Occupational Choice and Learning during Job Search: The Role of an Employment Agency (joint with Kenneth Mirkin)

Abstract:
We investigate learning about occupational fit during unemployment, and in particular the role of the employment agency in improving market efficiency. The unemployed are heterogeneous in and imperfectly informed about their ability in each occupation. In each occupation, better searching workers are more likely to be hired. Hence, workers learn about their ability from job search outcomes. Conceptually, this setting entails a bandit model of occupational choice embedded in a frictional labor market equilibrium. We characterize the search behavior that a planner would choose and compare it to equilibrium behavior. Among those with high beliefs in the more productive occupation, those who also have high beliefs in the other occupation should search longer in the productive occupation than they are themselves willing to. In turn, those having low beliefs in the other occupation should begin searching there sooner than they would individually prefer. Similar results are discussed for all other types. Policies that incentivize workers to search over a broader set of occupations exist in most OECD countries, and these results shed light on the types of workers that should be subjected to them and the ones that should not. Efficiency of learning during unemployment has not been yet characterized in the random search literature, and this offers a new perspective on the role of government employment agencies. Similarly, the bandit literature has not given much attention to equilibrium considerations that arise naturally due to search externalities. We prove these results by considering deviations from a symmetric equilibrium allocation in a continuous time model.