wPs .306
POLICY RESEARCH WORKING PAPER            3 061
Reciprocity in Free Trade Agreements
Caroline Freund
The World Bank
Development Research Group
Trade
May 2003



I POLICY RESEARCH WORKING PAPER 3061
Abstract
Freund uses detailed trade, tariff, and income data for     agreements because the value of a developing country
countries involved in 91 trade agreements negotiated        tariff preference in terms of its effect on trade
since 1980 to test for reciprocity in free trade            preferences from a rich country is quite small. The gains
agreements. The results offer strong evidence of            from unilateral liberalization are likely to far outweigh
reciprocity in North-North and South-South free trade       potential gains from using protection as a bargaining
agreements, but there is little empirical support for       chip in trade negotiations.
reciprocity in North-South trade agreements. In               The evidence is consistent with a repeated game model
particular, after controlling for other determinants of     of trade liberalization. The model presented shows that
trade preferences, the results suggest that a one percent   trade preferences granted are increasing in trade
increase in preferences offered leads to about a one-half   preferences received. This implies that countries can
of a percent increase in preferences received in North-     extract greater concessions from trade agreement
North and South-South trade agreements. Freund also         members if they have higher external trade barriers.
finds evidence that large countries extract greater trade   However, if a country's trade barriers are very large then
concessions from small countries. This leads to a           the gains from reneging on the agreement in the short
modified form of reciprocity in North-South agreements.     run will be high, making the agreement unenforceable
A large increase in access to a developing country market   despite offering long-term gains. So, there is a
leads to only a small increase in access to a rich country  reciprocity-credibility tradeoff. High tariffs may allow
market.                                                     countries to extract more concessions from potential
The results imply that there are incentives for           trade agreement partners, but they also make the country
countries to maintain protection in order to extract more   less credible in actually implementing agreed tariff
concessions from trade partners. But in general, such       concessions. If a country's external tariff is very high
perverse incentives should be less of a concern in          relative to other countries, then it will not be able to
developing countries involved in North-South                commit credibly to any free trade agreement.
This paper is a product of Trade, Development Research Group. Copies of the paper are available free from the World Bank,
1818 H Street NW, Washington, DC 20433. Please contact Paulina Flewitt, room MC3-333, telephone 202-473-2727,
fax 202-522-1159, email address pflewitt@worldbank.org. Policy Research Working Papers are also posted on the Web
at http://econ.worldbank.org. The author may be contacted at cfreund@worldbank.org. May 2003. (34 pages)
The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about
development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The
papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this
paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the
countries they represent.
Produced by Partnerships, Capacity Building, and Outreach



Reciprocity in Free Trade Agreements
Caroline Freund*
The World Bank
* I have benefited from comments by Daniel Lederman, Nuno Limao, Marcelo Olarreaga, John Romalis,
Karnal Saggi, Maurice Schiff, Robert Staiger and participants at presentations at the 2003 AEA meetings in
Washington, the World Bank, the University of Maryland, and the Regional Integration Network
conference in Punta del Este, Uruguay. Correspondence: 1818 H Street, NW, Washington D.C. 20433, e-
mail cfreundeaworldbank.org, ph: (202)473-7102, fax: (202)522-1159.






Reciprocity in Free Trade Agreements
I. Introduction
The two pillars of the GATT are nondiscrimination and reciprocity. Article XXIV of the
GATT, which allows for the formation of trade blocs, has been derided as antithetical to
the GATT because it permits members of a trade bloc to discriminate against
nonmembers. What is less clear, however, is whether article XXIV is in sync with the
other pillar of GATT--reciprocity. The language on reciprocity in the body of the GATT
is clear: governments seek a "balance of concessions" and when presented with the
withdrawal of a trade concession, its trade partner is permitted to withdraw a
"substantially equivalent concession". Article XXIV also includes language that could be
interpreted as pertaining to reciprocity, in that it calls for trade barriers "to be eliminated
with respect to substantially all trade between the constituent territories." Thus, by
definition, preferential trade agreements involve some degree of reciprocity because both
sides are expected to make full trade concessions. But, unlike traditional multilateral
negotiations, this does not necessarily yield equivalent concessions since an agreement
can involve members of various sizes with vastly different trade barriers, yielding gains
in market access that are far from symmetric. In addition, some sensitive sectors are
typically excluded, and many other types of trade barriers, such as antidumping claims or
technical standards can remain in place, or even increase to offset tariff concessions.
The purpose of this paper is to examine theoretically and empirically what role
reciprocity has played in regional integration agreements. The application of reciprocity
in multilateral tariff negotiations has strong theoretical foundations. In a series a papers,
Bagwell and Staiger show that terms-of-trade motives provide countries with incentives
2



to have positive tariffs, creating a prisoner's dilemma, whereby all countries would be
better off if they could cooperate and reciprocally lower tariffs.1 They argue that the
articles of the GATT offer negotiating rules that help governments undo inefficient trade
restrictions generated by the terms-of-trade externality. In other words, reciprocal tariff
reduction allows countries to credibly commit to lower tariffs and reach a higher welfare
level. Finger, Reincke, and Castro (1999) find some evidence of reciprocity in terms of
tariff cuts offered in the Uruguay Round negotiations.2
The value of reciprocity in regional agreements is less clear. Unlike in
multilateral negotiations, where reciprocity enhances overall trade liberalization,
reciprocity in regional agreements (by definition) furthers discriminatory tariff reduction.
Reciprocity may be especially damaging in North-South agreements, where asymmetries
in size suggest that low-income countries will have to make relatively larger trade
concessions to achieve an agreement with a high-income country. A need for reciprocity
also implies that some agreements should be infeasible; for example, a large country
would gain too little from a free trade agreement with a very small low-tariff country to
make the agreement worthwhile. Finally, reciprocity could provide incentives for low-
income countries to maintain higher trade barriers in order to obtain preferences from
high-income countries, and as a result generate greater trade diversion.
In this paper, we first examine the theoretical foundations for reciprocity in free
trade agreements. We follow Bagwell and Staiger and use a repeated game model to
analyze the question of what types of regional agreements are sustainable. In order to
achieve a trade agreement, countries must have an incentive to sign the agreement and
I See Bagwell and Staiger (2000) for a summary of the literature.
2 About 30 percent of concessions given are matched by concessions received.
3



the agreement must be self-enforcing. Using an oligopolistic model of trade, we show
that it is easier to form an agreement with a large country, a high-cost country, or with
one that has high tariffs. However, we also find that a country's credibility in enforcing
the agreement is decreasing in its tariff level. Thus, there is a reciprocity-credibility
trade-off; larger tariff reductions at home yield larger trade concessions abroad, but larger
tariff cuts are less likely to be self-enforcing. The model also predicts that reciprocity
should be more important in agreements between countries with similar cost structures.
The intuition is that when costs differ, in addition to enhancing competition, a
preferential agreement shifts production to the low-cost country, enhancing overall
welfare gains and making the agreement easier to sustain.
To examine reciprocity empirically, we use three measures of trade preferences.
The first uses detailecl tariff and trade data to calculate the bilateral trade-weighted tariff
for each country pair in a trade agreement. The second uses the same data to calculate
the tariff revenue that will be lost as a result of a trade agreement; it is effectively a
measure of the gain in producer surplus in the exporting country. The third focuses on
changes in market access subsequent to the formation of a trade agreement, where market
access is measured as changes in trade intensity indices (essentially trade shares adjusted
for income growth) following a regional agreement. Using all three measures, we find
that the trade preferences granted in a regional agreement are highly correlated across
country pairs. Controlling for relative country size and separately also for country fixed
effects, we find that preferential access is the single most important variable determining
preferences granted. Blecause of issues of simultaneity, we also instrument for
preferences with country-fixed effects. The results are robust to this innovation.
4



We also examine whether reciprocity in North-South agreements is different from
reciprocity in other agreements. The model implies that countries with similar costs
structures are more likely to require reciprocity to be self-enforcing. The variation in
costs is likely to be greater when agreements are formed between North and South
countries, suggesting that reciprocity would be less important in North-South agreements.
In addition, work by John Whalley and various coauthors shows that concessions in
North-South trade agreements, such as NAFTA, tend to be in areas other than tariffs.3 In
particular, the South gains insurance for its access to the North's market, while the North
gains concessions on environment and other non-tariff areas. Similarly, Staiger (1998)
notes that GATT language does not require reciprocity by small countries, suggesting
that reciprocity in preferential agreements between North and South countries is unlikely
to be driven by the same underlying forces as reciprocity in the GATT.4
Our results show little evidence of reciprocity in North-South agreements. In
particular, among North-South partners, preferences in one country are not correlated
with preferences in the other country. There is, however, a modified form of reciprocity;
North countries extract significantly more market access in South countries than South
countries extract from the North. Specifically, a ten percent reduction in the developing-
country tariff yields only about a 2 percent reduction in the rich-country tariff; in
3 Perroni and Whalley (2000) and Abrego et al. (1997).
4 Staiger's comment was made in reference to a paper by Davis and Kowalczyk (1998). They examine
tariff phase-outs in Mexico and United States following NAFTA and look for evidence of reciprocity.
Their main finding is that the length of tariff phase-outs by sector in both Mexico and United States is
increasing in the U.S. tariff. The paper, however, analyzes a very narrow form of reciprocity because we
typically think of reciprocity as being intersectoral. That is, we would expect the main trade concessions
that Mexico gives the United States to be in different sectors from the concessions that the United States
gives Mexico.
5



contrast, a 10 percent reduction in the large country tariff leads to an 33 percent reduction
in the poor country tariff.
The paper is organized as follows. The next section provides the theoretical
framework. Section 3 examines the empirical importance of reciprocity in trade
agreements. Section 4 concludes.
II. The Theoretical Framework
We use a three-country oligopolistic model of trade to evaluate the importance of
reciprocity in free trade agreements. This model is appropriate for analyzing trade bloc
questions because optimal tariffs are non-zero and because regional agreements expand
the members' share of the world market and hence enhance profits abroad. Gaining
preferential market access is an important and often clearly stated purpose of trade
agreements.5 The model is meant to be illustrative and highlight the importance of
country size, market structure, and reciprocity in achieving a free trade agreement.
There is one good, which is produced by a single profit-maximizing firm in each
country and segmented markets lead to trade in this good (as in Brander and Krugman
(1983)). Since large markets are likely to have more firms, the one-firm assumption
would be strong if we did not allow variation in production costs. Lower domestic
production costs and more intense domestic competition will both have similar effects on
trade. They will reduce the extent to which consumers benefit from international trade
and will also depress the profits of foreign firms.
5For example, the U.S. Trade Representative argues that because Canada and Mexico have trade
agreements with other countries in the region "U.S. businesses are losing marketshare. U.S. wheat and
potato farmers, for example, are now losing markets in Chile to Canadian exports" (Robert Zoellick, May
7, 2001)
6



We assume that the inverse demand function, pi(Q), in each country (X, Y, and Z)
is linear. Specifically, Pi (Q') = 1- a'Qt, and Q' = qi + qi + qz, where Q' is the total
quantity consumed in country i, q' is the quantity produced by the firm inj for market i,
and a' varies with the size of the market, smaller countries have larger a's.
Competition is Cournot. Profits to the firm from countryX, in country i, ir, are
(1)   or' = qP' (Q')-cXxq -t ,tq,
where, ci is the constant marginal cost of production in country i, and t' is the tariff that
the firm from countryj faces in market i. Solving for the profit maximizing quantity of a
firm from country X in market i, we have
(2)    qi =      (1-3(cy +tX)+cy +cz +t +tz)
(2)  q,,        ~~~4a'
Equation (2) shows that higher domestic costs reduce quantity, while higher foreign costs
increase quantity. A sufficient condition for each firm to sell in each market is that
t, < (1- 3cx + c, + c,)/ 3, which we assumes holds.
The government's welfare function, W, in country X, is the sum of consumer
surplus, producer surplus, and tariff revenue.
(3)    W = U(Qx) PxQx + E (qIP' -cxq' -t'q')+txq   +tq
x,y,z
If there are no free trade agreements then the government maximizes welfare,
equation (3) over t' and tz, taking other countries' tariffs as given. Under MFN, the
tariff on Y must be equivalent to the tariff on Z, resulting in an optimal tariff of
3 -cs -cy - cz
t =    x    O   . Note that in this model the optimal tariff in one country is not a
10
7



function of the tariffs implemented abroad or of the size of the market (a'). But it is a
function of marginal production costs-higher costs are associated with lower tariffs in
all countries. Regardless of what policy other countries follow, a single country is always
better off with a positive tariff. However, if each country installs its optimal tariff then
all countries are worse off than they would be at free trade. Each country is made better
off if some reciprocal bilateral or multilateral tariff reduction is achievable.
III.  A Free Trade Agreement
Two conditions must be met in order for countries to participate in a free trade
agreement. First, an incentive constraint must be satisfied, the agreement must make
countries better off. Second, the agreement must be self-enforcing, long-run gains must
make it worthwhile for countries to commit to the agreement, as opposed to maintaining
tariffs.
Incentive constraint
A free trade agreement will only be approved if the welfare of the member
countries is improved. That is, it must be the case that welfare after the bilateral tariff
reduction exceeds welf are with ex ante tariffs. Using the welfare function above,
substituting quantities in from Equation (2) and solving yields the incentive constraint for
an agreement between countryXand country Y. Specifically the constraint is:
(-6+18cy +2c - 14c2)tx _tx2   6(1-3cx +,cy +c)ty -3ty2
32aX                        16ay
The first term in equation (4) represents the net domestic loss that results from the decline
in a country's own tariff, the increase in consumer surplus and the loss in domestic profits
8



and tariff revenue. The second term is the gain in profits abroad that results from the
decline in the foreign tariff. An agreement is more likely to be welfare improving when
ax is large and a' is small-i.e. the small country gains more from the agreement. In
addition, it is more welfare enhancing when the cost in the other member country, cy, is
relatively large-that is, the low-cost country gains more from the agreement. A low
domestic tariff, t, and a high partner-country tariff, tP, also increase the value of the
agreement. The incentive constraint will always be satisfied for both countries, when
countries are similar in size, cost structure, trade policy, and the extent of the tariff
reduction-as is more likely to be the case in a North-North or a South-South agreement.
Alternatively, the incentive constraint may be satisfied for both parties if the smaller
country or the low-cost country offers greater trade concessions.
Sustainability
We use a repeated game framework to evaluate the extent to which a trade
agreement between two countries is self-enforcing.6 In order for a bilateral agreement to
be sustainable, the welfare gain from cheating on the agreement and then returning to the
MFN tariff equilibrium forever must not exceed the welfare level from committing to the
agreement. We consider cheating and punishment to consist of failing to install the
preferences (i.e. maintaining status quo tariff) and returning to the ex ante tariff level
forever. Alternatively, we could use the Nash tariff level as punishment, which would
further the extent to which tariff reduction is achievable since the punishment phase
would be more severe. However, since countries are constrained by the WTO, using the
more conservative status quo punishment is more realistic. Specifically,
6 See Freund (2000) for more details on the repeated game analysis in this type of model.
9



(4)    Wdevite +  W        <I
(+1 - *punishment - 1 a Wbilateral
Where 3 is the discount rate. We can solve this equation for maximum size of the tariff
cut that is achievable between any two partners.
Using the welfare function above and substituting quantities in from Equation (2)
and solving yields the cutoff discount rate (8 ) such that a free trade agreement is
feasible.
(   6-18cy --2c, +14C, +tx  ayt_   a
6-18c5 4 6cy +6cz -3ty  2axtY
Equation (5) implies that the cutoff discount rate that X needs to achieve an agreement
with Y is lower, i.e. commitment is easier, if Y's market is relatively large (ar/ax is small),
y is a relatively high cost country (cy is high relative to cr), and the preferential treatment
Y offers is relatively large (t/It is small). The intuition is that if the foreign market is
large, the foreign producer has high costs, and/or the foreign preferential treatment
granted is extensive then preferential market access is worth more and countries have less
incentive to cheat on the agreement.
This implies thiat there is a credibility-reciprocity tradeoff with respect to trade
policy and participation in free trade agreements. Countries have an easier time
committing to free trade agreements when the preferential access offered is relatively
small, but countries can extract greater market access abroad by offering more
preferential treatment in return. This provides an intuition for why some agreements may
never be implemented. Countries offer more to get more, but their concessions are
simply not credible in equilibrium.
10



To examine the set of possible agreements we rewrite equation (5) in terms of the
cutoff tariff level needed to make an agreement feasible.
(6)   tX< 6-18c, +6cy +6c, -3ty   2axtY
6 -18cy -2c, +1l4cz + t'  aY
If tariffs are low, the first term will be determined largely by cost structure and we can
rewrite equation (6) as
2axtY
Equation (7) says that the maximum tariff level such that an agreement is feasible is
nearly proportional to the tariff level abroad, the market structure, the relative size of the
two countries, and the discount rate. Taking logs of both sides yields
ax
(8)   In t' < In(M) + n(f(cx, cy )) + n(- +In t y=t x
By symmetry, a similar equation will also maintain for country Y.
(9)   In tY <- ln(245) + ln(f (cy,o) +,  n7)- )+ Intx = ty
a
Equation (9) implies that there is a range of preferences for each country such that
agreements are feasible. Figure 1 presents this range graphically. Country X's constraint
shows the minimum tariff preference that country Y must grant country X for a given
level of preference in X's market. Any point above this constraint is feasible from
country X's perspective since it gets more from country Y than is required for the
agreement to be self-enforcing. Similarly, country Y's constraint shows the minimum
preference that Y requires from X, for any given level of preference Y offers. From Y's
perspective, any point to the right of country Y's locus is feasible. The area between the
11



two lines displays the range of feasible agreements, and points A, B and C all represent
feasible agreements. In contrast, the agreement represented by point D is not feasible
from Y's perspective since the preferential treatment that X offers Y is not enough to
justify granting such large preferences to X in return.
The model shows that preferences granted will be increasing in the level of
preferences received. However, small changes in preferences granted may have no effect
on the level of preferences received, provided both constraints are satisfied. For example,
points A and B in Figure 1 represent two feasible agreements, but point B is better for
country X than point A because it gets more in return. This implies that point B might be
the more likely agreemnent if X has more bargaining power, and point A the outcome if Y
has more bargaining rower.
The closer the enforceability constraints are to each other, the smaller is the range
of feasible agreements and the more important is reciprocity in determining outcomes. If
the constraints are very far apart then reciprocity may not be as important in determining
preferences; outcomes will depend much more on other things, such as country
characteristics, relative bargaining power, and political issues.
Equations (8) and (9) imply that there is a range of tariff pairs where agreements
are feasible, and the range is increasing in the discount rate. If 8 is very small, then it is
possible that there are no agreements that are self-enforcing, as shown in Figure 2. The
intuition is that countries care very little about the future so they will cheat on any
agreement.
The distance between the constraints also depends on relative costs. The more
similar are the costs the more likely it is that the constraints are close to each other. To
12



see this, note that country X faces a constraint with f(cx,cy) and by symmetry, country Y
faces a constraint f(cy,c,). For feasible values of c,,, cy and c_,  x is negative, a  is
acx            acy
positive, and   <      . This implies that when costs structures are different, the
ac,   &cy
constraints are more likely to be far apart, all else equal. The intuition is that when costs
are different, a regional agreement not only increases production but also shifts more to
the low-cost producer, which magnifies the output effect. This cost-shifting effect makes
the agreement more welfare improving and hence easier to enforce.
By contrast, market size does not effect the extent to which the constraints bind
t = _  ; x .Relative to countries of an equal size, the small-country constraint shifts
out, indicating that it can support more concessions for a given level of preferences
received. The large country constraint shifts in by an identical amount, indicating that it
can grant less preferences for a given level of preferences received. Thus, both
constraints shift, but the distance between them is unaffected.
IV. Empirical Analysis
In this section, we test the extent to which reciprocity is important in determining
preferences offered in free trade agreements. The model shows that reciprocity plays a
critical role in determining the set of feasible regional agreements. This role is
heightened if the feasibility constraints are close together, as in Figure 3, where more
preferential treatment by one party must be reciprocated in order for an agreement to be
13



sustainable. In this case, reciprocity determines actual outcomes. In contrast, Figure 4
shows a case where reciprocity is important in determining the set of feasible agreements,
but where reciprocity has little influence on precise outcomes. The circles represent the
space of possible agreements, and the filled circles are actual agreements. In this case,
reciprocity determines which agreements are feasible, but observed outcomes are more
closely related to other variables. If the incentive constraints tend to be close together
then we should find that reciprocity is important in regional agreements.
To look for evidence of reciprocity we use three measures of actual trade
preferences that counties grant one another. First, using detailed trade and tariff data, we
calculate ex ante trade-weighted average bilateral tariffs. Second, we use tariff data and
exports before the agreement to estimate lost tariff revenue. The second measure also
captures the transfer of tariff revenue from the importing country to producer surplus in
the exporting country. Third, we create an index of the change in market share from the
agreement. This measures the extent to which the regional agreement led to increased
bilateral trade in each country. More specific descriptions of each measure follow.
The most basic measure of preferences is the trade-weighted tariff.
(7)    tariWfj = Y    J.
k         exp)orts,
where MFN; is the tariff in countyj on industry k, exports are exports from i toj in
industry k, and exports, are total exports. Bilateral tariffs are available from the
TRAINS database and total bilateral exports are from the IMF International Financial
Statistics.
14



Using this measure, the triangles in Figure 5 represent the preferences offered by
one party of a trade agreement versus the preferences received. The scatter plot offers
some evidence that reciprocity is important-the bilateral concessions of one trade
partner seem to be increasing in the concessions granted by the other partner-even
without controlling for country size and income levels. As a control, the diamonds show
preferences of each country in an agreement relative to the world, where preferences are
measured as the trade-weighted tariff via the world. Therefore, the diamonds represent
preference pairs if regional agreement members granted each other the preference that an
exporter with the world export structure would receive. The diamonds appear to be
uniformly distributed within a range, suggesting that if bilateral agreements involved
tariffs reductions to an average world partner then reciprocity would not be important.
The graph suggests that countries in regional agreements tend to offer bilateral
preferences that are more similar to each other than they are to the rest of the world.
One problem with the simple trade-weighted tariff is that it does not capture the
relative importance of an export market. For example, the U.S. and Canadian trade-
weighted tariffs on Mexican imports could be very similar, but since the United States is
a much larger market for Mexican goods, U.S. preferences are worth more to Mexico.
This relates to the notion in the GATT that reciprocity involves trade policy affecting an
equivalent amount of trade. Including GDP in the equation will control for this to the
extent that it is only market size that matters. But, it could also be tastes, proximity, and
endowments.
The second measure controls for this market relevance effect. It is a measure of
gain in producer surplus in the exporting country, or the forgone tariff revenue in the
15



importing country. It is the bilateral tariff in a six-digit sector multiplied by the trade in
that sector, summed across all sectors. It is calculated as follows:
(8)     producer surplus, = tariffy * exportsV,
It is thus the amount of tariff revenue that the importing country will forgo collecting as a
result of the trade agreement, assuming that all tariffs are removed.7 To the extent that
prices remain roughly unchanged, it also represents a gain in producer surplus to the
exporting country. Hence, the measure will be increasing in exports and also in tariffs,
provided the elasticity of trade to tariffs does not exceed one in absolute value.
While these measures provide a good proxy for the extent to which firms gain
from liberalization in the other country, and hence how balanced the agreement is, there
are some problems with both measures. First, as previously noted, not all tariffs are
necessarily removed when the agreement is signed, so they might overstate the gains in
some cases and perhaps fail to show reciprocity, even though reciprocity is actually
present. Second, tariffs are not the only trade barriers. Third, the data required to
calculate this measure are only available for 52 agreements concluded after 1989. Still, if
agreements at least approach genuine free trade agreements and tariffs are the primary
barriers that are negotiated, then only agreements that offer similar return should be
signed. In this case., this measure should roughly capture the extent of reciprocity.
The third measure focuses on increases in trade to capture the extent to which
concessions are actually made. It attempts to estimate the magnitude of all barriers, not
7More specifically, it is calculated as the average trade-weighted tariff in the year before the agreement, or
if that is not available the closest year to that year. We also attempt to match as closely as possible the year
of the tariff data in the partner countries. The trade weighted tariff is then multiplied by average trade in
the four-year period before the agreement was concluded in order to control for possible anomalies in trade
flows in a particular year.
16



just tariffs, that are lowered. It uses data on bilateral trade among members of 91
regional agreements negotiated between 1980 and 1999 to estimate changes in market
share, where market share is estimated using the so-called "export intensity index", Ijj.
(6)   Ivj = X 9
XROW,j
The numerator in Equation (6) isj's share of i's exports, and the denominator isj's share
of the rest-of-the-world's exports.8 The export intensity index describes how much i
exports toj relative to how much the rest of the world exports toj. An export intensity
index of unity implies thatj's share of i's exports is identical toj's share of rest-of-the-
world exports. The important feature of this index is that, with constant income elasticity
of trade across countries, it will not increase as a result of regional income growth in the
importing country. This is becausej's share of i's exports (the numerator of the index)
increases in exactly the same proportion asj's of world exports (the denominator of the
index). This measure of market access is defined as the average intensity in the four-year
period following the agreement less the average intensity in the four-year period
preceding the agreement and denoted it as AIij. Trade intensity indices are constructed
from the fo.b. bilateral trade data recorded in the IMF Direction of Trade Statistics.
For the purposes of computing all three indices, for the members of a new union
(eg MERCOSUR), we examine all bilateral pairs; for countries acceding to a well-
established union (eg. Austria to the EU), we consider only exports from the new member
to the union and vice versa, not each pair of countries. The intuition is that negotiations
for a new union take place at a country-to-country level, but negotiations between a
8 See Anderson and Norheim (1993) for a detailed description of trade intensity indices.
17



country and a union take place at a country-to-union level. Table 1 in the appendix lists
all of the agreements used for each measure.
In the first stage, we look at partial correlations between tgy and tji, PSi and PSji
and AIij and AIji. In addition to measures of preferences, theory suggests that preferences
granted should be a function of country size and market structure. Bigger countries
should get a relatively greater increase in market access abroad since the value of their
tariff reduction to other countries is worth relatively more. However, if there are
increasing returns to scale, or bigger countries simply have lower cost (or more) finns,
then the market-size effect will be dampened. The basic regression equation that we
estimate, for each measure, is:
(7)    preferencey, = a + /1preferencej, + fl2gdpi + /33gdp1 + ,
where preference is one of the three measures (ln(tariff), ln(producer surplus), or the
change in the intensity index), and small letters denote the natural log of the variables.
Theory predicts that ,B should be positive; an increase in preferences granted should lead
to an increase in preference received. Indeed, with perfect reciprocity the coefficient
should be unity. The null hypothesis is that J1 is zero-representing the case where the
gains to one country are uncorrelated with the gains to the other country. This would be
the self-enforcing constraints never bind, i.e. the gap between them is very wide, and
agreements are much more a function of other factors. For example, countries may have
varying bilateral trade barriers, join trade agreements at random, and give full trade
concessions. In this case, countries would get greater concessions from an agreement
with a high-tariff country and less from an agreement with a low-tariff country. Since
agreements are chosen randomly there is no reason to think that the preferences measures
18



should be correlated-country-fixed effects (used below) should account for most of the
variation in the measure of preferences.
The signs of 32 and 013 depend on the relative importance of market size effects.
If market size is very important then 02 should be positive and f3 should be negative-
and they should be of a similar magnitude. If costs (and extent of domestic competition)
play an important role then the signs of the coefficients is ambiguous.9
One econometric problem with this regression is that errors are likely to be
correlated across bilateral pairs. Each agreement enters as two observations, for example,
CUSFTA enters once with the US preferences in Canada as a dependent variable and
once with Canadian preferences in the US as a dependent variable. We correct the errors
for pairwise correlation as well as heteroskedasticity using Rogers (1993) and White
(1980), respectively.
The results are reported in the columns (1), (4), and (7) of Table 1. All of the
signs are as predicted, but only the coefficient on the preference variable is significant in
all of the specifications at standard confidence levels. In particular, the coefficient of
about.6 on the tariff variable implies that a 10 percent bilateral tariff cut in one country
leads to about a 6 percent cut in the other country's tariff. The coefficient of .5 on the
producer surplus variable implies that a ten percent increase in tariff revenue lost leads to
a 5 percent increase in producer surplus gained. The coefficient of .3 on the share
variable implies that a 10 percent increase in i's access toj's market is associated with
about a 3 percent increase inj's access to i's market. These estimates are somewhat
9 We also try including log population in the regression equation, to see if level of development is
important, but it was not significant and the coefficients on the preferences were unchanged (not reported).
19



higher than estimates by Finger, Reincke, and Castro (1999) for tariff concessions in the
Uruguay Round; they find that 29 percent of concessions were reciprocated.
The coefficient on GDPi and GDPj in the tariff and intensity indices regressions
are positive and negative, respectively, as market size effects would predict-implying
that the market access; that a country receives for a given level of preferences granted is
increasing in its relative size-but the coefficients are not significant.'0
Columns (2), (5) and (7) of Table 1 augment the basic specification to include
importer and exporter country-fixed effects, instead of GDP. Country-fixed effects will
capture the extent to w(hich certain countries always give/receive the same access, due to
market size and market structure effects. While these turn out to be important
determinants of preferences, they do not appear to be correlated with reciprocity since the
coefficients on preferenceji remain almost unchanged in each of the three cases."l On
average, the R-squares in the regressions suggests that about half of the explained
variation is due to reciprocity and about half is due to country fixed effects. The
remaining variation among agreements could be a result of bargaining, mis-measurement,
or other agreement issues that are not in the model such as labor, the environment or
investment treaties.
This simple regression implies that the null hypothesis-there is no correlation
between preferences granted and preferences received-can be eliminated. But, there are
'° We also test joint significance and in most cases the coefficients are not jointly significant at standard
levels.
l One additional concern is that the results could be generated by the type of trade agreement; for instance,
some agreements are deeper than others, and as a result might produce greater changes in trade. To some
extent this would be evidence of reciprocity, since countries choose how far to go, but it would be a
different kind of reciprocity--at the bloc level instead of the country level. To test for this, we augment the
analysis to include bloc fixed effects. Again, the results remain robust (not reported), suggesting that
bilateral reciprocity plays an important role in trade agreements.
20



still some econometric issues that need to be addressed. If the model is correct, the
estimated coefficient on preferences is biased because preferences are jointly determined
and because of measurement error. Specifically, since preferences in i are an increasing
function of preferences inj, and vice versa, the coefficient estimate is likely to be
overstated. On the other hand, since we are using tariffs and outcomes to proxy for true
preferential policy, both the dependent and independent variables are likely to be
measured with error. Assuming the measurement error in the dependent and independent
variables are normally distributed and uncorrelated with each other, the error in the
explanatory variable would tend to bias the coefficient on preferences towards zero.
To deal with these problems we use instrumental variables technique.'2 We
instrument for preferences with importer and exporter fixed effects. Country fixed
effects make ideal instruments because they are exogenous and are highly correlated with
preferences. Moreover, they should also be uncorrelated with the measurement error in
preferences. The results are reported in Columns (3), (6) and (9) of Table 1. In each
case, the coefficient on the preference variable is almost unchanged, suggesting that the
two effects roughly offset each other in the OLS regressions.
North-South Agreements: Are Some Agreements Less Equal than Others?
The model shows that agreements where marginal costs are very different are less likely
to require reciprocity. In this section, we consider this prediction by evaluating North-
South agreements, where costs structures are likely to be very different, implying that
reciprocity is less likely to prevail in North-South agreements.
12 The parameter in question is not identified in a simultaneous equations approach.
21



This prediction is also supported by other work on North-South agreements.
Staiger (1998) notes that the notion of reciprocity as intended in the GATT may not carry
over to North-South regional agreements, since even in multilateral negotiations small
countries have not been required to offer reciprocal concessions to large countries. The
1979 Enabling Clause encourages industrialized countries not to seek reciprocal
concessions that are "inconsistent with their individual development, financial, and trade
needs". In addition, industrialized countries may have other motivations for signing free
trade agreements with developing countries. The concessions by developing countries
may not directly relate to trade, but involve issues like investment, technical standards,
intellectual property rights, and competition policy. 13 Additional evidence that such
agreements may be different comes from the number of programs designed to
unilaterally give preferences to developing countries, such as Europe's Everything-But-
Arms initiative and the Africa Growth and Opportunity Act in the United States. On the
other hand, there has also been a recent emergence of agreements that are less one-sided.
The Cotonou Agreement, for example, rescinds the one-way preferences that Europe
gave countries in Africa and the Caribbean and replaces them with free trade agreements.
An alternative to Staiger's argument may be that precisely because small countries have
not been required to make reciprocal concessions in multilateral negotiations, large
countries are using regional agreements to extract concessions.
In this section, we examine whether North-South agreements are different. South
countries are defined as those with per-capita income below $10,000 in 1995. Table 2
reports the results from running the regressions separately on North-South country pairs.
13 See Limao for a model of how such agreements could be negotiated simultaneously with a trade
agreement.
22



For both the tariff and the producer surplus measures there is no evidence of reciprocity
in North-South agreements. However, market share changes are correlated in the OLS
regressions, though this result is not robust for North-South country pairs in the
instrumental variables regression. This could suggest that country-fixed effects are not
good predictors of preferences in North-South agreements. However, results from a
regression of preferences on country fixed effects in North-South agreements have an R-
square of 0.41. The implications of this are not clear. One possibility is that since
preferences are measured with error, there could be a spurious correlation in preferences.
To examine this question in more detail we split each sample according to the size
of the market, and use seemingly unrelated regressions techniques. Specifically, in one
sample the dependent variable is the increased market access the smaller country (the
South) gives to the larger country (North) and the other sample has the opposite. This
specification allows big countries and small countries to have different coefficients on the
preference variable. For example, if the North has most of the bargaining power then for
a given change in access to its market, the North should be able to extract relatively more,
i.e. the coefficient on North access offered should be greater than the coefficient on South
access offered. The results are reported in Table 3. For both the tariff and the market
access measures they suggest that South countries have relatively less bargaining power
than North countries. For example, for a 10 percent increase in the South's access to the
North, the North extracts 15 percent increase in access to the South. In contrast, for a 10
percent increase in access to the South, the South extracts only about a 5 percent increase
in access to the North. The asymmetry can not be a result of a higher average tariffs in
the South than in the North because the constant will pick up average tariffs. It is worth
23



noting that doing a similar exercise on other agreements produces somewhat different
results. Columns 5 and 6 of Table 3 report the results for these other agreements.
Conclusion
This paper has shown that reciprocity is important in free trade agreements. In
particular, the results suggest that a one percent increase in preferences offered leads to
about a one-half of a percent increase in preferences received. One exception is in North-
South agreements, where we find a modified form of reciprocity that is related to country
size: a one percent increase in preferences offered by the big (small) country leads to
more (less) than a one percent increase in preferences received.
From a policy ]perspective, the results suggest that incentives to maintain
protection to extract more concessions in a trade agreement are of concern. In general,
however, they should be less of a concern in North-South agreements since the marginal
value of a developing-country tariff reduction in terms of its effect on reciprocal
reduction is very small. The gains from unilateral liberalization are likely to far outweigh
potential gains from using protection as a bargaining chip.
References
Abrego, L, C. Perroni, J. Whalley, and R. Wigle (1997) "Trade and Environment:
Bargaining Outcomes fiom Linked Negotiations" NBER Working Paper #6216.
Anderson, K. and H. Norheim (1993) "History Geography and Regional Economic
Integration." In Regional Integration and the Global Trading System. New York: St.
Martin's Press, 19-51.
Bagwell K. and R. Staiger (2000) "GATT Think" NBER Working Paper 8005.
24



Brander J. and P. Krugman (1983) "A 'Reciprocal Dumping' Model of International
Trade." Journal of International Economics, 15, 313-323.
Davis D. and C. Kowalczyk (1998) "Tariff Phase-Outs: Theory and Evidence from
GATT and NAFTA" In The Regionalization of the World Economy, NBER Project
Report Series. Chicago and London: University of Chicago Press, 227:250.
Finger, J.M., U. Reincke, and A Castro (1999) "Market Access Bargaining in the
Ururguay Round: Rigid or Relaxed Reciprocity?" World Bank Working Paper #2258.
Freund C. (2000) "Multilateralism and the Endogenous Formation of Preferential Trade
Agreements," Journal of International Economics.
Limao, Nuno (2002) "Are Preferential Trade Agreements with Non-Trade Objectives a
Sumbling Bloc for Multilateral Liberalization" University of Maryland Center for
International Economics Working Paper 02-02.
Rogers, W. H. (1993) "Regression Standard Errors in Clustered Samples." Statistical
Technical Bullitin 13: 19-23.
Staiger R. (1998) "Tariff Phase-Outs: Theory and Evidence from GATT and NAFTA:
Comment" In The Regionalization of the World Economy, NBER Project Report Series.
Chicago and London: University of Chicago Press, 253:258.
Perroni, C. and J. Whalley (2000) "The New Regionalism: Trade Liberalization or
Insurance?" Canadian Journal of Economics, 1-24.
White, H. (1980) "A Heteroskedasticity-Consistent Covariance Matrix Estimator and a
Direct Test for Hereroskedasticity." Econometrica 48:817-830.
25



Figure 1: Feasible Trade Agreements
ln(tY)
Y's constraint
a         X's constraint
/*A/
ln(tx)
Figure 2: No Feasible Agreements
In(tY)
X's constaint
1.            Y's constraint
Ln(tx)
26



Figure 3: Enforceability Constraints Identifiable
ln(tY)
Y's constaint
X's constraint
0'S
ln(tx)
Figure 4: Enforceability Constraints Not Identified
ln(tY)
Y's constaint
0 
0 0  0  0
0          4
<  .     /'sc~~s onstraint
o <0
0    0
ln(tX)
27



.Figure 5: Preferences in Actual Agreements
25 -
000 0   00 0    
20-              A
A
~~ 15  ~~    ~     A                            o world
S6. 15 -          14 O^ ^   L A                      A, bilateral
<, 1-     8k     & A      A       A.  *
4.01         S     ^ 0i   0         0
OO,MO*    ot 0 0  A, OA   0
5 -   **i' A*> *   A
A     A
0 IL
0            10            20           30
tariff in X
28



Table 1: Determinants of Trade Preferences
tij    tij     tij    psij   psij    psij   Alij  Al\ij  Ali
_______    (1)    (2)     (3)    (4)     (5)    (6)       () ()       9
tji      0.58** 0.63** 0.57**
(4.40)  (3.47)  (6.02)
psji                           0.51**  0.40** 0.56**
(11.68) (3.91)  (9.12)
AIji                                                 .31**   0.31*  0.29**
(3.82) (2.33) (2.54)
gdpi     0.06           0.06   0.43**         0.38    0.02          0.02
(1.06)         (1.06)  (2.97)        (2.49)  (0.80)       (0.79)
gdpj     -0.02          -0.02   0.47**         0.42   -0.02         -0.02
(-0.39)        (-0.37) (6.65)        (5.45)  (0.92)       (-0.83)
Country  No      Yes    No     No      Yes    No      No     Yes    No
fixed
effects
Method   OLS     OLS    IV     OLS     OLS     IV     OLS    OLS    IV
NOB      104     104    104     104    104     104    182    182    182
R-       0.35    0.87   0.35    0.68   0.88    0.68   0.10   .44    0.10
SquareI                       II
Mlij is the percenage change in i's export intensity toj. Errors corrected for
heteroskedasticity and pairwise correlation. All regressions include a constant tenn,
values for the constant not reported. **Significant at the 1 percent level.*Significant at
the 5 percent level. In columns (3) and (4) country fixed effects are used as instruments
for Alji.
29



T'able 2: Reciprocity in North-South Agreements
tij _         PSi,                   AI,j
North-  Other  North-  Other   North-  Other  North-  Other
South          South           South          South
( __ _(1)  (2)  (3)   (4)     (5)      (6)     (7)    (8)
tji     -0.05  0.68**
(-0.45)  (6.53)
PSjM                    -0.12  0.50**
(-0.36) (11.98)
AIji                                   .39**   0.25*   -0.11  0.37**
(3.02)  (2.55)  (-0.54)  (3.87)
Ln(GDP1)    0.29   0.42   1.66**  0.43*   0.03     0.02   0.04    0.02
(1.42)  (0.67)  3.09   (2.04)  (0.72)  (0.44)  (0.63)  (0.58)
Ln(GDPj)    0.23   -0.00  1.63**  0.55**  0.00    -0.04   0.02   -0.04
(1.69)  (-0.03)  (3.38)  (5.68)  (0.11)  (-1.59)  (0.39)  (-1.67)
Method     OLS    OLS     OLS    OLS      OLS     OLS     IV     IV
NOB        22     82      22     82      58      124     58     124
R-Square   0.21    0.45    0.88   0.62    0.16     .07    0.00    .06
A constant is included in all regressions, value not reported. Errors corrected for
heteroskedasticity and pairwise correlation. *Significant at the 5% level. ** Significant at
the 1% level.
30



Table 3: Determinants of Large and Small Country Preferences
Total           North-South           Other
Large    Small     Large    Small     Large     Small
Access   Access    Access  Access to  Access    Access
to Small to Large  to Small   Large   to Small to Large
(1)      (2)       (3)      (4)       (5)       (6)
tji     1.39**   0.58**   3.31**    0.17**    1.38**   0.63**
(9.71)   (9.71)    (3.12)   (3.12)    (9.98)    (9.98)
PSji    2.11**    .41**     1.00      0.10    2.25**    0.39**
(11.04)  (11.04)   (1.05)    (1.05)   (10.05)   (10.05)
AIj;    .64**    .52**     1.59**    .43**     0.42**   0.56**
(6.07)   (6.07)    (5.89)   (5.89)    (4.11)    (4.11)
All regressions run with a constant and gdpi and gdpj using SUR. Only the
coefficients on preference variables from each regression are reported. Columns
(1), (3), and (5) report the results when the small country's preferences granted is
the dependent variable. Columns (2), (4), and (6) report the results when the large
countrys' preferences granted is the dependent variable.
31



Appendix Table 1: Regional Integration Agreements
Used in Tariff and
Agreement               Date of entry  Type of agreement   Producer Surplus
into force                           Measure
CER/ Australia-New Zealand            1-Jan-83        Free trade             No
______ _____  agreement               N
United States - Israel               19-Aug-85        Free trade             No
______ _____  agreement               N
EC accession of Portugal and Spain    1-Jan-86       Accession to            No
customs union            N
Mercosur/ Argentina Brazil Paraguay  29-Nov-91      Customs union            Yes
Uruguay                              29-Nov-91      Customs_union_Ye
EC - Czech Republic                   1-Mar-92        Free trade             Yes
_____________________                   ~agreem ent
EC - Slovak Republic                  1-Mar-92        Free trade             Yes
_________ _________ _________agreem            entYe
EC - Hungary                          1-Mar-92        Free trade             Yes
agreementYe
EC - Poland                           1-Mar-92        Free trade             Yes
____ ___ ____ ___ ____ ___ ___   ___ ____ ___ ag reem ent _ _ _ _ _ _ _ _ _
EFTA - Turkey                         1-Apr-92        Free trade             No
______ ______ ______ ______   . ___agreem ent
EFTA - Czech Republic                 1-Jul-92        Free trade             No
______________ ______________agreem            ent              N
EFTA - Slovak Republic                1-Jul-92        Free trade             No
_______ _______ _______ _______agreem          ent_       _  _ _ _ _ _ _ _
Czech Republic - Slovak Republic      1-Jan-93      Customs union            No
EFTA - Israel                         1-Jan-93        Free trade             No
____ ___ ____ ___ ____ ___ ___   ___ ____ ___ agreem ent        N
CEFTA1 Bulgaria Czech Republic
Hungary Poland Romania Slovak         1-Mar-93        FreeYes/ except Bulgaria
Republic Slovenia                                     agreement          Yes
EC - Romania                          1-May-93        Free trade             No
____ ___ ____ ___ ____ ___ ___   ___ ____ ___ agreem entY        e
EFTA - Romania                        1-May-93        Free trade             N
_____ ____ ____ ____ _____ _   __ ____ ____ agreem ent          N
Chile - Venezuela                     1-July-93        Free trade            Ye
____ ___ ____ ___ ____ ___ ___   ___ ____ ___ agreem entY        e
EFTA - Bulgaria                       1-Jul-93         Free trade             No
________ _______ ___   ____ _______     agreem ent              N
EFTA - Hungary                        1-Oct-93         Free trade             No
____ ___ ____ ___ ____ ___ ___   ___ ____ ___ agreem ent        N
EFTA - Poland                        15-Nov-93         Free trade             No
32__ _ _ _ _ _ _agreement
32



EC - Bulgaria                       31-Dec-93        Free trade             No
_________ _________ _________agreem            ent             N
Chile -Colombia                      1-Jan-94        Free Trade             Yes
____ ____ ____ ____ ___ ____   ___ ____ ___ A greem ent
NAFTA                                1-Jan-94         Free trade            Yes
.__ _ _ __ _ _ _ _ __ _ _ _ _agreem ent
BAFTAI Estonia Latvia Lithuania      1-Apr-94         Free trade            No
______ _____  agreem ent      _ _ _ _ _ _  _ _ _
Bolivia- Mexico                       1-Jan-95        agreement             Yes
Colombia-Mexico-Venezuela             1 -Jan-95       Free trade            Yes
.____ ____ ____ ____ ____ ____agreem ent
Costa Rica -Mexico                    1-Jan-95        Free trade            Yes
EC - Lithuania                        1-Jan-95        Free trade            Yes
agreementYe
EC - Estonia                          1-Jan-95        Free trade            Yes
agreement
EC - Latvia                           1-Jan-95        Free trade            Yes
agreementYe
EC accession of Austria, Finland and  1-Jan-95       Accession to           Yes
Sweden                                              customs union
EFTA - Slovenia                       1-Jul-95        Free trade             No
agreement
CEFTA accession of Slovenia           1-Jan-96     Accession to free         No
trade agreement
EC - Turkey                           1-Jan-96      Customs union           Yes
EFTA - Estonia                        1-Jun-96        Free trade             No
______________ _____________agreem             ent              N
EFTA - Latvia                         1-Jun-96        Free trade             No
______________ _____________agreem             ent              N
Slovenia - Latvia                     1-Aug-96        Free trade             No
EFTA - Lithuania                     1-Aug-96         Free trade             No
______________ _____________agreem             ent              N
Chile - Mercosur                      1-Oct-96        Free trade             Yes
______________ _____________agreem             entYe
Slovak Republic - Israel              1-Jan-97        Free trade            Yes
__________ _________ _________agreem           ent
Poland - Lithuania                    1-Jan-97        Free trade             Yes
agreementYe
Slovenia - Estonia                    1-Jan-97        Free trade             No
____ ____ ___ ____ ____ ___   ___ ____ ___ agreem ent           N
Canada - Israel                       1-Jan-97        Free trade             Yes
____ ____ ___ ____ ____ ___   ___ ____ ___ agreem entYe
EC - Slovenia                         1-Jan-97        Free trade             No
agreement
33



Slovenia - Lithuania                  1-Mar-97         Free trade             No
______ _____   agreement               N
Bolivia - Mercosur                    2-Mar-97         Free trade             Yes
agreement       _________
Israel - Turkey                       1-May-97         Free trade            Yes
____ ___ ____ ___ ____ ___ ___  ___ ____ ___ agreem entY          e
CEFTA accession of Rornania            1-Jul-97     Accession to free         Yes
trade agreement 1-Ju-97agremet_N
Slovak Republic - Latvia               1-Jul-97        Free trade             No
agreement
Slovak Republic - Lithuania            1-Jul-97        Free trade             No
agreement              No
Czech Republic - Latvia                1-Jul-97        Free trade             Yes
agreement              Yes
Canada - Chile                         5-Jul-97        Free trade             Yes
agreement
Czech Republic - Lithuania            1-Sep-97         Free trade             No
______ ______  agreementYe
Czech Republic - Israel               1-Dec-97         Free trade             Yes
Hungar-Turke  1-Apragreement               Yes
Romania - Turkey                      1-Feb-98         Free trade             Yes
Chile-Peru___ 1 -9agreement                Yes
Hungary - Israel                       1-Feb-98        Free trade             Yes
______ ______  agreementYe
Czech Republic - Estonia              12-Feb-98        Free trade             Yes
______ ______  agreementYe
Slovak Republic - Estonie             12-Feb-98        Free trade             No
______ ______  agreement
Poland - Israel                       1-Mar-98         Free trade 
agreement
Lithuania - Turkey                    1-Mar-98         Free tradeYe
____ ___ ____ ___ ____ ___ ___   ___ ____ ___ agreem entYe
Hungary -Turkey                        1 -Apr-98       Free tradeYe
agreementYe
Estonia - Turkey                       1-Jun-98        Free tradeYe
agreementYe
Chile - Pewu                           1-Jul-98        Free Trade             Yes
__________ __________ _________agreem           ent
Mexico - Nicaragua                     1-Jul-98        Free Trade             Yes
Agreement       ________
Czech Republic - Turkey               1-Sep-98         Free tradeYe
______ ______  agreementYe
Slovak Republic - Turkey              1-Sep-98         Free trade              N
____ ____ ____ ____ ___ ____ ____ ____ ___ agreem ent            N
Slovenia - Israel                     1-Sep-98         Free trade              N
________ _______ ________ __   ____ _______ agreem ent           No
34



Bulgaria - Turkey                      I-Jan-99         Free trade             No
agreement
CEFTA accession of Bulgaria            1-Jan-99     Accession to free          No
trade agreement-
Poland - Latvia                        1-Jun-99         Free trade             Yes
agreementYe
Chile - Mexico                         1-Aug-99         Free trade             Yes
____ ___ ____ ___ ____ ___ ___   ___ ____ ___ agreem entY          e
EFTA - Moroco                          1-Dec-99         Free trade              No
35__ _ _ _ _ _ _agreement
35



Policy Research Working Paper Series
Contact
Title                          Author                  Date            for paper
WPS3043 The Incentive-Compatible Design  Thorsten Beck         May 2003          K. Labrie
of Deposit Insurance and Bank                                           31001
Failure Resolution: Concepts and
Country Studies
WPS3044 Impregnated Nets Cannot Fully    Mead over             May 2003          H. Sladovich
Substitute for DDT: Field       Bernard Bakote'e                        37698
Effectiveness of Malaria Prevention  Raman Velayudhan
in Solomon Islands             Peter Wilikai
Patricia M. Graves
WPS3045 Causes and Consequences of Civil  Klaus Deininger      May 2003          M. Fernandez
Strife: Micro-Level Evidence from                                       33766
Uganda
WPS3046 Migration, Spillovers, and Trade  Ross Levine          May 2003          E. Khine
Diversion: The Impact of       Sergio L. Schmukler                      37471
Internationalization on Stock Market
Liquidity
WPS3047 Comparing Mortgage Credit Risk   Robert Buckley        May 2003          0. Himid
Policies: An Options-Based Approach Gulmira Karaguishiyeva              80225
Robert Van Order
Laura Vecvagare
WPS3048 Targeted Transfers in Poor Countries: Martin Ravallion  May 2003         P. Sader
Revisiting the Tradeoffs and Policy                                     33902
Options
WPS3049 Hidden Impact? Ex-Post Evaluation  Shaohua Chen        May 2003          P. Sader
of an Anti-Poverty Program      Martin Ravallion                        33902
WPS3050 The Changing Financial Landscape:  Wafik Grais         May 2003          R. Vo
Opportunities and Challenges for  Zeynep Kantur                         33722
the Middle East ancl North Africa
WPS3051 Regional, Multilateral, and Unilateral  Glenn W. Harrison  May 2003      P. Flewitt
Trade Policies of MERCOSUR for  Thomas F. Rutherford                    32724
Growth and Poverty Reduction in  David G. Tarr
Brazil                          Angelo Gurgel
WPS3052 Long-Run Impacts of China's WTO  Kym Anderson          May 2003          S. Lipscomb
Accession on Farm--Nonfarm Income Jikun Huang                           87266
Inequality and Rural Poverty   Elena lanchovichina
WPS3053 Economic Impacts of China's      Elena lanchovichina   May 2003          S. Lipscomb
Accession to the World Trade   William Martin                           87266
Organization
WPS3054 The New Comparative Economics    Simeon Djankov        May 2003          G. Sorensen
Edward Glaeser                           37088
Rafael La Porta
Florencio Lopez-de-Silanes
Andrei Shleifer
WPS3055 The Global Growth of Mutual Funds  Deepthi Fernando    May 2003          A. Yaptenco
Leora Klapper                            31823
Viktor Sulla
Dimitri Vittas



Policy Research Working Paper Series
Contact
Title                           Author                  Date             for paper
WPS3056 Avoiding the Pitfalls in Taxing   Patrick Honohan        May 2003          A. Yaptenco
Financial Intermediation                                                  31823
WPS3057 Teaching Adults to Read Better and  Helen Abadzi         May 2003           H. Abadzi
Faster: Results from an Experiment                                        80375
in Burkina Faso
WPS3058 Working for God? Evaluating Service Ritva Reinikka       May 2003           H. Sladovich
Delivery of Religious Not-for-Profit  Jakob Svensson                      37698
Health Care Providers in Uganda
WPS3059 Do Capital Flows Respond to Risk  Cesar Calder6n         May 2003           T. Tourougui
and Return?                     Norman Loayza                             87431
Luis Serven
WPS3060 World Market Integration through the  Rui Albuquerque    May 2003           T. Tourougui
Lens of Foreign Direct Investors  Norman Loayza                           87431
Luis Serven