Essays on industrial organization and health care markets

Abstract

This thesis comprises three essays on the industrial organization of health care markets. In the first essay, joint with Mert Demirer, I study how formulary-contingent rebates affect insurers formulary placement of branded statins. The prices charged for on-patent, branded pharmaceuticals represent a large, and controversial, component of medical spending in the U.S. In contrast to many countries and many other government programs, drug prices in the Medicare Part D program are determined by privately negotiated rebates between insurance plans and drug manufacturers. How large are these rebates? What would happen to formularies, consumer surplus, and firm profits if the government could increase the rebates of a blockbuster Medicare Part D drug? We estimate a simultaneous model of insurance demand and statin demand for the population of statin users in 2010. Our demand estimates allow us to quantify how insurer profits change under different statin formulary structures.

We use these profit functions to estimate the rebates for Crestor and Lipitor, two blockbuster drugs, using a moment inequality approach; we estimate rebates between 25% and 54% for branded statins. In counterfactuals, we analyze the effect of rebates on formulary design and consumer surplus. We show that increasing only Crestor rebates has no effect on consumer surplus because of offsetting effects on winners and losers. In contrast, increasing only Lipitor rebates does increase consumer surplus. If rebates reduced U.S. prices to match those paid in Canada, then consumer surplus would increase by up to 3.1% In the second essay, I compare estimates of formulary-contingent rebates using three empirical moment inequality models. Unobserved private rebates are an important determinant of the prices that insurers pay drug manufacturers in Medicare Part D. There is growing interest in understanding these negotiated rebates and there consequences on market equilibrium.

However rebates are secret and have proven difficult to estimate. In this paper I compare three moment inequality models that I use to estimate formulary-contingent rebates for branded statins. The first model, which only allows for measurement error, imposes the strong assumption that their is no rebate heterogeneity that is unobserved to the econometrician. Due to the fact that different insurers use different agents (Pharmacy Benefit Mangers) in rebate negotiations, this assumption is unlikely to hold. As a consequence I develop two models that allow for unrestricted insurer-specific unobserved rebate heterogeneity. Somewhat surprisingly, the measurement errors only model produces reasonable results in this context, however the rebates for Lipitor are approximately twice as large in my preferred model relative to the measurement errors only model.

In the third essay, also joint with Mert Demirer, I study the effects of government negotiated drug prices using Nash-in-Nash bargaining models. One of the most controversial aspects of Medicare Part D is that the government is prohibited from being involved in price negotiations despite the fact that it provides almost $100 billion to the program in subsidies each year. We model pharmaceutical drug price setting using Nash-in-Nash bargaining models. We compare two models: one where insurers negotiate drug prices and another where the government negotiates prices. We show that the ability for the government to improve consumer surplus depends on both upstream and downstream market structure. With many insurers and few drug manufacturers, the government can increase consumer surplus, but with few insurers the government cannot increase consumer surplus no matter how much bargaining power it has vis-a-vis drug manufacturers.

We also show that a Nash-in-Nash bargaining model where insurers and drug manufacturers negotiate over both manufacturer prices and copays can be used to estimate unobservable manufacturer prices and bargaining weights as long as there are profit spillovers across bilateral negotiations.