The assumption of constant returns in the matching function, embodied in mo
st bilateral search models, is crucial to ensure the uniqueness of the unem
ployment rate along a steady-state growth path. This article explores the e
mpirical viability of this assumption by estimating individual reemployment
probabilities on a sample of unemployment entrants. I apply hazard models
to survey data on both completed and uncompleted unemployment durations. Th
e hypothesis of constant returns to matching is not rejected, on the basis
of the evidence that the job-finding hazard depends only on local labor mar
ket tightness and is independent of its size.