| In this dissertation I examine changes in wage inequality in two chapters. In the first chapter, I examine the slowdown in the relative demand for college-educated labor in the U.S. since the early 1980s. A large literature suggests that this puzzling slowdown is primarily the result of non-monotone changes in the demand for skill, particularly since the mid-1990s, induced by the introduction of computers to the labor market. In these two essays, I develop a complementary result: I show that roughly 10--60% of the gap in the annual growth rates of the relative demand for college-educated workers between the 1963--1982 and 1982--2008 periods can be closed by adjusting for shifts in supply and demand within schooling groups; however, a slowdown in relative demand growth beginning in 1993, well-documented in the literature and potentially-related to recent technological changes, remains pronounced across all specifications.;In the second chapter, I examine changes in relative wages in Brazil. From 1995 to 2002, per capita GDP and schooling wage differentials in Brazil remained relatively stable; during the boom in international commodity prices from 2002 to 2011, per capita GDP grew a robust 2.7% annually while wage differentials along all points of the schooling distribution compressed dramatically. In contrast to a recent literature that interprets these patterns as evidence of trend changes in relative labor demand favoring less-educated workers, I show that a standard two schooling group model of the labor market appears inconsistent with observed movements of relative wages and supplies, potentially implying that demand shifts among male workers in Brazil from 1995 to 2011 were heterogeneous within broad schooling groups. As an alternative to grouping workers solely by their level of schooling, I propose an alternative model that compares three types of workers: low-skill (less-educated, less-experienced), high-skill (more educated, more-experienced), and middle-skill (either more-educated or more-experienced, but not both). |
