Labor Market Frictions and Dynamic Labor Demand.

Abstract

This dissertation explores the microeconomic and aggregate implications of an array of employment adjustment frictions. Chapter 1 investigates the JOBS Bank, a unique employment adjustment cost that prevailed in the domestic automobile industry for nearly 20 years. The JOBS Bank required manufacturers to pay full salary to a worker for each week spent on layoff beyond an allotment specified in the labor contract. The paper presents a model in which JOBS generates an option value of production: the firm produces more often than otherwise to safeguard its allotment of layoff weeks in case future vehicle demand deteriorates within the life of the contract. The paper tests this implication on plant-level data, and reduced-form analysis rejects the qualitative predictions of the model. To understand this result, the paper then estimates the structural model by indirect inference. This exercise enables a series of counterfactual simulations that indicate the features of the plant's environment which may have muted the effect of the JOBS Bank. In the models of Chapters 2 and 3, the firm faces a different set of adjustment frictions. In Chapter 2, the firm must post vacancies in order to match with unemployed workers and pays a cost to advertise each opening. The model allows for non-linear production technology and both idiosyncratic and aggregate risk. In a set of quantitative applications, the model is shown to provide a coherent account of a) the steady- state distributions of employer size and employment growth across establishments; b) the cyclicality of flows between employment and unemployment; c) the negative comovement of unemployment and vacancies; and d) the dynamics of the employer size distribution. Chapter 3 investigates the analytics of labor demand in the presence of a fixed cost of employment adjustment. It shows how the forward-looking policy rule nests the solution of the corresponding static, or myopic, problem. The paper then demonstrates that, for reasonable parameterizations of the model, the myopic policy provides a remarkably accurate approximation to forward-looking labor demand. This suggests that the myopic rule may serve as a useful guide to the mechanics of a rich class of dynamic models.