working papers

The Inverted Job Ladder in Skilled Professions [PDF]

How do workers initially match with firms, and how do these matches improve over time? A large job ladder literature devoted to this question proposes a unanimous surplus-ranking of firms in which poached workers move to better firms while displaced workers move to worse firms. Using a new historical dataset on lawyers, I rank law firms based on where their lawyers went to school, finding that poached lawyers move to worse firms while displaced lawyers move to better firms. Guided by these and several other stylized facts, I propose an alternative theory to the standard job ladder approach. In my model, each worker's surplus-maximizing firm assignment is a function of her talent. Incumbent firms privately learn how talented their workers are, and thus only allow adversely selected worker types to be poached. In equilibrium, poached workers move down in rank (to firms where they are more productive). Meanwhile, workers who are retained are revealed over time to have been under-placed, so random displacement shocks move them up in rank (to firms where they are more productive) by temporarily removing the adverse selection problem. The model is well-suited for quantifying the value of labor market institutions that publicly certify talent. By estimating the model, I find that more than 20% of output is lost to misallocation induced by informational frictions. Pre-job market screening devices can substantially raise average earnings, creating a rationale to regulate the timing of job recruitment to prevent it from disrupting informative competition in academics.

Collusive Capacity, [PDF] with Daehyun Kim

It is widely believed that cartels with too many members are destined to fail. The standard argument is that as the number of cartel members increases, shares of collusive profit diminish relative to deviation profits. We show that this argument is built on unreasonable assumptions about plant capacity. We add plant capacity choices to an otherwise standard dynamic oligopoly game. We consider the unsophisticated and easily enforced strategy in which each firm simply chooses plant capacity equal to static Nash output. Our main result is that as the number of firms goes to infinity, the critical discount factor required to sustain collusion on the monopoly price converges to less than 0.63. Thus, collusion is (quite) robust to the number of firms. This result applies to a broad class of demand functions and to both Cournot and Bertrand competition.