Essays in Health Economics – Habit Formation and Physician Referral

This dissertation examines two consumer and provider behaviors in the health care market – habit formation and physician referral. The existence, extent, and the determinants of these behaviors have profound implications on health policy making.In the first chapter, I extend the literature on habit formation to the theory of health capital and utilization of health care services. Using the Grossman model of health capital, I show that habitual preferences imply intertemporal dependency in utilization and the resistance of demand to price shocks. Using panel data from the Health and Retirement Study I exploit the known biases of OLS and fixed effect estimators in the dynamic panel model setting to bound estimates from the Arellano-Bond model. I find evidence of intertemporal dependency in the use of prescription drugs and dental care visits, but not several other health care services. By leveraging data on episodes of care from the Rand Health Insurance Experiment, I demonstrate that the main findings are not an artifact of supplier-induced demand. I then exploit the exogenous price change in health care services due to Medicare enrollment at age 65 to show that the use of nursing home nights, prescription drugs and dental visits, as well as several preventive care services respond slowly to price shocks. My findings suggest that policies designed to nudge consumption could have long term effects on health care utilization.In the second chapter, I use national physician-pair panel data to assess the effect of switching electronic health record (EHR) developers on patient-sharing patterns among outpatient primary care physicians and specialists. To address the potential endogeneity of individual physicians’ preferences for specific EHR developers, I exploit EHR switching decisions made by the physicians’ affiliated hospitals. I show that primary care physicians (PCP) increase their referrals to same-developer specialists by 5.3% after switching to a new EHR developer. My results imply EHR market agglomeration and a potential misallocation of referrals across specialists of different qualities.In the last chapter, I investigate the gender gap in response to EHR switching. Although researchers and policymakers have highlighted a persistent gender gap, little is known about whether gender plays a role in disparate responses to changes in the work environment. Utilizing a national primary care physician referral panel data, I measure the impact of EHR developer switching on physician referral networks. I model the gender gap in referral behaviors as treatment effect heterogeneity and obtain estimates from double machine learning causal forests. I find that men, generally, maintain more connections with other professionals and adapt to technological changes more readily than women. This pattern is consistent among less-experienced physicians and those affiliated with smaller hospitals. However, the gender gap diminishes and even reverses among physicians in the top experience and hospital size deciles.