the TA is an incomplete contract. However, as the repeated game lit-

erature has abundantly made clear, the renegotiation of punishments can have deleterious

effects on cooperation, because it can undermine the credibility of punishments. This is

the core of the argument in Ludema (2001), who argues that the WTO’s dispute settle-

ment pr ocedure can have adverse effects on cooperation if it encourages the renegotiation

of punishments. Interestingly,Watson et al. (2008) make a point that goes in the oppo-

site direction: they argue that, in the absence of a dispute settlement procedur e, after a

deviation countries would quickly renegotiate, thus undermining the enforcement of

the agreement, whereas the presence of a dispute settlement procedure can slow down

renegotiation and make it more costly, and this paradoxically can facilitate cooperation.

The above-mentioned papers raise an interesting question concerning the design of

dispute settlement procedures: to what extent should they encourage renegotiation and

settlement between governments? One clear point that emerges from this literature is that

it is critical to distinguish between two kinds of renegotiation, namely the renegotiation

of substantive policy rules and the renegotiation of enforcement rules: the former tends t o

be desirable, the latter tends to be harmful. But in reality the distinction between these

two forms of renegotiation can be blurred, and an interesting research question would

be whether and how a dispute settlement procedure can be designed to encourage one

form of renegotiation and discourage the other.

4. REGIONAL TRADE AGREEMENTS

In recent years there has been a considerable amount of theoretical and empirical research

on regional trade agreements (RTAs), spurred at least in part by the growing role of this

Another theoretical paper where the DSB facilitates the self-enforcement of TAs by changing the information

structure of the game, but in a two-country setting, is Park (2011). In Park’s model, each gover nment privately

observes a noisy signal of the other government’s trade policy, and the DSB can facilitate cooperation by converting

the privately observed signals into public signals. Here I also note that most o f the academic research on the enforce-

ment role of the DSB has been theoretical, but there are a few notable empirical papers on this topic, see in par ticular

Bown (2004a), Bown (2004b), Reinhardt (2001),andBusch and Reinhardt (2002).

370 Giovanni Maggi

type of agreement in the real world. In this section I will focus on three topics: (a) the

economic and political determinants of RTA formation; (b) the impact of an RTA on its

members’ external trade barriers and on multilateral trade liberalization;and (c) the design

of rules for trade negotiations. The common theme of this section is an emphasis on the

endogeneity of trade policies. I will leave aside, on the other hand, the older question

of how an exogenously formed RTA affects trade ﬂo ws and welfare (e.g. through trade-

diversion and trade-creation effects), which has been covered extensively in Baldwin and

Venables’ (1995) handbook chapter.

4.1. Determinants of RTAs

A question of obvious positive and normative r elevance concerning RTAs is: in a world

where governments are motivated by political economy considerations, under what con-

ditions are RTAs more likely to form? And in particular, is an RTA more likely to form

when it is beneﬁcial or detrimental to global welfare? This question is examined by

Grossman and Helpman (1995b), who consider a setting similar to Grossman and Help-

man (1995a) but with two small economies that can choose whether to form a free trade

agreement (FTA). Given the presence of producer lobbies, an FTA is more likely to be

adopted when it generates larger rents for the producers of both countries, a situation

that Grossman and Helpman label “enhanced protection.” In a given sector, enhanced

protection can occur under the following circumstances. Suppose that one country has

a lower external tariff than the other, and that as a result of the FTA producers from

the former country can export all of their output to the latter country without affecting

local prices in the latter country. Then producers in the former country get higher rents

while producers in the latter country are not affected. If the FTA generates enhanced

protection in a sufﬁcient number of sectors and in a relatively balanced way between the

countries, then it will be politically viable. But since enhanced protection is mor e likely

when the FTA causes trade diversion, the conclusion is that more trade-diverting FTAs

are more likely to be adopted.

A similar question is examined by Krishna (1998), but in the context of an oligopolis-

tic model with segmented markets, where governments care only about the proﬁts of

their domestic ﬁrms. In such an environment, an FTA is adopted if and only if it increases

proﬁts in both countries. This in turn is mor e likely to happen when the FTA leads to

a bigger reduction in the market share of non-member-country ﬁrms in the member

countries’ markets, or in other words, when the FTA induces more trade diversion.

The theoretical research on the trade-diversion and trade-creation effects of RTAs was developed mostly before 1995,

but there are some notable recent contr ibutions,in particular Panagariya and Krishna (2002), which extends the classic

Kemp-Wan result (which applies to customs unions) to the case of free trade agreements, and Freund (2000a),which

compares the welf are effects of alternative (exogenous) paths of trade agreements leading to global free trade. In recent

years there have been also some interesting empir ical studies on the trade-diversion and trade-creation effects of RTAs:

some prominent examples are Treﬂer (2004), Romalis (2007), Magee (2008),and Clausing (2001).

International Trade Agreements 371

Therefore Krishna’s analysis delivers a similarly pessimistic message as Grossman and

Helpman (1995b):FTAs are more likely to be adopted when they ar e more trade-diverting

and hence detrimental to welf are.

A counterpoint to the pessimistic message of the tw o papers discussed above is repre-

sented by the well-known argument (made for example by Kr ugman, 1991)thatRTAs

are more likely to form among “natural” trading partners, that is countries characterized

by especially large gains from mutual trade liberalization (for example because of their

geographical proximity or their comparative advantage structure). If this is the case, the

argument goes, the RTAs that emerge in equilibrium will be more likely to cause trade

creation than trade diversion, and hence more likely to increase welfare.

Given the seemingly contrasting theoretical arguments outlined above, the question

of the welfare impact of endogenously formed RTAs is ultimately an empirical one, and

indeed this question has been tackled by a number of interesting empirical papers. Krishna

(2003) takes a structural approach to this question, estimating a general equilibrium

model and using its structural parameters to examine the welfare effects of a number

of hypothetical RTAs. Krishna ﬁnds that 80% of these hypothetical RTAs would be

welf are-improving,but interestingly,he ﬁnds that neither geographical variables nor trade

volumes are signiﬁcantly correlated with the welfare gains, thus offeri ng little support for

the natural-trading-partners view of the world. Baier and Bergstrand (2004) ﬁnd that the

lik elihood of an RTA is higher when countries are closer to each other and are more

isolated from the rest of the world, a ﬁnding that supports the view that “natural” trading

partners are more likely to form RTAs. Baier and Bergstrand (2007) examine the impact

of RTAs on trade ﬂows when taking into account the endogeneity of RTA formation.

They ﬁnd that, when this endogeneity is recognized, the trade-creation effects of RTAs

appear to be much larger , and in particular, the impact of RTAs on trade ﬂows increases

ﬁvefold relative to estimates that take RTA formation as exo genous. Egger and Larch

(2008) ﬁnd results that are consistent with those of the above-mentioned papers using a

larger sample, and furthermore ﬁnd that an RTA is more likely to form when there are

pre-existing RTAs involving near-by countries.

Another question that has received some attention in the literature is how the like-

lihood of RTA formation is affected by multilateral trade liberalization. Freund (2000b)

examines this question within a repeated-game model and ﬁnds that deeper multilateral

trade liberalization leads to more RTAs, for two reasons: it increases the i ncentives to

form an RTA and it increases the likelihood that it is self-enforcing.

At the empirical

level, Fugazza and Robert-Nicoud (2012) ﬁnd some evidence that multilateral t rade lib-

eralization increases the likelihood of subsequent RTA formation. More speciﬁcally, they

ﬁnd that after the Uruguay Round, the US had a higher propensity to liberalize trade on

a preferential basis in goods where it had granted the deepest multilateral tariff cuts.

See also Ethier (1998) for an examination of the impacts of multilateral trade liberalization on subsequent RTAs.

372 Giovanni Maggi

Finally, an interesting question is: what determines the choice between an FTA and

a customs union (CU)? Empirically, most RTAs take the form of FTAs. Facchini et al.

(2013) propose a theoretical explanation for stylized fact, based on a three-country polit-

ical economy model with imperfect competition, where each country elects a represen-

tative to choose trade policies. Under a CU, since tariffs are chosen collectively by the

member countries, the voters of each country strategically delegate power to a more

protectionist representative. Facchini et al. show that, because of this strategic delegation

effect, an FTA tends to imply higher welfare for member countries and is more likely to

be politically viable than a CU.

4.2. Impacts of RTAs

In this section I will focus on two related themes: the impact of RTAs on member

countries’ trade barriers against outsiders (“external” trade barriers) and the impact of

RTAs on multilateral trade liberalization.

The impact of RTA formation on external trade barriers has been the subject of a

sizable theoretical literature. One point that emerges clearly from this literature is that

the qualitative impact of RTAs on external tariffs depends crucially on whether the

agreement tak es the form of an FTA or of a CU. I will focus ﬁrst on FTAs.

Various papers ha ve pointed out a strong tendency of FTAs to lead to lower external

trade barriers. This is generally known as t he “tariff complementarity” effect, but it is

important to point out that such effect can arise from two distinct mechanisms. The ﬁrst

one, highlighted by Richardson (1993), occurs when the FTA leads member countries to

compete for tariff revenue,thus inducing them to reduce external tariffs. Interestingly,this

mechanism can occur even for small countries that use tariffs only for political-economy

r easons and have no TOT power. A second mechanism was pointed out by Bagwell

and Staiger (1999b): an FTA leads member countries to import less from non-member

countries, and if member countries have TOT power this reduces their incentives to

manipulate TOT vis-à-vis non-members, in turn leading to lower external tariffs.

The tariff complementarity effect however is not the only possible effect at play in

determining the impact of FTAs on exter nal tariffs, and other effects may arise that mit-

igate or overturn it. Limão (2007) for example shows that an FTA can lead to higher

external tariffs if the FTA serves also non-trade objectives (such as enhancing cooperation

on labor standards or security issues). Several real-world FTAs appear to have this feature,

with one country (typically the US or the EU) granting tariff preferences, and the other

(typically a less developed country) making non-trade concessions. In this type of situ-

ation, the country that grants tariff preferences may be better off increasing its external

Other papers that have highlighted tariff-complementarity effects in different settings are Bond et al. (2004), Cadot

et al. (1999), Bagwell and Staiger (1997a), Ornelas (2005a,c),andSaggi andYildiz (2010).

International Trade Agreements 373

tariffs, because t his enhances the value of the preferences and hence allows it to extract

larger non-trade concessions.

In the case of CUs, tariff-complementarity effects may still be present, but two new

forces arise that push in the opposite direction. The ﬁrst one is known as the “market

power” effect: if two member countries import the same good fr om outsiders, once the

CU is formed they jointly have more power overTOT, and thus have a stronger incentive

to raise tar iffs against outsiders. T he second effect is known as the “coordination” effect:

if country A increases the tariff on imports of a certain good from country B, this has a

positive externality on all other countries (both importers and exporters of that good),

and a CU allows member countries to internalize this externality, thus leading to higher

external tariffs. The ﬁrst paper in the literature to highight these two effects was Kennan

and Riezman (1990). Note that the coordination effect can arise even in the absence of

the market power effect: t his is crystal-clear in a setting of “competing exporters,” where

each country is the sole importer of a given good. Once all the effects are taken into

account, a CU can still lead to lower external tariffs, but this is less likely than in the case

of an FTA (as shown by Bagwell and Staiger, 1999b).

In a series of papers, Emanuel Ornelas has established a link between the literature

on the impacts of RTAs and that on the deter minants of RTAs, by showing that taking

into account the impact of RTAs on external tariffs has important implications for the

lik elihood of RTA formation in the ﬁrst place.

Ornelas (2005a) considers a model similar to Grossman and Helpman (1995b),butin

which external tariffs are deter mined endogenously. Ornelas identiﬁes a “rent destruc-

tion” effect of the FTA, which arises from the fact that the rents from external tariff

protection spill over to partner countries under the FTA. The rent-destruction effect

lowers the incentives of special interest groups t o lobby for protection, and t his creates a

tendency of FTAs to induce reductions in external tariffs. Interestingly, when political-

economy motivations are stronger, the drop in external tariffs is larger, thus FTAs are

more conducive to multilateral trade liberalization.

The second point made by Ornelas (2005a) is that, since an FTA lower s the total

amount of rents that trade protection can generate, this has important implications for

the political viability of the FTA. Ornelas starts by considering a situation where there

is no ex-ante lobbying (that is, no lobbying to inﬂuence directly the decision to join

the FTA), and shows that in this case only FTAs that are sufﬁciently welfare-enhancing

can be politically viable: intuitively, a w elfare-r educing FTA cannot be attractive to the

gov ernment, because it reduces both the level of welfare and the a vailable amount of

Another effect that may work against the tariff complementarity effect was highlighted by Stoyanov (2009):ifforeign

lobbying is possible, producers from FTA partners may have stronger incentives to lobby for higher external tariffs

once the FTA is for med, since their gains from such exter nal tariffs are higher under the FTA.

The effects of a CU on external tariffs have been examined also by Krugman (1991), Bond and Syropoulos (1996),

Bagwell and Staiger (1997b),and Cadot et al. (1999).

374 Giovanni Maggi

rents. Ornelas then allows for ex-ante lobbying, and ﬁnds that in this case a welfare-

reducing FTA may in some cases be politically viable, but this is made less likely by the

rent-destruction effect of the FTA: i n particular, a welfare-reducing FTA can be viable

only if the governments’ valuation of welfare relative to contributions is neither too small

nor too large.T he bottom line of this paper then is that the rent-destruction effect reduces

the political viability of welfare-reducing FTAs.

The impact of RTA s on external trade barriers has also been the subject of recent

empirical work. Estevadeordal et al. (2008) focus on the effect of preferential trade lib-

eralization on external tariffs in Latin America from 1990 to 2001. An appealing feature

of this dataset is t he wide var iation in trade preferences across sectors and over time.

Employing a rich set of ﬁxed effects, these authors ﬁnd that preferential tar iff reduction

induces faster decline in external tar iffs. Furthermore, they ﬁnd that this effect is present

only for FTAs, not for CUs, and is stronger in sectors where the potential for trade

diversion is larger.

Interestingly, Limão (2006) and Karacaovali and Limão (2008) ﬁnd results that seem-

ingly diverge from those of Estevadeordal et al. (2008). These two papers examine the

impact of preferential trade liberalization by the US and the EU on multilateral trade lib-

eralization, and ﬁnd that the US and the EU liberalized less during the Uruguay Round

in sectors where they had granted tariff preferences, suggesting that preferential liber-

alization might hinder the cause of global free trade. What can explain the difference

in ﬁndings between these papers and Estevadeordal et al. (2008)? Theory can help us

answer this question. A key difference between the two approaches is that Limão (2006)

and Karacaovali and Limão (2008) focus on the US and the EU, whereas Estevadeordal

et al. (2008) focus on developing countries. Tariffs are considerably higher in developing

countries, so the potential for trade diversion is larger for these countries, and as the-

ory suggests, this implies a stronger tariff complementarity effect. Furthermore, Limão’s

(2007) theoretical analysi s suggests that preferential liberalization may hinder global free

trade if RTAs are formed also for non-trade reasons, and this is more often the case for

North–South RTAs than for South–South RTAs.

Thus far I have focused on how the formation of an RTA affects its member countries’

unilateral choices of external tariffs. Next I focus on the impact of RTAs on the political

viability of multilateral trade agreements.

Similar results are obtained by Ornelas (2005b,c) in the context o f an oligopolistic model similar to Krishna (1998).

Two other papers present ﬁndings that are consistent with those of Estevadeordal et al. (2008): Calvo-Pardo et al.,

(2009) ﬁnd that the formation of ASEAN led to a reduction in external tar iffs by its member countries, and Bohara

et al. (2004) ﬁnd that the increase in preferential imports from Brazil to Argentina that followed the formation of

MERCOSUR led to a decrease in Argentina’s external tariffs.

Related to this literature is also a recent paper by Baldwin and Jaimovich (2012),which focuses on the impact of RTAs on

subsequent RTAs.This paper extends Baldwin’s (1995) model of “domino”regionalism and tests its main empirical pre-

diction, namely that the formation of an RTA has a contagion effect and increases the likelihood that further RTAs will

be for med. Using a comprehensive panel of FTAs, Baldwin and Jaimovich ﬁnd support for the contagion prediction.

International Trade Agreements 375

Levy (1997) considers a model where gains from trade can arise from differences in rel-

ative factor endowments and/or from increased product variety, and takes a median-voter

approach to the choice of trade policies. He shows that, if the FTA pro vides a country’s

median voter with disproportionately large gains, it may raise his or her reservation utility

above the level offered by a multilateral agreement, thus under mining political support

for the latter. This undermining is more likely to occur in FTAs that involve countries

with similar relative factor endowments. For example, suppose Germany joins the EU.

Assuming that EU countries have similar relative factor endowments, this will beneﬁt

the median voter in Germany mostly through variety gains. In the next stage, Germany

considers signing a multilateral agreement. If Germany is relatively rich in capital and

its median voter is an unskilled worker, the multilateral agreement is likely to damage

the median voter through Stolper-Samuelson effects, without providing much additional

variety gains, so he/she will block the multilateral agreement. On the other hand, if the

same median voter were asked if he/she supports a multilateral agreement before joining

the EU, then the answer may be y es, because in this case the variety gains may outweigh

the adver se Stolper-Samuelson effects.

The model by Krishna (1998), already mentioned in Section 4.1, leads to a similar

result, in spite of the very different structure. Recall that Krishna focuses on a Grossman-

Helpman type model with oligopolistic competition. After establishing that politically

viable FTAs are more likely to be trade-diverting, Krishna examines how an FTA affects

the political viability of a multilateral agreement. The key ﬁnding is that the formation

of an FTA may increase producers’ opposition to a subsequent multilateral agreement,

because the latter may reduce or eliminate the rents created by the FTA, and it can even

rever se the preferences of producers,from supporting a multilateral agreement to opposing

it.Thus, taken together, the papers by Levy (1997) and Kr ishna (1998) deliver a pessimistic

message about the impact of RTAs on the political viability of multilateral agreements.

Finally, Bagwell and Staiger’s (1999a) model (already discussed in Sections 2.1 and

3.1.4) also delivers pessimistic implications for the impact of RTAs on multilateral trade

agreements, but for a ver y different reason than the models discussed above. A simple

corollary of Bagwell and Staiger’s analysis is that the presence of RTAs, by breaking t he

ability of the MFN rule to channel all international policy externalities into a single

world-price externality, undermines the effectiveness of the MFN and reciprocity rules

in guiding countries toward an efﬁcient policy outcome.

Here I will mention two other strands of literature that are quite interesting but distinct from the ones discussed in

the text. The ﬁrst one examines how RTAs affect the self-enforceability of a multilateral agreement. In particular,

Bagwell and Staiger (1997a,b) focus on the transition period during which an RTA is being negotiated, showing

that the anticipation of the RTA has an important impact on the sustainability of multilateral tariff cooperation; Saggi

(2006) considers the impact of RTAs on the sustainability of a multilateral agreement in an oligopolistic setting, ﬁnding

that the presence of an RTA under mines multilateral cooperation when countries are symmetr ic, but not necessarily

when countries are asymmetric; ﬁnally, Bond et al. (2001) focus on the transition period during which trade barriers

within a CU are phased out, and examine how the deepening of intra-CU trade liberalization affects the sustainability

376 Giovanni Maggi

4.3. Rules for Trade Negotiations

In this section I return to my emphasis on the design of rules, but this time I focus

on rules that constrain trade negotiations rather than the policy choices of individual

governments. A number of papers have been written on the question of what rules (if

an y) should regulate trade negotiations, but this literature can be hard to tame, as the

modeling approach and the exact nature of the question seem to shift from paper to

paper. In what follo ws I propose a simple conceptual framework that might be useful to

organize our thinking and deﬁne more clearly the relevant questions.

Consider the following senario: imagine that, at some ex-ante stage, all countries

involved in the trading system bargain over the rules that regulate future trade negotia-

tions. Also assume that at the ex-ante stage countries cannot write a complete agreement,

and in particular they cannot specify actual trade policies, but only negotiation rules (oth-

erwise there would be no reason to set negotiation rules in the ﬁrst place). One possible

reason why negotiation rules might be desirable is that governments may have incentives

to sign RTAs that exert negative externalities on non-member countries, in which case

the resulting outcome may be globally inefﬁcient.

In the scenario I just described, one can make a distinction between two types of rules

constraining trade negotiations: (1) Rules that constrain the nature of the agreement itself.

An example of this type of rule is one that prohibits RTAs, or equivalently, a rule requiring

that any agreement be approved by all countries. Another example is a rule that prohibits

CUs but not FTAs. I refer to these rules (for lack of a better expression) as rules-to-make-

rules; (2) Rules that constrain the policy content of the agreements. Examples include

reciprocity-type rules, which r equire that tariff r eductions be balanced acr oss countries,

and non-discrimination rules, such as the MFN rule. These rules do not constrain the

type of agreements that countries sign, but do constrain the policies (or policy changes)

that countries can agree upon. I refer to these as policy-content rules.

Note that the GATT-WTO currently does not impose rules-to-make-rules, but only

policy-content rules, and more speciﬁcally: (i) GATT Article I imposes the MFN rule,

of multilateral agreements. The other line of research I want to mention is the work by Martin et al. (2008, 2012),

who explore the two -way relationship between RTAs and military conﬂicts. At the theoretical level, they argue that

RTAs increase the opportunity cost of conﬂict, thereby reducing the likelihood of war, and conversely, countries with

a higher likelihood of conﬂict are more likely to sign RTAs as a way to promote peace. Furthermore, they argue

that this logic does not apply to multilateral trade agreements, because multilateral trade openness decr eases bilateral

dependence from trade with any given country and hence the cost of a bilateral conﬂict. At the empirical level, they

ﬁnd support for these predictions using a large dataset of military conﬂicts during the 1950–2000 period.

In the scenario outlined above I am assuming that at the ex-ante stage countries can only specify rules constraining

future negotiations, not r ules constraining individual government policies. In reality, of course, even from the very

beginning the GATT imposed both types of rules. I am separating the stages at which the two different types of rules

are designed for conceptual simplicity. Also, the distinction between policy-content rules and rules-to-make-rules can

be subtle but should be conceptu ally clear. For example, note that imposing the MFN rule does not logically imply

prohibiting FTAs, because an FTA in principle need not violate MFN (an FTA eliminates tariffs between its members,

but this does not prevent a member from respecting the MFN rule, which it can do by eliminating the relevant tariffs

vis-à-vis non-members).

International Trade Agreements 377

so any negotiated policy changes must be extended in a non-discriminatory way to all

member countries; (ii) GATT Article XXIV allows an exception to the MFN rule for

the case of an RTA that eliminates substantially all trade barriers among its members, but

requires the RTA member countries not to raise their external tariffs above pre-RTA

levels;

and (iii) GATT requires that negotiations adhere to the principle of reciprocity,

although as discussed earlier, this is an informal principle rather than a strict rule.

It is import ant, however, to consider also the possibility of rules-to-make-rules, both

from a positive perspective (why are there no such rules in GATT-WTO?) and from a

normative one (should such rules be imposed?). In particular , the simplest such rule—a

ban on RTAs—has a special theoretical interest. Indeed, an interesting question can be

posed here: if there are no frictions in multilateral bargaining (as assumed in most of the

existing models), why would a ban on RTAs not be optimal? If governments are forced

to choose trade policies at a multilateral bargaining table, why would this not achieve

efﬁciency?

I will return to this question at the end of this section, after surveying the

recent research in this area.

If one adopts the conceptual framework outlined above, one needs to take a stand

on what objective function the rules are supposed to maximize. There are two possible

questions that can be asked, each of which is interesting in its own right:

1. What rules maximize the likelihood of reaching global free trade? This is essentially

the question raised by Bhagwati (1993): are RTAs “building blocks” or “stumbling

blocks” on the path to global free trade?

2. What r ules maximize the joint payoff of all governments? This is a more positive

question and can lead to different answers than the previous one, if one allows for

political motivations in the governments’ objectives.

Ideally, a full answer to either question requires comparing the equilibrium conﬁguration

of trade agreements (or their path, if one takes a dynamic approach) under alternative

negotiation rules, of course allowing for RTAs as well as multilateral agreements. Models

that focus on the impact of exogenously formed RTAs on multilateral agreements, such

as those that I surveyed in the previous section, evidently can only shed partial light on

this question.

4.3.1. Policy-Content Rules

As I mentioned earlier, the key policy-content rules that regulate trade negotiations in

the GATT-WTO system are the MFN rule and—at a more i nformal level—the reci-

procity pr i nciple. In Section 3.1.4, I discussed papers b y Bagwell and Staiger (1999a) and

Exceptions to MFN are also allowed in a number of extraordinary circumstances, such as national-security or health

hazards (GATT Article III).

Of course it is possible that such a rule is not strictly needed, if global efﬁciency can be achieved also in the absence

of any r ule on trade negotiations, and in this case a ban on RTAs will only be weakly optimal. Indeed, this is the case

in some of the models I will survey below.

378 Giovanni Maggi

Ossa (2011) that examine the implications of these rules in the context of multilateral

trade negotiations, but these papers do not consider endogenous RTAs.

A paper that does consider the impact of these rules on the endogenous formation of

RTAs is Bagwell and Staiger (2005b).This paper considers a two-stage scenario where,in

the ﬁrst stage, countries can sign a multilateral agreement, and in the second stage , pairs

of countries can sign bilateral agreements. The ﬁrst point of the paper is that, if bilateral

negotiations are left unrestricted, a problem of “bilateral opportunism” is likely to arise:

after multilateral trade concessions have been exchanged, there is an incentive for a pair

of countries to take a further step and liberalize trade bilaterally, but this will erode the

v alue of the concessions that the excluded country had obtained in the initial multilateral

negotiation, and this in turn makes countries more reluctant to make multilateral trade

concessions in the ﬁrst place. The second point of the paper is that the bilateral oppor-

tunism problem described above may be solved if trade negotiations are disciplined by

the MFN rule in conjunction with a reciprocity rule. To gain intuition for this result,

suppose there are only two goods. Then the MFN rule ensures the existence of a single

relative world price (as explained in Section 2.1.2), and reciprocity ensures that this world

price i s effectively ﬁxed by the initial multilateral agreement; and given that the world

price is preserv ed in subsequent bilateral negotiations, the welfare of countries that do

not participate i n a bilateral negotiation is preserved as well.

In a related paper, Bagwell and Staiger (2010b) consider a setting in which coun-

tries can sequentially sign bilateral agreements in the presence of t he MFN rule. Two

inefﬁciencies tend to arise in this setting: the ﬁrst one is due to the bilateral oppor-

tunism problem highlighted above; the second is that, since later negotiating partners can

free-ride on the MFN concessions that a country makes to early negotiating partners, a

country may be induced to offer too little in the early negotiations (“foot-dragging”).

Bagwell and Staiger then argue that these inefﬁciencies can be removed if two additional

rules ar e imposed, along with MFN: the reciprocity rule and a “non-violation” rule along

the lines of GATT Article XXIII.1b. These rules together act as a device to “secure” the

concessions received by a country in early negotiations and protect them from potential

free riding and bilateral opportunism in the future.

A different argument for the desirability of the MFN rule is proposed by McCalman

(2002).This contribution is notable because it is a rare example of a model that explicitly

introduces bargaining frictions in multilateral trade negotiations. In this model, a large

country negotiates with N small countries over tariffs and transfers,and each small country

has private infor mation about its gains from an agreement. McCalman compares two

scenarios, one where the large country can make different offers to different countries,

and one where the MFN rule constrains the large country to make the same offer to

all countries. The large country of course is worse off under the MFN rule, since the

This is a good juncture to mention a paper by Ludema and Mayda (2009), which examines empirically the free-rider

problem associated with the MFN r ule, ﬁnding that this problem is indeed of ﬁrst-order empirical impor tance.

International Trade Agreements 379

latter constrains its choice, but global efﬁciency may be higher under MFN, and this is

more likely when N is larger. The reason lies in the fact t hat bargaining is inefﬁcient, due

to private information. If bargaining were frictionless, unconstrained bargaining would

always lead to efﬁciency, and the MFN rule would always be a bad idea. But in the

presence of private information, unconstrained bargaining is inefﬁcient, because the large

country is not able to appropriate the entire surplus from the negotiations, and as a

consequence it is possible that imposing MFN increases the welfare of the N countries

more than it reduces the welfare of the large country.

4.3.2. Rules-to-Make-Rules

The simplest rule-to-make-rules, and the one that has received the most attention in

the literature, is a rule that prohibits RTAs. Would the wo rld be more efﬁcient if all

agreements had to be multilateral in nature? In this section I discuss a number of recent

papers that speak to this question.

One of the ﬁrst papers to examine the desirability of a rule banning RTAs is McLaren

(2002), who focuses on the implications of irreversible investments for the formation of

trade agreements. McLaren considers a world with three countries: in the ﬁrst period,

individuals choose in what sectors to allocate their resources;in the second period,there is

a coalition-formation game among governments, which can yield an FTA, a multilateral

free trade agreement, or no agreement at all. Negotiations are costly, and this cost is

higher for a multilateral negotiation. This model may have multiple equilibria, including

an equilibrium where global free trade arises (and the allocation is efﬁcient),and equilibria

where two of the countries form an FTA (and the allocation is inefﬁcient).The latter type

of equilibr ium i s based on a self-fulﬁlling prophecy. If individuals expect an FTA between

countries A and B, they will make investment decisions that make these countries more

specialized relative to each other, thus increasing the gains from trade between them.

At the same time, countries A and B will become less specialized relative to the outside

country, thus decreasing the gains from trade between them and the outside country.

As a consequence, ex-post countries A and B will ha ve a str ong incentive to sign an

FTA, while the beneﬁts from a multilateral agreement will be small. Ex-ante, an FTA

equilibrium of this kind may be inefﬁcient if the allocation distortions associated with

the FTA outweigh the sa vings in negotiation costs. Intuitively, then, there is a region of

parameters in which an FTA equilibrium exists and is inefﬁcient, and therefore a ban on

FTAs is strictly desirable.

Another policy-content rule imposed by the GATT-WTO, as I mentioned in the previous section, is contained in

GATT Article XXIV. Mrazova et al. (2013) examine the implications of Article XXIV as it applies to CUs, requiring

that the common external tariff of the CU must not exceed the average of its members’ pre-CU tar iffs. Mrazova et al.

argue that Article XXIV increases the probability that free trade emerges in equilibr ium, but when free trade does not

arise in equilibrium, the constraints imposed by this rule may lead to a reduction in world welfare.

380 Giovanni Maggi

Goyal and Joshi (2006) adopt a network-formation approach to study the formation of

FTAs, using a particular notion of “stability” to determine the conﬁgurations of FTAs that

can arise in equilibrium. Focusing on a setting with a homogenous good and symmetric

countries, Goy al and Joshi ﬁnd that the complete FTA network, which yields global free

trade, is a stable network, thus suggesting that FTAs are “building blocks” toward global

free trade. Another network-theoretic model of FTA formation is Furusawa and Konishi

(2007), which differs from Goyal and Joshi (2006) in that it allows for differentiated

goods and a richer pattern of asymmetries between countries. Consistently with Goyal

and Joshi, this paper ﬁnds that when countries are symmetric,the complete FTA network

(global free trade) is stable. Howev er, if goods are highly substitutable, there may also be

other stable networks that do not yield global free trade. And if countries are asymmetric,

the complete FTA network may not be stable. Furusawa and Konishi conclude that FTAs

may under some conditions be “stumbling blocks” toward global free trade.

Before moving on, I will make a general observation about the network-theoretic

approach to the analysis of RTA formation, of which I just discussed two examples. In

these network models, there is no meaningful distinction between a sequence of RTAs

that leads to global free trade and a multilateral ag reement. Indeed, multilateral agree-

ments per se are not consider ed at all, so these models, though capable of generating

interesting insights, are arguably not well equipped t o evaluate t he desirability of rules-

to-make-rules, such as a ban on RTAs. The models that I discuss next, on the other hand,

do allow for multilateral agreements as well as RTAs, and hence are better equipped to

examine this question.

Seidmann (2009) examines a three-country bargaining model in which countries can

negotiate FTAs, CUs, and multilateral agreements. Countries are allowed to continue

negotiating after reaching an agreement, and for this reason, an RTA can be used by its

member countries to impr ove their bargaining position for subsequent trade negotiations

(“strategic positioning” effect).Tw o important features of the model ar e that international

transfers are available and that global free trade maximizes the three countries’ joint

surplus. Seidmann studies the equilibrium conﬁguration of agreements, highlighting for

example conditions under which a hub-and-spoke structure emerges. Regarding the

question of whether RTAs are building blocks or stumbling blocks toward global free

trade, in Seidmann’s model global free trade may or ma y not be reached if RTAs are

This is a good juncture to mention a paper by Yi (1996), who takes a coalition-formation approach to study the

for mation of CUs.Yi compares two possible games: a “unanimous regionalism” game, where a CU forms if and only

if all potential members agree to form the CU; and an “open reg ionalism” game, where each country chooses an

“address,” and then all the countries that have chosen the same address must be part of the same CU.Yi shows that the

grand CU (global free trade) is an equilibrium of the open-regionalism game, but typically is not an equilibrium of

the unanimous-regionalism game, and interprets this ﬁnding as suggesting that an “open regionalism” rule is desirable.

However it is not clear how to interpretYi’s notion of open regionalism in a way that has a meaningful counterpart

in the real world: Yi’s open-regionalism game implicitly assumes that a country is not free to leave a CU (if this were

the case then we would be in the unanimous-regionalism game), which seems like a far-fetched idea.

International Trade Agreements 381

feasible, while global free trade is always reached if RTAs are not feasible, thus a ban on

RTAs is always weakly desirable.

Saggi and Yildiz (2010) also consider a three-country bargaining model where gov-

ernments can negotiate RTAs as well as multilateral agreements. The negotiation game

is as follows: governments simultaneously name FTA partners, and an FTA is formed

if the announcements agree; if a government names both of the other countries, this is

interpreted as a proposal for a multilateral agreement, and if all governments propose a

multilateral agreement, it is implemented. A key feature of the model i s that international

transfers are not available. Saggi and Yildiz examine the implications of a rule banning

FTAs, by comparing a game where FTAs are allowed with one where they are not. Focus-

ing on coalition-proof Nash equilibria, they show that when countries are symmetric

global free trade is the only stable equilibrium, whether or not FTAs are allowed. But

when countries are asymmetric it may happen that global free trade is a stable equilib-

rium only if FTAs are allowed, so a ban on FTAs can make global free trade less likely.

To understand this result intuitively, recall that there are no international transfers in the

model, so global free trade is not the only Pareto-efﬁcient outcome. Indeed, a ban on

FTAs may lead away from global free trade, but not away from Pareto-efﬁciency. If FTAs

are banned, it is possible that in equilibrium two countries agree to liberalize trade while

the third does not: this outcome is skewed in favor of the country that free rides, but is

Pareto efﬁcient. Thus, Saggi and Yildiz’s result should be interpreted as suggesting only

that FTAs may be needed to achieve global free t rade, not that FTAs may be needed to

achieve global efﬁciency.

Saggi et al. (2013) build on Saggi andYildiz (2010) by considering the case of CUs.

The main ﬁnding of this paper is that, in contrast with the case of FTAs,in the case of CUs

a “stumbling block”scenario is possible,in the sense that the freedom to pursue CUs may

prevent the attainment of global free trade. Interestingly, the reason for this difference in

results is not that a CU has a more harmful impact on outsiders than an FTA, but rather

that it implies a stronger incentive for insiders to deny access to outsiders.Taken together,

the two papers just discussed suggest that under some conditions it might be desirable to

ban CUs but not FTAs; but this is only suggestive, because in order to mak e this point

rigorously one would need to consider a model where CUs and FTAs ar e both options

available to governments, which is not the case in either of the abo ve-mentioned papers.

Aghion et al. (2007) also examine the building-block vs stumbling-block question

within a three-country model where countries can negotiate FTAs as well as multilateral

agreements, but with some key differences relative to Saggi andYildiz (2010).Inpartic-

ular,Aghion et al. assume that there is a leading country that chooses whether to engage

in sequential bilateral bargains or in a single multilateral bargain, and allow for politi-

cal econom y motivations in the government objectives. In addition, they assume that

international transfers are available. Aghion et al. deﬁne payoffs to be “grand-coalition

superadditive” if the payoff of the grand coalition is larger than the payoff of all countries

382 Giovanni Maggi

combined in alternative coalition structures. This property is satisﬁed for example if free

trade is Pareto-efﬁcient and each government maximizes national welfare. A key result

of the paper is that, if payoffs are grand-coalition superadditive, then the leading country

may prefer sequential bargaining or multilateral bargaining, depending on the nature of

coalition externalities, but in either case global free trade must emerge in equilibrium.

Aghion et al. then examine environments where grand-coalition superadditivity fails,

which can happen when political-economy motiv ations are strong. In this case, it is

possible that global free trade is attained only if FTAs are permitted (a building-block

scenario), and it is also possible that global free trade is attained only if FTAs are banned

(a stumbling-block scenario). Note that, as in Saggi andYildiz (2010), it may happen that

a ban on FTAs leads away fr om global free trade , but not that a ban on FTAs leads away

from Pareto-efﬁciency. However, unlike Saggi-Yildiz, this is not due to the absence of

international transfers, but rather to the possibility that governments may not maximize

welfare; indeed, a ban on FTAs can lead away from free trade only if governments do not

maximize welfare.

A related point is made by Ornelas (2008): if governments do not maximize welfare,

it is possible that FTAs have the effect of bringing the world closer to global free trade

relative to a multilateral agreement, and hence a ban on RTAs may lead the world away

from global free trade. However, this possibility arises in Or nelas (2008) for ver y different

reasons than in the papers mentioned above: here the reasons are that FTAs lead member

countries to lower their external tariffs, and that there is a tendency for FTAs to emerge

in equilibrium when they are trade-creating rather than trade-diverting.

A common message suggested by the papers discussed above is that,if rules are designed

ex-ante when governments are in “constitution-writing” mode and seek to maximize

global welfare, but ex-post government objectives may diverge from welfare, then a ban

on RTAs may be harmful, because RTAs may lead the world closer to global free trade

than a multilateral agreement.

I conclude this section by returning to the question I posed earlier: why has the

GATT-WTO not banned RTAs? As discussed above, the possible answers suggested by

the literature are two: (i) because rules are designed ex-ante when governments are in

constitution-writing mode and maximize welfare, whereas ex-post government objec-

tives may diverge from welfare; or (ii) because efﬁciency might also be achieved by means

of other rules, such as MFN and reciprocity. A third, and conceptually simpler, reason

why banning RTAs might not be a good idea is the presence of important frictions in

multilateral bargaining.This consideration is arguably of ﬁrst-order empirical importance,

but has received little attention in the formal literature.

If multilateral bargaining is less

efﬁcient than bilateral bargaining, for example because of the complexity of negotia-

tions when a large number of countries is in volved, then allowing RTAs may be strictly

One notable exception is McLaren (2002), but he models multilateral bargaining frictions in a very reduced-form way,

through a parameter that captures the extra cost of multilateral negotiations.

International Trade Agreements 383

desirable on efﬁciency grounds. But a more complete understanding of this issue would

require introducing bargaining frictions explicitly in our models and examining how they

depend on the set of countries and on the set of issues in volved in the bargain.

5. CONCLUSION

Coming back to the pessimistic statement made by Krugman (1997) (see beginning of

Section 2), one is tempted to ask: have the last 15 years of research pr oved Krugman

wrong? This is a matter of debate, b ut I think that, at a minimum, the literature has

demonstrated that the logic of economics can go a long way toward explaining the pur-

pose and design of trade agreements. As I have argued in this chapter, more progress is

needed along several dimensions, but we have made signiﬁcant advances toward under-

standing the motivations that drive countries to sign trade agreements and the reasons

why trade agreements are designed the way they are, and we are now in a better position

to evaluate possible reforms of existing rules from a normative standpoint. Most of the

research has been at the t heoretical level, but in the last few years there has been an

acceleration in empir ical research, spurred by the availability of new and better data sets.

What’s next for the economics of trade agreements? At a broad level, I think that

this research area is ready—both in terms of theoretical tools and data availability—to

follow a path that other research areas in international economics have fruitfully taken,

namely a tighter i ntegration between theoretical and empirical analysis. Some of the

most important questions on the table require counterfactual analysis, and this in turn

calls for structural and/or calibration approaches. Some recent papers that I discussed in

this chapter have mo ved in this direction, but we ar e still at the beginnings.

In terms of substantive questions, I will point to a few directions that seem promising

to me. One question that we still know little about is the empirical importance of “New

Trade” and domestic-commitment motives for trade agreements. Another important set

of open questions concerns multilateral trade negotiations in the presence of bargaining

frictions: why has the Doha round failed? Is it because there are no mutual gains left

on the table, or because of bargaining frictions due to the large number of countries

and issues on the table? If bargaining frictions are part of the problem, can bargaining

protocols be designed in a smarter way to facilitate more efﬁcient outcomes?

If history is of any guidance, the release of new important datasets can trigger new

waves of empirical and theoretical research. This will hopefully be the case for the recent

release by the WTO of an unprecedented dataset that includes extremely detailed infor-

mation about the bargaining that took place during GATT negotiation r ounds. This

dataset may help answ er new questions, such as those related to the nature of bargaining

frictions and the importance of bargaining protocols, as well as old questions, such as the

extent to which the MFN rule generates free-rider problems and whether bilateral trade

agreements are building blocks or stumbling blocks toward global free trade.

384 Giovanni Maggi

REFERENCES

Aghion, P.,Antràs, P., Helpman, E., 2007. Negotiating free trade. Journal of International Economics 73 (1),

1–30.

Amador, M., Bagwell, K., 2013.The theory of optimal delegation with an application to tariff caps. Econo-

metr ica 81 (4), 1541–1599.

Amador, M., Bagwell, K., 2012. Tariff revenue and tariff caps. American Economic Review Papers and

Proceedings 102 (3), 459–465.

Antras, P., Staiger, R.W., 2012a. Offshoring and the role of trade agreements. American Economic Review

102 (7), 3140–3183.

Antras, P., Staiger, R.W., 2012b. Trade agreements and the nature of inter national price determination.

Amer ican Economic Review Papers and Proceedings 102 (3), 470–476.

Bacchetta, M., Piermar tini, R., 2011. TheValue of Bindings. WTO Staff Working Paper, ERSD-2011-13.

Bagwell, K., 2008. Remedies in the WTO: an economic perspective. In: Janow, M.E., Donaldson, V.J.,

Yanovich, A. (Eds.),The WTO Governance, Dispute Settlement & Developing Countries. Juris Pub-

lishing, pp. 733–770 (Chapter 40).

Bagwell,K.,2009. Self-EnforcingTrade Ag reements and Private Information. NBERWorking Paper 14812.

Bagwell, K., S taiger, R.W., 1990. A theory of managed trade. The American Economic Review 80,

779–795.

Bagwell, K., Staiger, R.W., 1997a. Multilateral tariff cooperation during the formation of free trade areas.

International Economic Review 38, 291–319.

Bagwell, K., Staiger, R.W., 1997b. Multilateral tariff cooperation during the formation of customs unions.

Journal of International Economics 42, 91–123.

Bagwell, K., Staiger, R.W., 1999a. An economic theory of GATT. American Economic Review 89 (1),

215–248.

Bagwell, K., Staiger, R.W., 1999b. Regionalism and multilateral tariff cooperation. In: Piggott, J.,Woodland,

A. (Eds.), International Trade Policy and the Paciﬁc Rim. MacMillan, London, pp. 157–185.

Bagwell, K., Staiger, R.W., 2001. Domestic policies, national sovereignty and international economic insti-

tutions. Quarterly Journal of Economics 116 (2), 519–562.

Bagwell, K., Staiger, R.W., 2002. The Economics of the World Trading System. MIT Press, Cambridge,

MA.

Bagwell, K., Staiger, R.W., 2003. Protection over the business cycle. A dvances in Economic Analysis and

Policy 3 (1), 1–43.

Bagwell, K., Staiger, R.W., 2005a. Enforcement, private political pressure, and the GATT/WTO escape

clause. Journal of Legal Studies 34 (2), 471–513.

Bagwell, K., Staiger, R.W., 2005b. Multilateral trade negotiations, bilateral opportunism and the rules of

GATT/WTO. Journal of International Economics 67, 268–294.

Bagwell, K., Staiger, R.W., 2006. Will international rules on subsidies disrupt the world trading system?

Amer ican Economic Review 96 (3), 877–895.

Bagwell, K., S taiger, R.W., 2009. Delocation and Trade Agreements in Imperfectly Competitive Markets,

Mimeo.

Bagwell,K.,Staiger, R.W.,2010a.TheWTO: theor y and practice. Annual Review of Economics 2, 223–256.

Bagwell, K., Staiger, R.W., 2010b. Backward stealing and forward manipulation in the WTO. Journal of

International Economics 82 (1), 49–62.

Bagwell, K., Staiger, R.W., 2011. What do trade negotiators negotiate about? Empirical evidence from the

world trade organization. American Economic Review 101 (4), 1238–1273.

Bagwell, K., Staiger, R.W., 2012a. Proﬁt shifting and trade agreements in imperfectly competitive markets.

International Economic Review 53 (4), 1067-1104.

Bagwell,K.,Staiger,R.W.,2012b.The economics of trade agreements in the linear delocation model. Journal

of International Economics 88 ( 1), 32–46.

Bagwell, K., Mavroidis, P.C., Staiger, R.W., 2007. Auctioning counter-measures in the WTO. Journal of

International Economics 73 (2), 309–332.

Baier, S.L.,Bergstrand,J.H.,2004. Economic determinants of free trade agreements. Journal of International

Economics 64 (1), 29–63.

International Trade Agreements 385

Baier, S.L., Bergstrand, J.H., 2007. Do free trade ag reements actually increase members’ international

trade? Journal of International Economics 7 1 (1), 72–95.

Baldwin, R.E., 1987. Politically realistic objective functions and trade policy. Economics Letter s 24 (1),

287–290.

Baldwin, R.E., 1995. A domino theory of regionalism. In: Baldwin, R., Haaparanta, P., Kiander, J. (Eds.),

Expanding members hip of the European Union. Cambridge University Press, pp. 25–53.

Baldwin, R.E., Jaimovich, D., 2012. Are free trade ag reements contagious? Journal of International Eco-

nomics 88 (1), 1–16.

Baldwin, R.E.,Venables, A.J., 1995. Regional economic integration. In: Grossman, G., Rogoff, K. (Eds.),

Handbook of International Economics, vol. 3. North-Holland,Amsterdam, pp. 1597–1644.

Battigalli, P., Maggi, G., 2002. Rigidity, discretion and the costs of writing contracts. American Economic

Review 92 (4), 798–817.

Battigalli, P., Maggi, G., 2003. International Agreements on Product Standards:An Incomplete-Contracting

Theory. NBER Working Paper 9533.

Bajona, C., Chu, T., 2010. Reforming state owned enterprises in China: effects of WTO accession.

Review of Economic Dynamics 13 (4), 800–823.

Bajona, C., Ederington, J., 2012. Domestic Policies, Hidden Protection and the GATT/WTO, Mimeo.

Beshkar, M., 2010a. Optimal remedies in international trade agreements. European Economic Review 54

(3), 455–466.

Beshkar, M., 2010b. Trade skirmishes and safeguards: a theory of the WTO dispute settlement process.

Journal of International Economics 82 (1), 35–48.

Beshkar, M., 2011. Arbitration and Renegotiation in Trade Ag reements, Mimeo.

Beshkar, M., Bond, E.W., 2012. Cap and Escape in Trade A greements, Mimeo.

Beshkar, M., Bond, E.W., Rho,Y., 2011. Tariff Binding and Overhang:Theory and Evidence, Mimeo.

Bhagwati, J., 1993. Regionalism and multilateralism: an overview. In: de Melo, J., Panagariya,A. (Eds.), New

Dimensions in Reg ional Integration. Cambridge Unive rsity Press, pp. 22–51.

Bohara,A.K., Gawande, K., Sanguinetti, P., 2004. Trade diversion and declining tariffs: evidence from M er-

cosur. Journal of International Economics 64, 65–88.

Bond, E.W., Park, J.H., 2004. Gradualism in trade agreements with asymmetric countries. Review of Eco-

nomic Studies 69, 379–406.

Bond,E.W.,Syropoulos,C., 1996.The size of trading blocks:market power and world welfare effects. Jour nal

of International Economics 40, 412–437.

Bond, E.W., Syropoulos, C.,Winters, L.A., 2001. Deepening of regional integ ration and multilateral trade

agreements. Journal of International Economics 53, 335–362.

Bond, E.W., Riezman, R., Syropoulos, C., 2004. A strategic and welfare theoretic analysis of free trade areas.

Journal of International Economics 64, 1–27.

Bown, C., 2004a. On the economic success of GATT/WTO dispute settlement. Review of Economics and

Statistics 86, 811–823.

Bown, C., 2004b. Trade disputes and the implementation of protection under the GATT: an empir ical

assessment. Journal of International Economics 62, 263–294.

Bown, C., Crowley, M.A., 2006. Policy externalities: how US antidumping affects Japanese exports to the

EU. European Journal of Political Economy 22, 696–714.

Bown,C.P.,Crowley,M.A.,2013a. Self-enforcing trade ag reements:evidence from time-varying trade policy.

Amer ican Economic Review 103 (2), 1071–1090.

Bown, C.P., Crowley, M.A., 2013b. Import protection, business cycles, and exchange rates: evidence from

the great recession. Journal of International Economics 90 (1), 50–64.

Brander, J., Spencer, B., 1984. Ta riff protection and imperfect competition. In: Kierzkowski, H. (Ed.),

Monopolistic Competition and International Trade. Clarendon Press, Oxford.

Brander, J., Spencer, B., 1985. Export subsidies and market share rivalry. Journal of International

Economics 18, 83–100.

Broda, C.,Weinstein, D., 2006. Globalization a nd the gains from variety. Quar terly Journal of Economics

121 (2), 541–585.

Broda, C., Limão, N., Weinstein, D.E., 2008. Optimal tariffs and market power: the evidence. American

Economic Review 98 (5), 2032–2065.

386 Giovanni Maggi

Brou, D., Ruta, M., 2009. A Commitment Theory of Subsidy Agreements, Mimeo.

Busch, M., Reinhardt, E., 2002. Testing international trade law: empirical s tudies of GATT/WTO dis pute

settlement. In: Kennedy, D.L.M., Southwick, J.D. (Eds.),The Political Economy of International Trade

Law: Essays in Honor of Robert E. Hudec. Cambridge University Press, pp. 457–481.

Cadot, O., de Melo, J., Olarreaga, M., 1999. Reg ional integration and lobbying for tariffs against non-mem-

bers. International Economic Review 40, 635–657.

Calvo-Pardo, H., Freund,C.,Ornelas, E.,2009.The ASEAN free trade agreement:impact on trade ﬂows and

external trade barriers. In:Barro,R.,Lee,J. (Eds.),Costs and Beneﬁts of Regional Economic Integration.

Oxford Univer sity Press.

Chisik, R., 2003. Gradualism in free trade agreements: a theoretical justiﬁcation. Jour nal of International

Economics 59 (2), 367–397.

Clausing, K., 2001. Trade creation and trade diversion in the Canada–United States Free Trade Agreement.

Canadian Journal of Economics 34, 678–696.

Conconi,P.,Perroni,C.,2009. Do credible domestic institutions promote credible international agreements?

Journal of International Economics 79, 160–170.

Conconi, P., Sahuguet, N., 2009. Policymakers’ horizon and the sustainability of international cooperation.

Journal of Public Economics 93, 549–558.

Conconi, P., Facchini, G., Zanardi, M., 2012. Fast-track authority and international trade negotiations.

American Economic Journal: Economic Policy 4, 146–189.

Costinot, A., 2008. A comparative institutional analysis of agreements on product standards. Journal of

International Economics 75 (1), 197–213.

Dekle, R., Eaton, J., Kortum, S., 2007. Unbalanced trade. American Economic Review Paper s and

Proceedings 97 (2), 351–355.

DeRemer, D., 2011. The Evolution of International Subsidy Rules, Mimeo.

Devereux, M.B., 1997. Growth, specialization and trade liberalizaton. International Economic Review 38,

565–585.

Dutt, P., Mihov, I.,Van Zandt,T., 2011. Does WTO Matter for the Extensive and the Intensive Margins of

Trade? M ime o.

Ederington, J., 2001. International coordination of trade and domestic policies. American Economic

Review 91, 1580–1593.

Ederington, J., 2002. Trade and domestic policy linkage in international agreements. International

Economic Review 43, 1347–1367.

Egger, P., Larch, M., 2008. Interdependent preferential trade agreement memberships: an empir ical analysis.

Journal of International Economics 76, 384–399.

Eicher,T.S., Henn, C., 2011. In search of WTO trade effects: preferential trade agreements promote trade

strongly, but unevenly. Journal of International Economics, 83 (2), 137–153.

Estevadeordal, A., Freund, C., Ornelas, E., 2008. Does regionalism affect trade liberalization toward non-

members? The Quarterly Journal of Economics 123 (4), 1531–1575.

Ethier,W.J., 1998. Regionalism in a multilateral world. Journal of Political Economy 106, 1214–1245.

Facchini, G., Silva, P.,Willmann, G., 2013. The customs union issue: why do we observe so few of them?

Journal of International Economics 90 (1), 136–147.

Freund,C., 2000a. Different paths to free trade: the gains from regionalism. Quarterly Journal of Economics

115, 1317–1341.

Freund, C., 2000b. Multilateralism and the endogenous formation of preferential trade agreements. Journal

of International Economics 52 (2), 359–376.

Furusawa ,T., 1999.The Negotiation of sustainable tariffs. Journal of International Economics 48, 321–346.

Furusawa,T., Konishi, H., 2007. Free trade networks. Journal of International Economics 72, 310–335.

Furusawa ,T., Lai, E.L.-C., 1999. Adjustment costs and gradual trade libealization. Journal of International

Economics 49, 333–361.

Gawa nde, K., Krishna, P., Robbins, M .J., 2006. Foreign lobbying and US trade policy. The Review of

Economics and Statistics 88 (3), 563–571.

Goldberg, P., Maggi, G., 1999. Protection for sale: an empirical investigation. American Economic Review

89 (5), 1135–1155.

Goyal, S., Joshi, S., 2006. Bilateralism and free trade. International Economic Review 47, 749–778.

International Trade Agreements 387

Grossman, G., Helpman, E., 1994. Protection for sale. American Economic Review 84 (4), 833–850.

Grossman,G.,Helpman,E.,1995a.Trade war s and trade talks. Journal of Political Economy 103 (4),675–708.

Grossman, G., Helpman, E., 1995b. The politics of free-trade agreements. American Economic Review 85,

667–690.

Handley, K., 2012. Exporting Under Trade Policy Uncertainty: Theory and Evidence. Mimeo.

Handley, K., Limão, N., 2012. Trade and Investment under Policy Uncertainty:Theory and Fir m Evidence,

NBER WP 17790.

Handley, K., Limão, N., 2013. Policy Uncer tainty, Trade and Welfare: Theory and Evidence for China,

Mimeo.

Hor n, H., 2006. National treatment in the GATT. Amer ican Economic Review 96 (1), 394–404.

Hor n,H.,Maggi,G.,Staiger,R.W.,2010.Trade agreements as endogenously incomplete contracts. American

Economic Review 100 (1), 394–419.

Jackson, J.H., 1997. The WTO dispute settlement under standing–misunderstandings on the nature of legal

obligation. American Journal of International Law 91 (1), 60–64.

Jackson, J.H., 2006. Sovereignty, the WTO, and changing fundamentals of international law. Cambridge

University Press.

Johnson, H.G., 1953. Optimum tariffs and retaliation. Review of Economic Studies 21 (2), 142–153.

Karacaovali, B., Limão, N., 2008. The clash of liberalizations: preferential vs. multilateral trade liberalization

in the European Union. Journal of International Economics 74 (2), 299–327.

Kennan, J., Riezman, R., 1990. Optimal tariff equilibria with customs unions. Canadia n Jour nal of Eco-

nomics 23, 70–83.

Kreinin, M.E., 1961. Effect of tariff changes on the prices and volume of imports. American Economic

Review 51, 310–324.

Krishna, P., 1998. Regionalism and multilateralism: a political economy approach. Quar t erly Journal of

Economics 113, 227–251.

Krishna, P., 2003. Are reg ional trading partners natural? Journal of Political Economy 111, 202–226.

Krishna, P., Mitra, D., 2005. Reciprocated unilateralism in trade policy. Journal of International Economics

65 (2), 461–487.

Krugman, P., 1980. Scale economies, product differentia tion, and the pattern of trade. The American Eco-

nomic Review 70 (5), 950–959.

Krugman, P., 1991. Is bilateralism bad? In:Helpman,E, Razin,A. (Eds.),InternationalTrade andTrade Policy.

MIT Press, pp. 9–23.

Krugman, P., 1997. What should trade negotiators negotiate about? Journal of Economic Literature 35,

113–120.

Lapan, H.E., 1988. The optimal tariff, production lags, and time consistency. American Economic Review

78 (3), 395–401.

Lee, G.M., 2007. Trade agreements with domestic policies as disguised protection. Journal of International

Economics 71 (1), 241–259.

Lee, G.M., 2011. Optimal International Agreement and Treatment of Domestic Subsidy, Mimeo.

Levy, P.I., 1997. A political-economic analysis of free-trade agreements. American Economic Review 87,

506–519.

Limão, N., 2005. Trade policy, cross-border externalities and lobbies: do linked agreements enforce more

cooperative outcomes? Journal of International Economics 67 (1), 175–199.

Limão, N., 2006. Preferential trade agreements as stumbling blocks for multilateral trade liberalization: evi-

dence for the US American Economic Review 96, 896–914.

Limão, N., 2007. Are preferential trade agreements with non-trade objectives a stumbling block for multi-

lateral liberalization? Review of Economic Studies 74, 821–855.

Limão, N., Maggi, G., 2013. Uncertainty and Trade Agreements, Mimeo.

Limão,N.,Saggi,K.,2008.Tariff retaliation versus ﬁnancial compensation in the enforcement of international

trade agreements. Jour nal of International Economics 76 (1), 48–60.

Limão, N., Tovar, P., 2011. Policy choice: theory and evidence from commitment via international trade

agreements. Journal of International Economics 85 (2), 186–205.

Liu, X., 2009. GATT/WTO promotes trade strongly: sample selection and model speciﬁcation. Review of

International Economics 17 (3), 428–446.

388 Giovanni Maggi

Liu, X., Ornelas, E., 2012. Free Trade Areas and the Consolidation of Democracies, Mimeo.

Ludema, R.D., 2001. Optimal international trade agreements and dispute settlement procedures. European

Journal of Political Economy 17 (2), 355–376.

Ludema, R.D., Mayda, A.M., 2009. Do countries free ride on MFN? Journal of International Economics

77, 135–150.

Ludema,R.D.,Mayda,A.M.,2010. DoTerms-of-Trade Effects Matter forTrade Ag reements? Evidence from

WTO Countries, M imeo.

Ludema R.D., Mayda, A .M., Mishra, P., 2010. Protection for Free: The Political Economy of US Tariff

Suspensions, Mimeo.

Magee, C., 2008. New measures of trade creation and trade diversion. Journal of International Economics

75, 340–362.

Maggi, G., 1994. Essays onTrade Policy and International Institutions under Incomplete Information. Ph.D.

Dissertation, Stanford University.

Maggi,G.,1999.The role of multilateral institutions in international trade cooperation. American Economic

Review 89 (1), 190–214.

Maggi, G., Rodriguez-Clare,A., 1998. The value of trade agreements in the presence of political pressures.

Journal of Political Economy 106 (3), 574–601.

Maggi, G., Rodriguez-Clare, A., 2007. A political-economy theory of trade agreements. American

Economic Review 97 (4), 1374–1406.

Maggi, G., Staiger, R.W., 2011. The role of dispute settlement procedures in international trade agreements.

Quarterly Journal of Economics 126, 475–515.

Maggi, G., Staiger, R.W., 2012. Optimal Design of Trade Agreements in the Presence of Renegotiation,

Mimeo.

Maggi, G., Staiger, R.W., 2013. Trade Disputes and Settlement, Mimeo.

Martin, A.,Vergote,W., 2008. On the role of retaliation in trade a greements. Journal of International Eco-

nomics 76 (1), 61–77.

Martin, P., Mayer, T., Thoenig, M., 2008. Make trade not war? Review of Economic Studies 75 ( 3),

865–900.

Martin,P.,Mayer,T.,Thoenig,M.,2012.The geography of conﬂicts and reg ional trade agreements.American

Economic Journal: Macroeconomics 4 (4), 1–35.

Mayer,W., 1981. Theoretical considerations on negotiated tariff adjustments. Oxford Economic Pa per s 33

(1), 135–153.

McCalman, P., 2002. Multilateral trade negotiations and the most favored nation clause. Journal of Interna-

tional Economics 57 (1), 151–176.

McLaren,J., 1997. Size, sunk costs, and judge Bowker’s objection to free trade. American Economic Review

87 (3), 400–420.

McLaren, J., 2002. A theory of insidious regionalism. Quarterly Journal of Economics 117, 571–608.

Milner, H., Rosendorff , P., 1996. Trade negotiations, information and domestic politics: the role of domestic

groups. Economics and Politics 8 (2), 145–189.

Milner, H., Rosendorff, P., 2001. The optimal design of international institutions: uncertainty and escape.

International Organization 55 (4), 829–857.

Mitra,D.,2002. Endogenous political organization and the value of trade agreements. Journal of International

Economics 57 (2), 473–485.

Mrazova, M., 2011. Trade Agreements When Proﬁts Matter, Mimeo.

Mrazova,M.,Vines,D.,Zissimos,B.,2013. Is theWTOArticle XXIV bad? Journal of International Economics

89 (1), 216–232.

Or nelas, E., 2005a. Rent destruction and the political viability of free trade agreements. Quarterly Journal

of Economics 120, 1475–1506.

Or nelas, E., 2005b. Endogenous free trade agreements and the multilateral trading system. Journal of Inter-

national Economics 67, 471–497.

Or nelas, E., 2005c. Trade creating free trade areas and the undermining of multilateralism. European Eco-

nomic Review 49, 1717–1735.

Or nelas, E., 2008. Feasible multilateralism and the effects of regionalism. Journal of International Economics

74, 202–224.

International Trade Agreements 389

Ossa, R., 2011. A new trade theory of GATT/WTO negotiations. Journal of Political Economy 119 (1),

122–152.

Ossa,R.,2012. Proﬁts in the new trade approach to trade negotiations. American Economic Review,Paper s

and Proceedings 102 (3), 466–469.

Ossa, R., 2013. Trade Wars and Trade Talks with Data, Mimeo.

Panagariya,A., Krishna, P., 2002. On necessarily welf are-enhancing free trade areas. Journal of International

Economics 57, 353–367.

Park, J.H., 2000. International trade agreements between countries of asymmetric size. Journal of

International Economics 50 (2), 473–495.

Park, J.H., 2011. Enforcing international trade agreements with imperfect private monitoring. Review of

Economic Studies 78 (3), 1102–1134.

Pauwelyn, J., 2008. Optimal Protection of International Law: Navigating between European Absolutism

and A mericanVolunteerism. Cambridge University Press.

Reinhardt, E., 2001. Adjudication without enforcement in GATT disputes. Journal of Conﬂict Resolution

45 (2), 174–195.

Richardson, M., 1993. Endogenous protection and trade diversion. Journal of International Economics 34

(3–4), 309–324.

Romalis, J., 2007. NAFTA’s and CUSFTA’s impact on inter national trade. Review of Economics and

Statistics 89, 416–435.

Rose, A., 2004a. Do we really know that the WTO increases trade? Amer ican Economic Review 94,

98–114.

Rose, A., 2004b. Do WTO members have more liberal trade policy? Journal of Inter national Economics

63 (2), 209–235.

Rose,A., 2013. The March of an Economic Idea? Protectionism Isn’t Counter-Cyclic (Anymore), Mimeo.

Saggi, K., 2006. Preferential trade agreements and multilateral tariff cooperation. International Economic

Review 47 (1), 29–57.

Saggi, K., Sara, N., 2008. National treatment at the WTO: the roles of product and country heterogeneity.

International Economic Review 49, 1367–1396.

Saggi, K.,Yildiz, H.M., 2010. Bilateralism, multilateralism, and the quest for global free trade. Journal of

International Economics 81, 26–37.

Saggi, K.,Woodland,A.,Yildiz, H.M., 2013. On the relationship between preferential and multilateral trade

liberalization: the case of customs unions. American Economic Jour nal: Microeconomics 5 (1), 63–99.

Schwartz, W.F., Sykes, A.O., 2002. The economic structure of renegotiation and dispute resolution in the

WTO-GATT system. Jour nal of Legal Studies 31, 179–204.

Segal, I.R.,Whinston,M.D., 2002.The Mirrlees approach to mechanism design with renegotiation. Econo-

metr ica 70 ( 1), 1–45.

Seidmann, D., 2009. Preferential trading a rrangements as strateg ic positioning. Jour nal of International

Economics 79, 143–159.

Sovey,A., 2012. States Held Hostage: Political Hold-Up Problems in International Politics, Mimeo.

Staiger, R.W., 1995a. International rules and institutions for trade policy. In: Grossman, G.M., Rogoff, K.

(Eds.), Handbook of International Economics, vol. 3. North-Holland,Amsterdam.

Staiger,R.W.,1995b. A theory of gradual trade liberalization. In: Deardorff,A.,Levinsohn,J.,Stern,R. (Eds.),

New Directions in Trade Theory. University of Michigan Press.

Staiger, R.W., Sykes,A.O., 2011. InternationalTrade,NationalTreatment and Domestic Regulation. Journal

of Legal Studies 40 (1), 149–203.

Staiger, R.W.,Tabellini, G., 1987. Discretionary trade policy and excessive protection. American Economic

Review 77 (5), 823–837.

Staiger, R.W.,Tabellini, G., 1999. Do GATT rules help governments make domestic commitments? Eco-

nomics and Politics 11 (2), 109–144.

Stoyanov, A., 2009. Trade policy of a free trade agreement in the presence of foreign lobbying. Journal of

International Economics 77 (1), 37–49.

Subramanian,A.,Wei, S.J., 2007. The WTO promotes trade, strongly but unevenly. Journal of International

Economics 72, 151–175.

390 Giovanni Maggi

Tang, M .K.,Wei, S.J., 2010. The value of making commitments externally: evidence from WTO accessions.

Journal of International Economics 78 (2), 216–229.

Tomz, M., Goldstein, J.L., Rivers, D., 2007. Do we really know that the WTO increases trade? Comment.

Amer ican Economic Review 97, 2005–2018.

Tornell, A., 1991. Time inconsistency of protectionist programs. Quarterly Journal of Economics 106,

963–974.

Tovar, P., 2011. Lobbying costs and trade policy. Journal of International Economics 83 (2), 126–136.

Treﬂer, D., 2004. The long and short of the Canada–US free trade ag reement. American Economic Review

94, 870–895.

Venables, A.J., 1985. Trade and trade policy with imperfect competition: the case of identical products and

free entry. Journal of International Economics 19, 1–20.

Venables,A.J., 1987.Trade and trade policy with differentiated products: a Chamberlinian-Ricardian model.

Economic Journal 97, 700–717.

Watson,J.,Klimenko,M.M.,Ramey,G.,2008. Recurrent trade agreements and the value of external enforce-

ment. Journal of International Economics 74 (2), 475–499.

Whalley,J.,1998.Why do countries seek regional trade agreements? In:Frankel,J. (Ed.),The Regionalization

of the World Economy. University of Chicago Press.

Winters,A.L., Chang,W., 2000. Regional integration and the prices of imports: an empirical investigation.

Journal of International Economics 51, 363–377.

Winters,A.L.,Chang,W.,2002. How regional blocks affect excluded countries: the price effects of Mercusor.

Amer ican Economic Review 92, 889–904.

Yi, S.S., 1996. Endogenous formation of customs unions under imperfect competition: open regionalism is

good. Journal of International Economics 41, 153-177.