Explaining frictional wage dispersion

Matching the magnitude of frictional wage dispersion has been difficult for most frictional search models of the labor market; only 1/30–1/4 of the frictional wage dispersion observed in the data can be accounted for using realistic calibrations. In this paper, I develop a general equilibrium model with ex ante homogeneous workers but heterogeneous firm productivities that accounts much better for the magnitude of frictional wage dispersion. The key ingredient that makes this model different from wage posting mod-els is the bargaining and the wage determination structure. When a firm meets an employed worker, the new firm and the incumbent firm engage in Bertrand competition in the worker’s share of the match surplus; consequently the cur-rent wage of the worker depends not only on the firm’s productivity, but also on the history of previous offers. I demonstrate that the latter generates about 3/4 of the frictional wage dispersion, and that standard wage posting models cannot achieve similar magnitudes of frictional wage dispersion with the cali-bration used in the paper. In contrast to the structural estimation literature, the parametrization of the model is very parsimonious. The model is calibrated to match the worker flows, the standard deviation of log productivity and the replacement ratio, and it delivers frictional wage dispersion that matches the data very closely. ∗I would like to thank Per Krusell for advising me in this work. Marc Melitz, Nobu Kiyotaki, Esteban Rossi-Hansberg and Christopher Sims also deserve special recognition. I am grateful for comments from Alisdair McKay, Mikael Carlsson, Felipe Schwartzman, Ádám Zawadowski, and the seminar participants at the Sveriges Riksbank, the Macro Seminar at Princeton University, and audiences at my job market flyouts. I would like to thank the Sveriges Riksbank for financial and research support during the summer of 2008. The customary disclaimer applies.