Comments

Jorge Tovar: Despite the abundant literature on the empirical effects of trade liberalization for both developed and developing countries, the real consequences remain unknown. Giovannetti and Menezes-Filho focus on one of the most complex implications of trade liberalization: the evolution of the labor market. In particular, the paper studies the relation between trade liberalization and the evolution of the demand for skilled labor in Brazil. The authors argue that the reduction in tariffs caused a decline in the prices of imported intermediate goods, which embody advanced technologies that use skilled workers. Trade liberalization thus leads to an increase in the relative demand for skills.

The authors test their hypothesis using Brazilian household and manufacturing data for the period 1990 to 1998, excluding 1991 when no annual industrial survey was conducted. The main conclusion of the paper is that the decline in input tariffs had a significant effect on skill upgrading in Brazil. The strategy followed to identify the effects of trade liberalization is to estimate a series of regressions controlling for input tariffs, which serve as proxies for technology diffusion. The paper finds that the relative demand for skilled workers increased. This result has a potentially interesting policy implication: if the relative demand for skilled workers increases, the market should make an effort to increase the supply of such workers in the long run. This could be seen as an indirect benefit of trade liberalization.

The paper does a great job questioning the differential effects of trade liberalization on workers with different skill levels. However, the authors ignore the political economy of tariff reductions, something that is discussed extensively in the literature. They argue that endogeneity issues do not affect their regressions, that is, the error term and tariff reductions are not correlated because the government's objective was to reduce all tariffs to a common level. The inclusion of industry fixed effects is expected to control for industries that systematically receive more protection. The inclusion of time dummies solves the potential correlation between tariffs and other macroeconomic events. As Goldberg and Pavcnik argue, however, two potential sources of endogeneity remain.¹ First, unobserved time-varying political economy factors could simultaneously affect tariffs and industry wages (or labor demands). Second, workers could choose certain industries over others based on unobserved time-varying characteristics. If, for example, trade liberalization caused the most productive workers to leave sectors that experienced large tariff cuts, tariff coefficients would be biased upwards. An instrument to correct for this potential endogeneity can be based on the solution proposed by Goldberg and Pavcnik.² Essentially, as long as tariffs are reduced over a short period and the industry structure does not vary significantly, one can build an instrument using the information contained in the lags of the tariff levels per industry.

The potential lack of robustness in the estimates can be seen in table 3, where the inclusion of fixed effects (to control for endogeneity) has an important impact on the value of the coefficient. The paper strongly suggests that differentials by type of worker do exist, but the real difference might be hidden behind these unexplored political economy effects.

The paper explores a relevant topic with important policy implications. The authors' use of data at the firm and individual level marks a significant improvement over previous work, and the paper provides new insight into the true empirical effects of trade liberalization on labor markets. The strategy followed by the authors, however, makes one wonder why Brazil is such an unusual case that no political economy issues arose when the trade liberalization process was implemented.

Osmel Manzano: Giovannetti and Menezes-Filho provide an interesting discussion of the effects of trade liberalization on the demand for labor of different skill levels. As mentioned in the paper, this is an important issue in the literature on the effects of trade liberalization. Given that developing countries are relatively abundant in unskilled labor, trade liberalization should increase the relative demand for this type of worker and a corresponding decline in the skill premium. The evidence so far is inconclusive, however. This paper sheds light on the debate by separating the effects...