We augment the standard matching function by allowing for heterogeneity of job seeker input and introducing a simple model of endogenous search effort. We then decompose changes in the average employment transition rate into changes in inputs and in aggregate matching efficiency. First, the matching function elasticity with respect to vacancies (α) and the groups’ search effort elasticities are not separately identified, i.e., data on group-specific transition rates are consistent with low α and pro-cyclical search effort as well as with high α and counter-cyclical effort. Second, matching efficiency is identified up to a positive scalar. Third, data provide evidence for variable search effort; and the decline in the matching efficiency in the Great Recession was exceptional, even after controlling for variable search effort. Fourth, a model with group-specific variable search effort and only one exogenous shock — aggregate matching efficiency (as opposed to a number of group-specific exogenous shocks) — captures well both co-movement and relative movement in group-specific transition rates. If the variable search effort is shut, as is typical in the literature, the data then require large group-specific shocks to capture transition rates. Finally, in contrast to the standard approach, the matching function with heterogeneity and variable search effort requires larger aggregate matching efficiency for larger α. Joint with Andreas Hornstein.

Key words: Matching efficiency. Search effort. Matching elasticity. Aggregate matching function.