In many industries, multi-product sellers offer bundled loyalty discounts to buyers who refrain from purchasing competing products from rivals. These contracts can increase sales volumes by helping sellers establish a strong customer base across their product portfolios, but they also risk foreclosing competing sellers, even those offering superior products. I examine the welfare implications of bundled loyalty discounts in the context of physician-administered vaccines, which medical providers purchase and administer directly to their patients. I use prior antitrust cases involving vaccines to recover the proprietary terms of bundled discount contracts. I then develop a model of medical providers’ demand and pharmaceutical companies’ supply of vaccines. Using all-payer claims data from Colorado to estimate the model, I recover medical providers’ preferences, vaccines’ marginal costs, and the magnitude of bundled discounts that rationalizes providers’ observed purchasing patterns. I use the model to evaluate the equilibrium and welfare effects of banning bundled discounts, allowing sellers and buyers to re-optimize their choices. I show that, in the absence of discounts, providers purchase fewer and lower-quality products, particularly at the expense of optional vaccines not required for school access. Insurers benefit in the short term through lower immunization coverage costs. The effect on pharmaceutical companies varies, depending on the scope and composition of their vaccine portfolios.

U.S. pharmaceutical companies invest nearly $100 billion annually in R&D, while the National Institutes of Health (NIH) allocates around $50 billion to biomedical research. Experts argue these investments should reflect public health needs and scientific opportunity, yet systematic evidence on allocation patterns remains limited. Using machine learning, we compile a novel panel dataset linking NIH funding and industry clinical trials across major diseases over two decades to disease-specific characteristics: burden, scientific opportunity, revenue potential, and the race and gender composition of affected populations. We find that both sectors invest more in high-burden diseases, but private firms also prioritize high-revenue conditions. Strikingly, the private sector underinvests in diseases that disproportionately affect Black Americans—even after adjusting for burden, revenue, and opportunity. NIH funding partially offsets this gap but compensates for only about 10% of the dollar shortfall and 80% of the "missing" trials for conditions affecting Black Americans. We identify disease areas where gaps remain largest, offering a roadmap for targeted policy action. Our findings highlight NIH’s essential—but incomplete—role in correcting racial disparities in biomedical innovation and show how recent cuts to NIH and Medicaid risk deepening these inequities.

I provide novel evidence that physicians respond to changes in insurance reimbursement for in-office treatments, although the direction and potential spillover effects of these responses may vary. To estimate physician elasticity to private insurance reimbursement, I exploit a plausibly exogenous regulatory change in Medicare payment. Specifically, I document how private reimbursement typically follows Medicare reimbursement (Clemens and Gottlieb, 2016; Chan and Dickstein, 2019) and use the estimated payer-specific relationship between the two prices to construct a valid instrument. My findings show that supply generally exhibits a positive slope—lower reimbursement reduces treatment provision. However, for certain discretionary treatments, the supply curve is negatively sloped. I provide suggestive evidence that physicians compensate for reduced reimbursement by increasing provision of these and related treatments that can be easily combined in single patient visits.