Dissertation Title： "Three Essays on the Economics of Medical Devices"

Financial relationships between physicians and industry are vital to biomedical innovation, yet create the potential for conflicts of interest in medical practice.

In the first thesis paper, I propose two models for the role of these relationships in medical technology markets, each with different implications concerning the need for regulation of the relationships. In the inducement model, financial relationships induce physicians to use more of the devices of the firms they consult for, which medical technology firms exploit by developing relationships with their most valuable physician customers. Oligopolistic firms use consulting payments to physicians to compete for market share, but coordinated restrictions on all firms' payments allow profits to increase. In the service model, firms compensate those physicians who provide them expertise in product design and development and in training other physicians; payments have no influence on medical practice. To test the models, I exploit a policy shock, whereby government monitoring of payments to joint replacement surgeons resulted in declines of over 60% in both total payments and in the number of physicians receiving payments from 2007 to 2008. Using hospital discharge data from three states, I find that the loss of payments leads physicians to switch 7 percentage points of their device utilization from their sponsoring firms' devices to other firms' devices. Altogether, tests of five key predictions support the inducement model, four of which refute the pure service model. I also find evidence of short-term increases in medical productivity following the policy intervention, although long-term effects on productivity remain unclear. Finally, I provide conditions under which regulation of financial relationships would be socially beneficial.

The second paper explores two inter-physician forces that drive the adoption of new technologies in medicine. Peer effects occur where individuals’ adoption decisions influence those of other members of their social networks, most often in a positive direction. These effects are strongest among physicians within hospitals, the health care organizations that structure physician work and social networks. Yet specialist physicians are also firms in imperfectly competitive health care markets, and strategic adoption of technologies their competitors do not also use may allow them to differentiate themselves, increasing their market power and softening price or quality competition. This would lead to negative effects of doctors’ adoption decisions on those of other doctors across their entire market. I estimate the magnitude of these countervailing positive and negative effects among surgeons adopting orthopedic implant technologies in two states, allowing the net effects to differ between physicians within and across hospitals. I find strong peer effects within hospitals for five out of seven technologies, but substantial heterogeneity in inter-physician effects across hospitals, with service differentiation and peer effects alternately dominating for different technologies. This suggests that a technology-specific policy approach to health care variations within markets may be neccessary. Interactions between hospitals and physicians have profound implications for the operation of health care markets, but our understanding of them is limited by the lack of observable outcomes of these interactions.

In the third paper, Jonathan Kolstad and I propose that the observable payments that implantable medical device firms make to physicians -- plausibly to induce utilization of their firm’s devices -- may respond to hospital-physician interactions and offer a window to shed light on these relationships. We develop a model of hospital-physician bargaining over the utilization of medical devices, then show that device makers concentrate their payments in markets where physicians have the greatest relative bargaining power advantage over hospitals, in order to maximize the payments’ impact on device utilization. We test the theory using nationwide data on hospital market concentration and on payments by orthopedic device firms to physicians. Results support the theoretical prediction, but also indicate high payments in markets where physicians have the lowest relative bargaining power, so that payments display a U-shaped relationship with physician relative bargaining power. We suggest possible extensions to the model to fully explain the empirical findings.