Can Capital-Skill Complementarity Explain the Rising Skill Premium in Developing Countries? Evidence from Peru

Abstract

Abstract: The factors behind the increase in the relative wages of skilled workers in developing countries are still not well understood. The authors use data from Peru to analyze the determinants of within-industry share of skilled workers. They use a translog cost function for gross output and are therefore able to incorporate the effects of materials, both domestic and imported, in addition to capital. The authors find that capital accumulation can explain a large fraction of the increase in the wage bill share and relative wages of skilled labor. This finding is contrary to the commonly held view that unobservable technological change is responsible for the rising skill premium in both developing and developed economies. A test for separability indicates that a gross output cost function is the appropriate one to use, and therefore share equations based on value-added cost functions could be misspecified. JEL classification: F16, J31, O12, O54, E22 Key words: skill premium, capital-skill complementarity, capital accumulation, Peru Introduction Many developing countries have experienced an increase in the relative wages of skilled workers following trade liberalization, as has been well documented by now (see, for example, Robbins 1996, Wood 1997, and Hanson and Harrison 1999). Since this is contrary to the predictions of standard Heckscher-Ohlin model of trade, skilled biased technological change has been cited as a possible explanation for this phenomenon. (1) The fact that the rise in the skill premium has been observed in developing countries as well has also been interpreted as an additional piece of evidence to support the claim that it is technology rather than trade that is driving the rise in relative wages in the industrialized world (Berman and Machin 2000). A common way to demonstrate that skilled biased technological change has taken place is to show that the share of skilled workers in the wage bill within industries has increased over time (Berman, Bound, and Griliches 1994 and Berman and Machin 2000). If the overall wage share of skilled workers in the economy is increasing due to trade, then one would expect to see that reallocations of labor between industries (from those with a low wage share of skilled workers to those with higher wage shares) driving this shift. On the other hand, if technology were the culprit, then these shares would be changing in the favor of skilled workers within industries. The papers cited above find that most the changes can be explained by within-industry changes. (2) Changes in the wage share of skilled labor within industries can be a good measure of shifts in relative demand. An increase in the wage share of skilled labor in the face of a rising skilled wage premium will indicate a shift in favor of this type of labor if we assume that the elasticity of substitution is greater than 1. The literature cited in the previous paragraph has attributed these wage share changes to technology. However, within-industry changes in the share can be driven by factors other than technology. Since technological change is not directly observable, the studies mentioned above interpret the residual component of share changes that cannot be explained by measurable factors as the effects of technology. (3) The factors other than technology that could also influence the shares of skilled and unskilled labor include capital and materials. Skilled labor could be more complementary with physical capital than unskilled labor. Griliches (1969) formalized the hypothesis for such capital-skill complementarity and provided evidence for it. Krusell et al. (2000), in fact, argue that capital-skill complementarity can explain most of the change in the share of skilled labor in the United States (a view that is contrary to the rest of the literature). It is also possible for materials to affect the share of skilled labor. For example, material imports could substitute for unskilled labor in industrialized nations, as Feenstra and Hanson (1999) point out. …