Please click here if you are unable to view this page.
TOPIC:
PHYSICIAN LEARNING AND TREATMENT CHOICES: EVIDENCE FROM BRAIN ANEURYSMS
ABSTRACT
I examine how two types of physician learning jointly shape their medical decision-making: Bayesian learning that updates beliefs about treatment-patient match quality and learning by doing that improves surgical skills. Using discharge-level data on the history of brain aneurysm treatments by over 400 new physicians in New York State, I first document empirical evidence that both types of learning are present and affect the choices of forward-looking physicians. I then develop and estimate a dynamic structural model of learning and treatment choices for heterogeneous patients. I quantify the impacts of learning and find that (i) learning leads to a substantial deviation from the myopic best choices, hurting short-term patient outcomes but improving the overall treatment success rates by 18-25%; (ii) learning explains about one-third of the supply-side variation in the choices of care, with Bayesian learning contributing to the convergence of choices and learning by doing the divergence of choices. In the counterfactual experiments, I evaluate how alternative payment schemes and incentive programs would affect the treatment choices and patient outcomes.