Research

The wage gap between black and white Americans has narrowed between the 1960s and the 1970s, but its progress stalled since 1980. This study argues that routine biased technological change (RBTC) contributed to dampening wage gap convergence in 1980-2000, having a differential impact across races and along the wage distribution. Thus, I present new empirical evidence on occupational patterns by race and on determinants of wage disparities along the wage distribution, and rationalize them with an RBTC model in which firms engage in statistical discrimination. I show that, surprisingly, the share of employment in routine intensive occupations has increased for black workers, in contrast with a significant decrease observed for white workers. I decompose the wage gap changes using the Oaxaca-RIF methodology and show that differences in occupational sorting of the workforce increase wage disparities, thwarting wage convergence between races at the bottom of the wage distribution. Together, these new empirical findings and model provide insights to better understand the mechanisms behind racial disparities at the end of the 20th century.

Autor, Levy, and Murnane (2003) ask whether vulnerability to automation,measured by task content, can rationalize employment trends. We invert their approach, asking what technological changes best explain skill-content and employment changes. We combine a tractable GE model with three editions of the Dictionary of Occupational Titles, the 1960, 1970, and 1980 Censuses, and March Current Population Surveys to estimate changes in the relative productivity of skills. We conclude that finger-dexterity productivity grew rapidly while abstract-skill productivity lagged, a form of ‘skill bias’. Together with substitutability between abstract and routine inputs, these results explain changes in skill use within occupations. 

What are the labor market consequences of changing the legal status of the largest immigrant group in a country? We exploit the enlargement of the European Union in 2007 to study the consequences on the Italian labor market of the permanent legalization of immigrants from Romania and Bulgaria. We adopt an IV-DID design, combining the aforementioned natural experiment with variation in the growth of EU07 migrants across local labor markets, instrumented by the change in the share of migrants in a previous legalization episode. We use a unique administrative employer-employee dataset that covers the universe of private sector workers in Italy, to study firms' responses in terms of personnel choices, and how their productivity and investments are affected. We find short-lived effects (2 years) on firm-level employment: an increase for EU07 migrants, at the expenses of natives, accompanied by a rise in hirings and separations for the former, and a corresponding drop for the latter. We provide evidence that the findings are mainly driven by the migrants’ legal status change, rather than by the arrival of new workers in the country. We also observe a reduction in per-capita labor costs, revenues and operative added value, confirming that the effects are probably driven by the legalization of previously undocumented workers.