Research

Publications 

Working Papers

We draw upon newly merged administrative data sets to study the relationship between payments from medical technology firms to physicians and medical device procurement by hospitals. These payments (and the interactions that accompany them) may facilitate the transfer of valuable information to and from physicians. However, they may also influence physicians’ treatment decisions, and in turn hospital device procurement, in favor of paying firms. Payments are pervasive: 87 percent of device sales in our sample occurred at a hospital where a relevant physician received a payment from a device firm. Payments are also highly correlated with spending within a firm-hospital pair: event studies suggest that a large positive increase in payments to a given hospital from a given firm ($438 per physician on average, or 112 percent of the mean) is associated with 27 percent higher expenditures on the paying firm’s devices post-event. Finally, we explore how payments mediate the relationship between expertise and device procurement patterns. Hospitals affiliated with the top Academic Medical Centers (AMCs), which plausibly represent an expert benchmark, purchase a different mix of devices than other hospitals, and payments to hospitals outside the top AMCs are correlated with larger deviations from the procurement patterns of top AMC hospitals.

I provide estimates of the causal effects of payments from pharmaceutical companies on the prescribing habits of Medicare Part D physicians. Identifying the causal effects of payments associated with pharmaceutical "detailing" (marketing to physicians) is confounded by dynamic selection in and out of payment assignment by the drug producer. I employ a novel identification strategy which uses exogenous variation in brand-related payments due to pharmaceutical producers undergoing acquisitions by other firms. The analysis focuses on aminosalicylates, the drug class used to treat inflammatory bowel disease, and finds that stopping payments to a physician reduces the probability that she will prescribe any drug from the aminosalicylates class by around 5 percentage points. Decomposing payment effects to persistent and temporary components, I find that at most 50 percent of brand-specific effects and 80 percent of class-specific effects are persistent. Payment effects in other drug classes seem to exhibit similar patterns.

We propose and test a model of the strategic interaction between public and private insurers in the physician services market. We depart from the standard healthcare service pricing model and allow physicians to (partially) adjust patient access based on price differences between insurers. Analysis of private and public sector insurer prices and public sector quantities supports the hypothesis that physicians respond to insurer price differences: A 10% increase in private prices decreases public beneficiaries’ share of physician services by 4%, equivalent to a per capita decrease of 1%-14% in services per public beneficiary, ceteris paribus.

I estimate the effects of interscholastic athletic competitions on student performance in public high schools in Texas. I use discontinuities in assignments of schools to interscholastic athletic conferences to identify these effects, and utilize the dynamic RD estimator developed by Cellini et al. (2010) to overcome identification challenges resulting from persistent conference assignments. I find that increased levels of competition have an adverse effect on student performance, with school average ACT scores decreasing by 1.4 percent and rate of students passing the TAKS exam dropping by 3 percentage points. Potential effect mediators are discussed and analyzed.

Works in Progress