Empirical matching functions: Searchers, vacancies, and (un-)biased elasticities

This paper shows that conventional empirical estimates of matching functions are systematically biased, because unobservable and endogenous search behaviour, such as on-the-job search by workers or the use of different search channels by firms, is neglected. I propose an approach for recovering unbiased elasticities under different scenarios of competition for workers and jobs without the need for constructing proxies for unobserved stocks of searchers and vacancies. Using German administrative data, I estimate the biases affecting conventional studies and derive bias-corrected estimates of the matching elasticities under different scenarios.