Choice and consequences, economics of pharmaceuticals and health insurance
MetadataShow full item record
This thesis is composed of three essays on consumer behavior in pharmaceuticals and health insurance markets. The first essay quantifies the effect of changes in local access to medical marijuana dispensaries on the use of prescription opioids in Medicare Part D. While concerns about marijuana as a gateway drug are valid in a small portion of this Medicare sample, in aggregate this effect is dominated by substitution away from opioids. Furthermore, access to marijuana improves uptake and efficacy of opioid replacement therapy for the treatment of opioid addiction. These findings suggest that the expansion of medical marijuana access helps alleviate rather than exacerbate the opioid epidemic at least among Medicare beneficiaries. The second essay estimates the effect of poor adherence to statin medication on medical costs using MarketScan claims data on commercially insured Americans suffering from cholesterol imbalance. Using variation in statin-related side effect rates to instrument for adherence, I find that a 30-day supply of statins induces a 1 percent decrease in total medical costs the following year. Enrollees save proportionally from statin consumption, yet poor adherence is prevalent. Estimates indicate insurers can profit by reducing out-of-pocket cost of statin medication. The third essay explores health insurance choice under the privatized Medicare "5-star Special Enrollment Period" reform of 2012, which allowed beneficiaries to switch to 5-star rated plans outside of open enrollment. Exploiting heterogeneity in exposure to the reform, via the lack of 5-star plans in some markets, we assess the extent to which enrollees switch plans, both within and across years. We find a positive and significant increase in the within-year change in enrollment caused by the reform: switches to 5-star contracts amount to a 7 to 16 percent increase in the contract enrollment base. Contrary to adverse selection concerns, we find that insurer risk pools improve, suggesting that the policy was effective in stimulating 5-star plan enrollment without driving up 5-star plan costs.