Dissertation Title： "How Does Managed Care Manage Care? Evidence from Public Insurance"In the United States, the fraction of individuals with public insurance is growing and the policies and markets that serve them are changing. Over two-thirds of Medicaid recipients are now enrolled in managed care organizations (MCOs), but little is known about how these plans operate. To study this market, I use data from New York Medicaid where managed care recipients are randomly assigned to plans. In chapter one, I estimate how physician and hospital networks impact health care use and spending. In chapter two, I study how production and selection vary across the managed care plans in the market. Finally, in the third chapter, I turn to the Medicare program and examine how health care use and spending change at age 65 as adults switch from private to public coverage. Taken together, my dissertation chapters provide new perspective on how differences within public programs and between public and private programs, impact health care consumers.

In chapter one, "Do Provider Networks Impact Health Care Use and Spending? Evidence from Random Assignment in Medicaid Managed Care," I use the random assignment of Medicaid recipients to health plans to assess the causal impact of provider network breadth on health care outcomes. A key distinction between the managed care plans that serve Medicaid recipients is the breadth of their provider networks (i.e. the number of providers they contract with). To study this, I use data from New York Medicaid where managed care recipients are randomly assigned to plans. Each plan contracts with a different set of physicians and hospitals, providing an ideal setting to study the impact of network breadth. To measure network breadth, I use estimates from a structural model of demand for physicians and hospitals. Combining these measures with administrative health records for over 100,000 randomly assigned Medicaid recipients, I find that the effects of physician and hospital networks differ significantly. Using variation in networks within plans, I find that broader physician networks are associated with increases in utilization and spending, fewer avoidable hospitalizations, and greater plan loyalty. Although also associated with greater plan loyalty, broader hospital networks have no effect on utilization or spending. Regulations that encourage broad networks should account for the tradeoff between improved access and higher spending.

In chapter two, "Are all Managed Care Plans Created Equal? Evidence from Random Plan Assignment in Medicaid," (with Mike Geruso and Tim Layton) I open the black box of managed care and study how health plans competing in the same market may vary in their approaches--and ultimately in their ability--to constrain healthcare spending. I examine this issue in the context of Medicaid Managed Care in New York City, in which some beneficiaries make active plan choices across a large number of privately-operated managed care plans, and other beneficiaries are randomly auto-assigned to the same set of plans. I exploit the random assignment to identify plan effects that are purged of selection, which I show would be an important confounder in this setting. My findings reveal significant variation in both "production" and "selection" across ostensibly similar managed care plans competing in the same local market. Specifically, I find that plans differ in spending on identical beneficiaries by as much as 33%. These differences are even larger for recipients with high baseline spending. I show that differences in negotiated upstream prices explain only a fraction (3-10%) of this difference, and that plan characteristics significantly affect the healthcare consumption of recipients. These findings are important for the continued reform of healthcare, in which managed care is often touted as the single most important tool for constraining healthcare spending growth.

In chapter three, "Traditional Medicare Versus Private Insurance: How Spending, Volume, and Price Change at Age 65 for the Previously Insured," (with Zirui Song) I examine how spending, volume and price differ between public and private insurance. To slow the growth of Medicare spending, policymakers have advocated raising the Medicare eligibility age. Despite its policy importance, little is known about how this proposal would impact national health care spending. For the majority of affected adults, the proposal would delay entry into Medicare and leave them with private insurance. We examine how health care spending differs between Medicare and private insurance by exploiting longitudinal data on imaging and surgery for a national cohort of individuals that switch from private insurance to Medicare at age 65. Using a regression-discontinuity design, we find that spending falls by $38.56 per beneficiary per quarter (32.4%) upon entry into Medicare at age 65 (P<0.001). In contrast, we find no changes in volume at age 65. For the previously insured, entry into Medicare leads to a large drop in spending driven by lower provider prices, which may reflect Medicare’s purchasing power as a large insurer. For the majority of adults affected, these findings imply that an increase in the Medicare eligibility age would raise national health care spending by replacing their Medicare coverage with private insurance where health care spending is higher due to higher provider prices.