Dissertation: Essays on Adverse Selection and Access to Specialty Care in Medicaid


Public health insurance programs in the United States are increasingly administered through the use of private managed care plans. Under this arrangement, private plans compete to offer coverage to beneficiaries, and the government partially or fully subsidizes the insurance premium. In Medicaid, the nation’s health insurance program for low-income individuals and families, more than 70% of beneficiaries are now enrolled in managed care. By transitioning to managed care, state policymakers hoped to achieve higher value from their Medicaid programs. However, a concern in managed care markets is adverse selection, which can result in inefficiently low levels of coverage for services that are valued by sicker, higher-cost beneficiaries. My dissertation investigates whether Medicaid plans are incentivized to restrict access to specialty care due to adverse selection, the extent to which policies like risk adjustment mitigate these incentives, and whether health insurance plans respond empirically by rationing access to selected services.

Chapter 1: Adverse Selection and Access to Specialty Cancer Care

In the first chapter, my coauthors (Timothy J. Layton, Mark Shepard, and Jacob Wallace) and I investigate how adverse selection influences the decision of Medicaid managed care (MMC) plans to provide access to advanced specialty cancer care. We model a health plan’s decision to cover specialty hospitals in settings where consumers do not pay premiums and plan revenues are exogenously determined by a regulator. We demonstrate that the decision depends on two parameters: (1) the take-up of the hospital among inframarginal beneficiaries, and (2) the profitability of the marginal beneficiaries that select the plan due to its coverage of the specialty hospital. We test our model’s predictions using a natural experiment. In 2005, a MMC plan added a well-known specialty cancer hospital to its provider network. We demonstrate that including the cancer center increased the plan’s market share among beneficiaries with cancer by 50%, with no evidence of an increase in market share for Medicaid beneficiaries without cancer. We investigate heterogeneity in the demand response, finding that higher-cost beneficiaries, particularly those with metastatic cancers, were differentially likely to enroll in the plan. The plan dropped the cancer center from its network one year later, and its market share among beneficiaries with cancer declined substantially. While the inclusion of the hospital induced higher-cost beneficiaries with cancer to enroll in the plan, we find no evidence of take-up of the hospital among inframarginal beneficiaries. The results suggest that selection is the primary impediment to inclusion of the specialty hospital in managed care plans’ networks and that the current equilibrium is inefficient.

Note: A newer version of this chapter has been released as an NBER working paper.

Chapter 2: Assessing Selection Incentives by Physician Specialty

In the second chapter, I examine whether Medicaid plans are incentivized to restrict access to physician specialists due to adverse selection. I focus on one of the largest MMC markets in the United States, New York City, and quantify the selection incentives associated with 29 different physician specialties. I first measure incentives in a naive payment system with no risk adjustment. Then, I test the performance of risk adjustment at mitigating incentives. Finally, I examine the incentives associated with each provider specialty using an alternate measure from the literature that accounts for beneficiaries’ ability to predict their specialty utilization, and compare this to my primary measure. I find that adverse selection creates particularly strong incentives for MMC plans to restrict access to oncologists, infectious disease specialists, thoracic surgeons, and neurosurgeons, with plans spending up to 4,000/month on beneficiaries who use these specialties. While risk adjustment and other transfers to plans mitigate these incentives, they remain, with plans losing up to 3,000/month on these beneficiaries even in a setting with risk adjustment and inpatient stop-loss payments. As managed care becomes the predominant mechanism for providing insurance benefits to low-income individuals and families in the United States, it is important that states understand which services might be most subject to under-provision in order to ensure adequate access.

Chapter 3: How Do Plans Respond?

In the final chapter, I test whether MMC plans ration access to adversely selected physician specialties. There are two primary mechanisms through which plans might restrict access to care. The first is by constructing narrow provider networks of selected specialties. Since Medicaid beneficiaries generally do not pay premiums to enroll in coverage and plan benefits are standardized, networks are a salient characteristic for plan choice; thus, plans may use their provider networks as a screening mechanism to deter unprofitable beneficiaries from enrolling. The second is through managed care utilization management techniques. Managed care plans commonly restrict access to services with strategies such as prior authorization requirements. Plans might use such managed care strategies efficiently, in order to discourage low-value care, or inefficiently, to encourage unprofitable beneficiaries to disenroll from coverage. First, I test whether plans explicitly exclude specialists that treat unprofitable patients from their provider networks. Then, I analyze utilization of specialty care by beneficiaries in managed care plans vs. the fee-for-service Medicaid program. I find a statistically significant, but small, relationship between selection incentives and MMC plans’ provider networks. In addition, I find evidence that Medicaid beneficiaries use disproportionately less care in adversely selected specialties when they are enrolled in MMC (relative to FFS).

Click here for a link to my dissertation.