Healthcare is a large and important part of the United States economy. In 2013, the healthcare sector accounted for 17.4% of GDP and employed 13% of the US workforce. Healthcare expenditures grew at an annual rate of 3.6%, faster than both the economy’s growth rate (2.2%) and the rate of inflation (1.5%). As Americans spend more on healthcare, policymakers are faced with the challenge of guaranteeing health insurance coverage to all Americans, while also making it affordable (Sanger-Katz et al. 2014). However, research shows that the US spends more on healthcare than any other OECD country, and yet our health outcomes are no better than most OECD countries (Squires 2012). Therefore, policymakers are looking to reform the healthcare system in ways that will make it more efficient and improve the quality of care that it provides. My research focuses on how healthcare providers decide to treat their patients. It contributes to a growing literature in health economics that identifies inefficiencies in the healthcare system by identifying inefficiencies in how providers treat patients. Examples of inefficiencies include situations in which healthcare providers supply medical care that does not provide health benefits or situations in which the same medical care could have been provided in a lower cost setting. Inefficiencies persist in healthcare markets due to information asymmetries. Healthcare providers know more about medical care than their patients and patients are often unable to differentiate between high and low-quality providers. As a result, research has shown that similar patients often receive different care from different healthcare providers. Motivated by the idea that the same patients can receive different medical care from different healthcare providers, my dissertation discusses three reasons why physicians have different practice styles. The first chapter focuses on the fact that physicians have different medical training and experience. The second chapter discusses how physicians are exposed to different degrees of medical malpractice liability. The third chapter describes how physicians are members of different health insurance networks. I find that all three reasons explain physician practice style, but that health insurance networks have the largest effects on healthcare efficiency. As a health economist, it is important to use quasi-experimental research methods to isolate the determinants of physician practice style, so each chapter in my dissertation relies on a different quasi-experimental method to show how physicians make decisions involving patient care. These methods help to overcome the biases inherent in health economics research, such as when patients select their physicians or when physicians select their patients. My dissertation also uses detailed data on physicians and patients to show not only how patient care varies across physicians, but also how patient care varies within physicians over time. In this way, my dissertation reveals the degree to which physician practice style is malleable and responsive to incentives. In the following paragraphs, I provide a brief description of each dissertation chapter and I discuss my goals for future research. Chapter 1 asks whether physician characteristics explain efficient practice style, where an efficient physician generates the lowest possible costs for a given set of health outcomes. This chapter was motivated by the economics of education literature where researchers seek to find the characteristics of effective teachers. Chapter 1 focuses on patients with minor injuries who should not receive different medical care from different physicians. The quasi-experimental method uses the fact that, within hospital emergency rooms (ERs), patients with minor injuries are as good as randomly assigned to ER physicians. The results reveal that the only observable physician characteristic that explains differences in practice style across physicians within the same ER is physician experience. Experienced physicians prescribe fewer procedures and charge less per visit, with no differences in health outcomes. Similar to the teacher experience literature, however, the gains to physician experience are largest within the first two years and then they quickly taper off after that. Therefore, inexperienced physicians are the least efficient providers of urgent, but routine healthcare. Chapter 2 pivots away from physician characteristics and towards malpractice liability. Several papers have shown how changes in tort law affect physician practice style, but no papers have shown whether physicians change their labor supply in response to their own malpractice claims. This chapter uses an event study method and a propensity-score matched difference-in-difference method to show how physicians adjust their labor supply in response to their own malpractice claims. Malpractice claims represent adverse medical events and I measure physician responses from the date that the event was reported. I focus on physicians who regularly treat hospitalized patients and I test whether they are less likely to work in hospitals after they receive malpractice claims. The results reveal that physicians treat fewer patients as attending physicians in hospitals after they receive claims. The same physicians, however, do not reduce their outpatient volume or their operating room volume. Moreover, physicians with malpractice claims related to patient deaths drive the labor supply responses. The labor supply responses persist for at least three years, suggesting that some types of malpractice claims can have permanent effects on physician practice style. Chapter 3 investigates whether patients receive different hospital care depending on their health insurance coverage. The quasi-experimental method relies on a Florida Medicaid reform that mandated that Medicaid beneficiaries in certain counties switch from the state’s fee-for-service (FFS) system to managed care plans. One of Chapter 3’s goals is to see whether Medicaid patients received different hospital care post-reform. I find that some Medicaid managed care patients received lower-cost hospital care compared to the FFS patients, primarily because the managed care patients were treated by lower-cost ER physicians. However, I also find that managed care plans in different markets constructed their physician networks in different ways. For example, I find that health maintenance organizations (HMOs), which are owned and operated by insurance companies, were less likely to expand their low-cost ER physician networks when they faced competition from a vertically integrated, hospital-owned plan. This finding explains why not all Medicaid managed care patients received different hospital care; HMOs were not able to contract with low-cost ER physicians in every market. Though most of my research has been focused on physician-level decision-making, my interests are evolving to include market-level analyses. In addition to determining how physicians respond to changes in incentives, I would like to understand why incentives change in the first place. Over the next year I will work with two new data sets, one of which I will use to provide additional evidence on how managed care plans choose their provider networks in different markets. The other data set I will use for a project related to financial incentives and physician prescribing behavior. In addition to these two projects, I am interested in managed care plan entry and exit into public health insurance markets and vertical integration in public health insurance markets. There is very little empirical research on vertical integration in healthcare markets because it is a relatively new phenomenon. Nevertheless, Gaynor, Ho, and Town (2014) write, “Various forms of restraints and integration between physicians, hospitals, and insurers are being developed, which provides opportunities for industrial organization economists to learn about the impacts of these arrangements."