Abstract
We develop a theoretically grounded extension of the two-way fixed effects model of Abowd et al. (1999) that allows firms to differ both in the wages they offer new hires and the wages required to poach their employees. Expected hiring wages are modeled as the sum of a worker fixed effect, a fixed effect for the “destination” firm hiring the worker, and a fixed effect for the “origin” firm, or labor market state, from which the worker was hired. This specification is shown to nest the reduced form for hiring wages delivered by semi-parametric formulations of the sequential auction model of Postel-Vinay and Robin (2002b) and its generalization in Bagger et al. (2014). Using Italian social security records that distinguish job quits from firings and layoffs,we demonstrate that our fixed effects model captures well differences in wage growth between workers involved in voluntary and involuntary job transitions. Bias correcting a variance decomposition of hiring wages, we find that origin effects explain only 0.7% of the variance of hiring wages among job movers, while destination effects explain more than 23% of the variance. Across firms, destination effects are more than 13 times as variable as origin effects. Interpreted through the lens of the sequential auction model, this finding requires workers to have implausibly strong bargaining strength. Studying a cohort of workers entering the Italian labor market in 2005, we find that differences in origin effects yield essentially no contribution to the evolution of the gender gap in hiring wages, while differences in destination effects explain the majority of the gap at the time of labor market entry. Our results suggest that where a worker is hired from is relatively inconsequential for his or her wages in comparison to where he or she is currently employed.