Essays on Provider Behavior in Health Care Markets

This dissertation consists of three essays that investigate the determinants of hospital and physician decisions and the consequences of those decisions for social welfare.;Chapter 1 examines how physicians combine their own experience with evidence from the clinical literature to form beliefs about the probability of rare complications. Focusing on a particular obstetric context, I present event study evidence that physicians respond to their own idiosyncratic experience. I combine this event study evidence with an analysis of the available clinical literature in order to calibrate a model of physician learning. I find that the calibrated model can account for much of the cross-sectional variation in practice patterns in this setting and may be able to account for the dramatic changes in practice patterns in this setting over the last three decades.;Chapter 2 (which is co-authored with Amitabh Chandra) examines the welfare implications of hospital adoption of percutaneous coronary intervention (PCI) since the early 1990s. We find that PCI adoption over this period was approximately welfare neutral, but would likely have generated positive net benefits if utilization had been limited to high-benefit patients. We also find that hospitals' profits on high-benefit patients are not sufficient to cover the fixed costs of adoption. Achieving the efficient outcome would therefore require policy changes that both discourage utilization among low-benefit patients and allow hospitals to capture more of the surplus generated among high-benefit patients.;Chapter 3 exploits day-to-day fluctuations in patient demand to estimate the marginal return to increasing per-patient hospital resources. Such estimates can shed light on the benefits of additional hospital capacity and on theories about the cross-sectional pattern of productivity in medicine. I find that patients arriving when a hospital has a large number of patients in its intensive care unit are less likely to be admitted to intensive care beds and more likely to face delays in the receipt of time-sensitive care. The hospital responses are consistent with a model in which hospitals ration care based on expected clinical benefit. There is no evidence that hospital congestion adversely affects health outcomes, but firm conclusions will require more data.