Over the past three decades, labor markets in developing countries went through important changes in terms of wage differentials and the employment composition of their workforce. One of these changes was the fast rise in the supply of workers with intermediate levels of education, both in absolute terms and relative to the supply of workers with higher and lower education levels. Moreover, the wage differentials between college-educated workers and those with intermediate levels of education rose, while the differentials between the latter and workers with up to elementary education fell. These changes had important implications for poverty and inequality.1
Many Latin-American countries also went through trade liberalization processes in the 1980s and 1990s as part of a package of market-orientated reforms. These reforms induced important changes not only in their product markets, by changing relative prices and increasing the productivity of manufacturing firms, but also in the labor markets, by spurring labor reallocation and transition into the informal sector.2 A significant body of research links the changes in the wage differentials reported above to the trade liberalization process, following a literature that first examined the impact of trade liberalization on inequality in the United States and other developed countries.3 [End Page 1] The theoretical framework used by most of these studies is the traditional Heckscher-Ohlin model. According to this model, since skill is a relatively scarce resource in most Latin American countries, opening up trade with more developed countries should raise the prices of low-skill-intensive goods. The production of these goods would therefore increase, intensifying the demand for unskilled workers and generating a decline in the skill premium. Many researchers were thus surprised when the data revealed that the wage differentials between college-educated workers and workers with up to secondary education in developing countries were actually rising.
One explanation for this apparent rejection of the Heckscher-Ohlin model is that countries do not move instantaneously from a closed economy to an open one, but from a pattern of high tariffs to one of low tariffs. Moreover, the changes in tariffs tend to vary substantially across sectors, depending on the structure of protection that prevailed before trade liberalization. In several cases, this structure was unrelated to the pattern of comparative advantages across industries. The changes in relative prices during a trade liberalization process therefore follow the changes in tariffs, and the impact of trade liberalization on inequality depends on the correlation between the changes in tariffs and the skill intensity across manufacturing sectors. Using this framework, Hanson and Harrison, as well as Robertson, find that trade liberalization was associated with a rise in wage inequality in Mexico, whereas Gonzaga, Menezes-Filho, and Terra report that the predictions of the Heckscher-Ohlin model are actually consistent with the decline in the wage differentials between workers with secondary education or more and workers with less than secondary education in Brazil in the early 1990s.4
An important limitation of these papers is that the framework they use only allows for two skill groups, while the descriptive evidence summarized above consistently points to different movements in the wage differentials between skilled and semiskilled workers, on the one hand, and between semiskilled and unskilled workers, on the other. How can trade opening explain both the rise in returns to college education and the fall in returns to secondary education that took place in many developing countries in the 1980s and 1990s? Two hypotheses were put forward to explain these trends. First, the movements in wage differentials may reflect the decline in the relative supply of college-educated workers, which did not increase as quickly as the supply of workers with secondary education. Second, the relative demand for college-educated workers may have risen. Schady and Sánchez-Páramo find [End Page 2] that the relative demand for college-educated...