Trade Liberalization, Macroeconomic Fluctuations, and Contingent Protection in Latin America
Latin America has pursued very aggressive trade liberalization policies since the mid-1980s. Both tariff and nontariff barriers have declined significantly as a result. This process of trade liberalization was not reversed even during the strong macroeconomic shocks that affected the region in the mid-1990s (the so-called tequila crisis), the late 1990s (the Brazilian devaluation), and 2001–02 (the Argentine devaluation).1 As expected, the move toward trade liberalization was accompanied by a strong surge of trade flows. This increased the pressure on local producers, which must now compete with worldwide imports in domestic markets. Governments in the region thus faced renewed demands for some type of import-relief measures. Authorities responded by implementing contingent protection, in particular, antidumping actions.2
The application of antidumping measures is governed by specific rules whose main principles are established in the World Trade Organization (WTO) Antidumping Agreement. This allows governments to impose these measures if it is determined, after an investigation conducted in accordance with this agreement, that dumping is occurring, that the domestic industry producing the similar [End Page 147] product in the importing country is suffering injury, and that there is a causal link between the dumping and the injury. Under this normative approach, antidumping actions should be associated with episodes of depressed economic activity or real exchange rate appreciations (or both), given that these developments favor the finding of industry injury or dumping practices and thus make the case for trade remedy measures easier to justify.
An alternative hypothesis is that these policies simply represent another form of protection that eventually replaces the protection inherent in tariff levels. Their prevalence thus depends on factors related to lobbying activity, which could also be stronger in times of adverse macroeconomic conditions. If this were the main driving force, antidumping measures and tariff levels would display a negative relation, and contingent protection should be responsive to political economy variables such as industry concentration.
The main purpose of this study is to investigate the above hypotheses for Latin America. We start by presenting an updated description of the use of antidumping actions in this region since the late 1980s. We identify the number of antidumping investigations initiated each year, the number of these investigations that led to the imposition of import duties, the sectors involved, and the exporting countries subject to these measures. We show that Argentina, Brazil, and Mexico are the main users in Latin America, with more than 80 percent of all cases (85 percent of definitive measures). Among the subject countries, Brazil and China are important for Argentina, while the United States and China are the main affected nations in the case of Brazil. In Mexico, the United States accounts for almost 30 percent of initiations, followed by China with 14 percent. We find a clear pattern of sectoral concentration of antidumping measures: chemicals and basic metals are antidumping-intensive industries in all three countries, and Argentina also exhibits significant antidumping activity in metal products, machinery and equipment, and engines and electrical equipment.
We then explore the relation between the implementation of these actions and the behavior of key macroeconomic variables, such as the level of economic activity and the real exchange rate. Several authors use regression analysis to study this relation empirically for developed countries.3 Niels provides one of the few studies of a developing economy in his research on Mexico.4 We extend his analysis on various dimensions. First, we include Argentina and Brazil in our sample. Second, we exploit the bilateral variability observed both in the dependent [End Page 148] variable (that is, antidumping actions are defined as country pairs) and in some of the explanatory variables (namely, the bilateral real exchange rate) to empirically identify the response of antidumping initiations to the variables of interest. Third, we incorporate industry-sector variability in the data: the dependent variable varies...
