EMPIRICAL ANALYSIS OF INCENTIVES AND SELECTION IN HEALTH CARE MARKETS

Abstract

The common theme throughout these collection of essays is the analysis of incentives and selection effects in health care markets. The first chapter investigates the effect that a behavioral nudge when choosing health care plans can have on risk selection into High-Deductible Health Plans (HDHPs). I use novel administrative data on enrollee plan choices and claims from a large employer that implemented an automatic enrollment policy, changing the default option to the company’s HDHP. My findings show that the automatic enrollment policy more than doubled the fraction of HDHP enrollees but did not mitigate the risk selection of low-cost individuals. I then formulate a machine learning frameworkto test for the presence of private information, where the results confirm a significant degree of private information leading to the overprediction of costs for HDHP switchers. The second chapter studies how changing the cost-sharing differences between in-network and out-of-network care under a health insurance plan affects how spending and utilization is allocated across in versus out-of-network providers. I exploit a quasi-experimental design based on a large employer that merged employees on the PPO and POS health plans together, holding fixed the network design while changing the relative cost-sharing generosities between in and out-of-network care. The results indicate that changing the cost-sharing attributes had an insignificant effect on in-network and out-of-network spending. The third chapter investigates how network breadth is affected by the number of insurance options in California’s Medicaid Managed Care system. Using within county variation in the number of insurers over the years 2002-2011, I estimate a 0.515 percentage point decrease in the probability that a hospital is in a plan’s network for every insurer a county adds. The results suggest that insurers are disincentivized from offering a broad network as they face more competitors.