Learning and Job Search Dynamics during the Great Recession

I document two new facts about job search during the Great Recession: (i) Search effort permanently increases after individuals receive (and reject) job offers, and (ii) search effort decreases with cumulative failed search. Motivated by these facts, I introduce a model in which Bayesian job seekers learn about the arrival rate of offers through their idiosyncratic search experiences. The model yields a tractable characterization of search effort in terms of an individual's past job offers and past search effort. I use the model to decompose the effect of learning on job search into static and dynamic components: Failing to find work exerts a negative influence on search by reducing the perceived opportunity cost of leisure in the current period, but also stimulates search by reducing the option value of unemployment in future periods. Because these effects vary endogenously over the spell, the model delivers rich – and potentially nonmonotonic – dynamics in search behavior. I estimate the model and demonstrate that learning accounts for the empirical profiles of search time, offer arrivals, and hazard rates over the unemployment spell.