Matching by Luck or Search? Empirical Evidence from the Executive Labor Market

This paper provides a dynamic extension of Peters (2010) directed search model. The point is to characterize the evolution of wage outcomes over time. The primary result of that paper, which is reproduced here, is that workers use random application strategies when they are searching for new jobs. As a consequence, matching markets will be characterized by kind of mismatch of worker and firm types. This mismatch varies in a systematic way with worker types, making it possible to look for evidence of this mismatch in market data. The main predictions are that lower worker types should have a larger variance of lifetime income than high types do, and that there should be a limit on the auto correlation of lifetime income. In particular, these predictions make it possible to distinguish between the model presented here and earlier models, like Peters (2001) where market outcomes are uncorrelated over time, or Eeckhout and Kircher (2010) where outcomes are perfectly correlated over time. The paper then explores a dataset on the executive labor market from 1993 to 2009. Using wage histories to identify the unobservable types of the various workers, the paper proceeds to check some of the main predictions of the model. In addition, the 'type' information recovered from the data can be used to increase the explanatory power of the wage equation by up to 22 percentage points relative to what is is accomplished using observable characteristics alone.