UCLA

All three chapters of this dissertation study different aspects of internal organization, and two of the chapters go a step further to understand how decisions within the firm impact markets outside the firm.

The first chapter of this dissertation studies empirically how worker task assignments within firms interact with labor and product markets. Using data from a software company, I observe millions of task assignments within hundreds of hair salons, many of which are competitors. I develop a measure of organization complexity, which is the amount of information required to implement a given task assignment, to provide evidence of firm-specific organization costs, which grant complex salons a comparative advantage in producing high-quality products.

Based on these facts, I develop a model where oligopolistic firms with different organization costs choose their internal structure. Complexity is costly, but it allows firms to improve product quality by better matching workers with multidimensional skills to tasks. I characterize the profit-maximizing organization, and use results from the literature on rational inattention and information theory to identify and estimate the model for Manhattan hair salons. Counterfactuals reveal that allowing internal organization to be heterogeneous and endogenous changes the equilibrium effects of policy. A sales tax cut increases specialization and therefore the productivity of all workers, while a minimum wage increase generates new types of wage spillovers.

The second chapter of this dissertation studies empirically how voluntary labor supply decisions within an organization impact workplace injury rates using novel data on the payroll and workers' compensation claims of Los Angeles traffic officers. I use the leave taken by coworkers as an instrument to estimate the causal effect of daily labor supply decisions on workplace injury. Self-selection via voluntary labor supply reduces injuries by 48 percent compared to the underlying injury rate. The majority of the effect is driven by private factors, implying decentralized overtime assignment mechanisms like shift auctions can be used to reduce injury rates.

The third chapter of this dissertation (joint with Stepan Aleksenko) studies theoretically how the use of recruiters impacts the types of workers hired. The chapter considers a model where a firm delegates search for a worker to a recruiter. Productivity is uncertain prior to hire with recruiter beliefs characterized by an expectation and variance. Delegation occurs using a refund contract which is common in the industry. We analyze how delegation in this setting shapes search behavior and the composition of hires. We demonstrate that delegation is theoretically equivalent to making the search technology less accurate. This generates inefficiency: search effort and social surplus are lower under delegation than in the first-best benchmark. We show this inefficiency is driven by moral hazard with a multitasking flavor. The recruiter wastes search effort finding low variance workers at the expense of high expectation workers. As a result, as workers become more homogeneous with respect to productivity variance, delegation becomes more efficient. Our model provides a microfoundation for variance-based statistical discrimination.