Why firms pay different wages? Empirical evidence suggests substantial differences in firm pay controlling for worker skill. Moreover, these differences are uncorrelated with worker skill suggesting the lack of sorting. First, we show that the face value interpretation of the facts is inconsistent with evidence on coworker segregation, which we document using Brazilian matched data. We interpret the evidence with a model of sorting and show that the first correlation is biased. We isolate the reason for the bias in non-monotonicities in wages, and show that the coworker moment measures sorting accurately. The model explains many facts about labor markets, but misses some moments for the same reason as the bias.

This paper circulated with a previous title "Sorting in the Labor Market: Theory and Measurement".

Job-search theory has been used successfully to explain cross-sectional patterns of wages and mobility patterns of workers. This paper confronts central implications of these models with a robust and pervasive fact: real wage cuts upon job-to-job transitions (JJT) are more frequent that for those remaining in one job. We provide two tests of the workhorse model of the related literature, both self-reliant of parametric assumptions. The resulting rejection suggests that wage cuts in JJT are not due to measurement error alone. We investigate these cuts further by examining the subsequent wage performance of these workers, concluding that this only explains the cuts of the highly educated.


"Wage Dynamics and Distribution in an Equilibrium Search Model"

We estimate the equilibrium search model in Moscarini (05) using data from the NLSY-79.
Rafael Lopes de Melo,
Mar 19, 2013, 10:24 PM