t the trade effects of the FTAA will exceed the impact of its tariff and quota changes.
For instance, to the extent that the FTAA requires closer cooperation on sanitary and phytosani-
tary issues, as is the case with the North American Free Trade Agreement (NAFTA), member
countries are likely to adjust their import standards so that they do not restrict trade unnecessari-
ly. Moreover, the FTAA is likely to have a myriad of indirect effects that ultimately expand
trade, even though these changes may not be spelled out in the agreement. Many developments
of this sort took place following the implementation of NAFTA and the Common Market of the
South (MERCOSUR, or Mercado Com�n del Sur). Examples include increased investor confi-
dence within the two regions, the further exploitation of scale economies, and the upgrading of
transportation linkages along new and existing routes of trade.
To better understand the potential breadth of the FTAAs influence, this chapter assesses the
impact that NAFTA, MERCOSUR, and related agreements have had on agricultural trade within
the Western Hemisphere. Focusing on U.S. agricultural exports from 1980 through 1999, the
chapter employs a series of modified gravity models, as suggested by Cheng and Wall (1999), to
identify noteworthy changes in trade coinciding with these agreements. A main strength of this
approach is that it distinguishes the impact of a trade agreement on U.S. exports to a specific
country from the relative closeness of that countrys bilateral trade relationship with the United
States. However, the variables used to identify trade agreements may also capture the influence
of other factors that were contemporaneous to these reforms.
To develop a complete picture of regionalisms impact on U.S. agricultural exports, separate
models are estimated at the aggregate level and for 32 individual commodities. This analysis
generates several important findings:
(1) Unilateral trade reforms undertaken by Mexico during the late 1980s and early 1990s
have provided a sizable boost to U.S. agricultural exports to Mexico. According to grav-
ity-model estimates, these unilateral reforms accounted for 39 percent of U.S. agricul-
tural exports to Mexico from 1989 through 1999, or an average of roughly $1.7 billion
per year. Thus, one of NAFTAs main benefits to U.S. agriculture has been to lock in
reforms that Mexico had made prior to NAFTA.
(2) NAFTAs estimated influence on U.S. agricultural exports to Mexico is positive and sta-
tistically significant for four of the commodities studied (grapes, tobacco products, yarn
and thread, and leather), and it is positive but statistically insignificant for 18 other com-
modities. Although the model differentiates NAFTA and Mexicos unilateral reforms,
both were components of an integrated strategy for market reform that Mexico has pur-
sued since the mid-1980s. Mexican trade liberalization, both unilateral and through
NAFTA, accounted for an average annual increase in U.S. agricultural exports to
Mexico of $3.1 billion during 1994-99.
(3) The estimated impact of the Canada-U.S. Free Trade Agreement (CFTA) and NAFTA
on U.S. agricultural exports to Canada is not statistically significant. This finding,
Economic Research Service/USDA
U.S. Agriculture and the Free Trade Area of the Americas/AER-827 39 which is observed both at the aggregate level and for all the individual commodities
studied, may reflect the fact that most barriers to U.S.-Canada agricultural trade were
relatively low prior to CFTA, while several important sectorsdairy, poultry, and
eggswere exempted from trade liberalization.
(4) MERCOSURs application of a common external tariff has lowered some barriers to
U.S. agricultural exports, creating new opportunities for trade. Relatively high levels of
U.S. agricultural exports during the MERCOSUR period are observed at the commodity
level for all four MERCOSUR countries and at the aggregate level for Argentina,
Paraguay, and Uruguay. In the cases of Argentina and Brazil, several consumer-oriented
food products from the United States appear to have benefited from tariff reductions
linked to MERCOSURs common external tariff, although the value of this trade is still
small compared with exports to Canada and Mexico.
(5) MERCOSUR appears to have had a trade-diverting effect on U.S. wheat exports to
Brazil. With the creation of MERCOSUR, Argentina has dramatically increased its
share of the Brazilian wheat market, while U.S. wheat exports to Brazil have declined.
Argentine wheat enters Brazil duty free, while U.S. wheat faces MERCOSURs com-
mon external tariff for the product.
The rest of the chapter contains a methodological overview of the modified gravity models and
an extensive discussion of their findings. Technical aspects of the models are discussed in appen-
dix 2-1, while the International Bilateral Agricultural Trade (IBAT) database, the source of the
export data used in the chapter, is profiled in appendix 2-2.
Gravity Model Methodology
In its most basic application, the gravity model of international trade posits that the level of
exports from one country to another is a function of each countrys gross domestic product
(GDP) and its population, as well as the distance between the two countries. To estimate the
trade effects of regional trade agreements, a number of gravity modelers (such as Frankel,
1997; Endoh, 1999; and Soloaga and Winters, 2001) have included additional explanatory vari-
ables that indicate a countrys membership in a specific trade agreement or trade bloc. These
variables, however, do not distinguish the influence of a trade agreement from the long-term, rel-
ative closeness of a specific bilateral trading relationship. Nor do they account for the strong
likelihood that the impact of a trade agreement varies from one participant to another.
To overcome these shortcomings, this chapter features a different specification of the gravity
model (table 2-1). Following Cheng and Wall, the modified models include two sets of fixed
effects (variables with the value of one or zero) that respectively identify specific importing
countries and specific years. The fixed effects for importing country play a crucial role in the
analysis, as they control for the importing countrys long-term bilateral trading relationship with
the United States. This increases the likelihood that the trade-agreement variables capture the
effects of those agreements, rather than the general closeness of a particular bilateral relation-
ship. Moreover, the trade-agreement variables are country-specific in order to address the possi-
bility that the impact of an agreement varies among its participants. Table 2-2 provides a defini-
tion of each trade-agreement variable.
While the modified gravity models provide an improved framework for assessing regional trade
agreements, the trade-agreement variables may still capture the influence of unrelated develop-
ments that are contemporaneous to these accords. Unusual weather patterns are an obvious
40 U.S. Agriculture and the Free Trade Area of the Americas/AER-827
Economic Research Service/USDA Economic Research Service/USDA
U.S. Agriculture and the Free Trade Area of the Americas/AER-827 41
Table 2-1Comparison of basic and modified gravity models
Basic gravity model
Dependent variable:
Log of exports from country
i to country j
Explanatory variables:
Intercept
Log of GDP of country
i
Log of GDP of country
j
Log of population of country
i
Log of population of country
j
Log of distance between country
i and country j
Other variables selected by researcher, such as dummy
variables to denote trade flows corresponding to particular
trade agreements
Econometric Approach:
Ordinary least squares (usually)
Modified gravity model, as used in this chapter
Dependent variable:
Log of U.S. agricultural exports to country
i in year t
(in U.S. dollars)
Explanatory variables:
Intercept
Log of GDP of country i in year t (in U.S. dollars)
Fixed effects denoting importing country
For example, the fixed effect for Mexico equals one if
Mexico is the importing country and zero otherwise. For
purposes of comparison, no fixed effect is included for Canada.
Fixed effects denoting year
For example, the fixed effect for 1998 equals one if the year
is 1998 and zero otherwise. For purposes of comparison,
no fixed effect is included for 1999.
Trade-agreement variables dummy variables that identify
country
is participation in a particular trade agreement in year t
Econometric approach:
Tobit
Source: Economic Research Service.
Table 2-2Trade-agreement variables in the modified gravity models
NAFTA countries
Unilateral-Mexico
Equals one for exports to Mexico during 1989-99 and zero otherwise
NAFTA-Mexico
Equals one for exports to Mexico during 1994-99 and zero otherwise
CFTA-Canada
Equals one for exports to Canada during 1989-99 and zero otherwise
NAFTA-Canada
Equals one for exports to Canada during 1994-99 and zero otherwise
MERCOSUR countries
Full members:
Argentina/1991-99
Equals one for exports to Argentina during 1991-99 and zero otherwise
Argentina/1994-99
Equals one for exports to Argentina during 1994-99 and zero otherwise
Brazil/1991-99
Equals one for exports to Brazil during 1991-99 and zero otherwise
Brazil/1994-99
Equals one for exports to Brazil during 1994-99 and zero otherwise
Paraguay/1991-99
Equals one for exports to Paraguay during 1991-99 and zero otherwise
Paraguay/1994-99
Equals one for exports to Paraguay during 1994-99 and zero otherwise
Uruguay/1991-99
Equals one for exports to Uruguay during 1991-99 and zero otherwise
Uruguay/1994-99
Equals one for exports to Uruguay during 1994-99 and z