تجارت و تخصیص استعداد همراه با نواقص بازار سرمایه

Trade liberalization in the 1980s and 1990s has been associated with a sharp increase in the skill premium in both developed and developing countries. This is in apparent conflict with neoclassical theory, according to which trade should decrease the relative return on the relatively scarce factor, and thus decrease the skill premium in skill-scarce developing countries. We develop a simple model of trade with talent heterogeneity and capital market imperfections, and show that trade can increase the skill premium in a skill-scarce South that opens up to a skill-abundant North, both in the short run as well as in the long run. We show that trade has two effects: it reduces the skilled wage, and therefore drives non talented agents out of the skilled labor force. It also reduces the cost of subsistence, thereby allowing the talented offspring of unskilled workers to go to school. This compositional effect has a positive effect on the observed skill premium, potentially strong enough to outweigh the decrease in the skilled wage. In our framework, trade liberalization may trigger an increase in the skill-premium in both the North and the South.

One of the most important results in Heckscher–Ohlin models of international trade, the Stolper–Samuelson theorem, predicts that when a country opens up to international trade – and thus, its relative price of skill-intensive goods decreases – the return of unskilled workers should increase, relative to the return of skilled workers.1 This prediction has been confirmed in a number of unskilled labor-abundant “early globalizers” (such as Italy, Singapore, South Korea and Taiwan) where trade has increased the unskilled wage relative to the skilled wage (thus decreasing the skill premium). However in the case of unskilled labor-abundant countries that have globalized in the 1980s and 1990s (such as most of Latin America, India and Hong Kong), trade seems to have increased the skill premium, rather than reducing it. 2 This fact, sometimes called the “skill premium puzzle”, has attracted a fair bit of attention. On the one hand, the trade literature has sought to reconcile the Latin American experience with Heckscher–Ohlin theory (HO from now on) by arguing that trade liberalization disproportionately affected unskilled labor-intensive industries (Revenga, 1997), or that countries such as China, Indonesia and Pakistan made the world outside Latin America actually unskilled labor-abundant (Davis, 1996 and Wood, 1999). In these contexts, HO theory would correctly predict an increase in the skill premium in Latin America. One problem with these interpretations is that they predict that skill intensity should have decreased across sectors in Latin America, a prediction that has not been confirmed in the data.3 In response to these shortcomings, the literature has turned to alternative trade models to explain the generalized increase in wage inequality,4 or to non-trade explanations, such as, skill biased technical change. In this paper, we propose a way to reconcile a traditional HO model of trade liberalization between an unskilled labor-abundant South and a skill labor-abundant North with an increase of the skill premium in the South as well as in the North. We do so by enriching the baseline model with talent heterogeneity, human capital accumulation, and credit constraints. The literature on trade liberalization in the presence of credit constraints (discussed below in detail) has shown that trade may increase human capital accumulation by relaxing the credit constraints faced by the poor, thereby improving their access to the education system. We show that when this is the case, trade may improve the allocation of talent to the skilled labor force, both in the short and in the long run. This compositional effect generates an upward pressure on the observed skilled wage, which can be strong enough to overturn the Stolper–Samuelson prediction of a lower skill premium in the South following trade liberalization. While reconciling the Stolper–Samuelson theorem with the Latin American experience, our model preserves the other main features of standard HO theory, including the fact that all industries in the South become more skill-intensive after trade liberalization.5 Our mechanism works as follows. Because of capital market imperfections, young agents cannot borrow to pay for their subsistence while attending school. Thus, only those whose parents have a high wage can possibly go to school. In an economy with little human capital, unskilled wages are low, and the cost of subsistence is high relative to the income of unskilled workers. This creates one equilibrium in which there are few skilled workers, the skilled wage is high, and all and only the offspring of skilled workers go to school. With heterogeneous talent, this equilibrium is “bad” in efficiency terms, in that many talented offspring of unskilled workers are prevented from going to school while many offspring of skilled workers go to school despite being non-talented. This is in contrast to a “good” equilibrium in which there are many skilled workers, the skilled wage is low, and all and only the talented workers go to school independently of the economic status of their families. We consider an economy that is skill-scarce because it is stuck at the bad equilibrium, and study its reaction to the liberalization of trade with a skill-abundant world. By putting a downward pressure on the skilled wage, trade may induce many non-talented skilled workers to drop out of the skilled labor force. At the same time, it reduces the cost of subsistence for unskilled workers, thus making it easier for their offspring to go to school. Because many of these previously-excluded agents are highly talented, they may still find it optimal to join the skilled labor force despite the trade-induced drop in the skilled wage. These two effects may move the economy from its initial equilibrium to the good equilibrium, thus increasing the average quality of the skilled labor force. This creates an upward force on the average observed skill premium, that can more than compensate the negative effect of trade on the skilled wage. Thus, the skill premium may increase in the skill-scarce country, both in the short run and in the long run. 6 Our results suggest that the Stolper–Samuelson theorem needs to be modified in the context of talent heterogeneity and imperfect credit markets, to account for the possibility of compositional changes in the skilled labor force. The literature on trade with capital market imperfections is now quite large. An important part of it has focused on how comparative advantage and the pattern of trade are determined by cross country heterogeneity in the efficiency of capital markets (see for example Kletzer and Bardhan, 1987 and Wynne, 2005; and Manova, 2008). Although our result is compatible with the idea that comparative advantage in the export of skill-intensive products may be driven by differences in capital market development, the focus of our paper is different. More connected to our paper is the literature on trade liberalization and skill acquisition in the presence of credit market frictions. This literature has studied several ways in which trade liberalization may affect domestic credit constraints and, through this channel, skill acquisition. In an important contribution Cartiglia (1997) shows that trade liberalization reduces the cost of schooling in a skill-scarce South by reducing the relative wage of skilled workers à la Stolper–Samuelson, thus making it easier for poor, credit constrained households to send their children to school. This effect may be large enough to offset the standard result that trade discourages the accumulation of the scarce factor (via the Stolper–Samuelson decrease in its relative return; see Findlay and Kierzkowski, 1983, and Grossman and Helpman, 1991), thus creating a positive association between trade liberalization and skill accumulation in the South. Ranjan, 2001a and Ranjan, 2003 enriches the setting in Cartiglia (1997) by studying how trade may affect credit constraints also through the distribution of income and wealth. The main intuition here is that trade increases (decreases) the wage income and long-run wealth of unskilled workers in the South (North). Assuming that credit constraints affect mainly the children of unskilled workers, trade results in a lessening of credit constraints in the South, and a strengthening of credit constraints in the North (unless credit constraints are institutionally less present in the North). 7 Building on this latter result, Chesnokova and Krishna (2009) show that the supply of skill-intensive goods in the North may actually decrease following trade liberalization, due to a strengthening of credit constraints. This carries the intriguing implication that trade may decrease welfare in such a country. 8 We borrow from this literature the basic insight that, in the presence of credit constraints, trade may increase the supply of skilled labor in the South. In particular, our result that trade may shift South from a low-skill equilibrium to a high-skill equilibrium in the long run has much in common with the results in Ranjan (2003). Our main innovation lies in the introduction of the kind of talent heterogeneity that maps into heterogeneity in productivity per worker. This allows us to investigate the compositional effects of the trade-induced increase in the skilled-labor supply. 9 Our main finding – that trade may lead to an increase in the observed skill premium – is novel to the literature. 10 It points to the importance of considering compositional effects of trade alongside standard effects on relative wages in efficiency units, which is what the previous literature has implicitly focused on. The paper is organized as follows. Section 2 presents our argument in an intuitive way. Section 3 develops the formal model, while Section 4 provides some generalizations. Finally, Section 5 concludes.

In this paper we develop a model of trade liberalization and occupational choice, with capital market imperfections. When an economy is unskilled labor-abundant because of credit constraints that affect the schooling decisions of agents, trade liberalization may have a non-standard effect on the skill premium. This is for two reasons. First, credit constraints may have allowed a large number of non-talented agents in the skilled labor force. Having been attracted to the skilled labor force by a high autarchic skilled wage, these agents may find it optimal to join the unskilled labor force when trade puts a downward pressure on the skilled wage. Second, credit constraints may have kept many talented agents out of the skilled labor force. To these agents, a trade-induced decrease in the cost of subsistence implies a lessening of credit constraints, and thus a better chance of upward mobility. Both of these effects result in an increase in the average talent of the skilled labor force, which may lead to an increase in the skill premium despite the trade-induced decrease in the skilled wage per efficiency unit of labor. Our results provide a possible explanation for the fact that trade liberalization in unskilled labor-abundant Latin America led to an increase in the skill premium in both Latin America and its skill-abundant trade partners. One implication of this is that the increase in the skill premium in Latin America does not necessarily need to result in a massive increase in income inequality, as it may be (at least partly) due to a better allocation of talent and more intergenerational mobility. While reconciling the predictions of the Stolper–Samuelson theorem with the Latin American experience, our mechanism is not incompatible with alternative explanations that have highlighted the role of skill-biased technical change or of quality upgrading. In fact, one interesting extension of our model is to consider the interaction of talent re-allocation with these other trade-induced changes in the structure of production. Other potential extensions include studying our mechanism in the context of more structural sources of comparative advantage (such as differences in physical capital, quality of schooling, etc.), and letting the decisions of agents be affected by the wealth distribution. Our results lead to several empirical predictions. First, trade liberalization should result in a higher proportion of not-so-talented educated children of well-off families to “leave” the skilled labor force. Such shifts could take place in the context of normal (or enhanced) labor market turnover, whereby these agents become less numerous among the newly-hired in the more competitive skilled professions, since they prefer something easier (and, in addition, earn rent income). In this context, our specific prediction – one that, admittedly, may be hard to test – would be that the average quality of newly-hired skilled workers should increase following trade liberalization. Second, we expect trade liberalization to lead to a higher degree of upward mobility in the South. In particular, we should observe that, for countries where initially it is mostly skilled workers who send their children to school, trade liberalization would lead to a larger number of unskilled workers to also send their children to school. Equivalently, we should observe that the family background of skilled workers displays an increasing proportion of disadvantaged backgrounds following trade liberalization. Evidence of the latter kind would provide support for our mechanism, but would also provide support for the broader literature on trade liberalization and skill acquisition, in the presence of credit market frictions (e.g. Ranjan, 2003). Finally, our other main empirical prediction is that, following trade liberalization, the average talent of people completing their education in a credit-constrained South should increase. While changes in the talent composition are typically not observable, the most recent literature on the evolution of the US skill premium has devised techniques to identify such changes (see, for example Carneiro et al., 2011, and Carneiro and Lee, 2011). Carneiro and Lee (2011) find that a rise in college enrolment in 1960–2000 was associated with a decrease in the average quality of college graduates, and this had a substantial effect on the college premium. While our focus is on a credit-constrained South, this result is compatible with what the model would have predicted for the impact of trade liberalization on a non credit-constrained North.45 Since our model suggests that the compositional effect of trade liberalization may be even more substantial in the South than in the North – as it implies not only the exit of marginal agents from the skilled labor force, but also the movement of more talented agents into the skilled labor force – we believe an interesting avenue for future research would be to apply these identification techniques to the case of a developing country that has opened up to international trade.