Empirical Matching Functions: Estimation and Interpretation Using Disaggregate Data

In this paper, we estimate matching functions using disaggregate data. We find strong support for the matching approach, with most specifications implying slightly increasing returns to scale. This finding does not appear to arise from our inclusion of additional controls or from the level of disaggregation, and so we conclude that earlier findings of constant returns in the US may be due to the various approximations needed to construct an aggregate time series. We also find evidence of endogenous job competition between the employed and nonemployed, so that the estimated parameters from a matching function cannot be interpreted as structural parameters.