Matching with Phantoms

Searching for partners involves informational persistence that reduces future traders' matching probability. In this paper, traders who are no longer available but who left tracks on the market are called phantoms. We examine a discrete-time matching market in which phantoms are a by-product of search activity, no coordination frictions are assumed, and non-phantom traders may lose time trying to match with phantoms. The resulting aggregate matching technology features increasing returns to scale in the short run, but has constant returns to scale in the long run. We embed this matching function in the canonical equilibrium search unemployment model. Although the model may feature sunspot fluctuations, its typical calibration on monthly US data displays the saddle-path property. The model predicts the same monthly job-finding probability and quarterly aggregate volatility as the standard model with a Cobb-Douglas matching function.