Why Are Health Care Report Cards So Bad (Good)?: Theory and Partial Evidence

Issuing quality report cards for health care providers has drawn from profession and academia criticisms largely based on intuitions, conventional wisdom, and stand-alone empirical approaches. I provide a signaling-game theoretical foundation, upon which a new empirical framework is proposed, to study the effects of the report cards. I find that, when patients and health care providers are matched randomly, multidimensional measures in the existing report cards render them the optimal mechanism that reveals the provider types without causing providers to select patients. However, asymmetric distributions of patient types between providers, caused by the referring physician, may force the high-quality provider to shun patients in order to signal himself. Despite this imperfection, the existing report cards cause the minimum selection compared with alternative report mechanisms. The theoretical results imply that treatment effects of the report cards vary from the signaling period to the signaled period, entailing estimating each effect with a difference-in-differences estimate in each period. I show that, in contrast to previous literature, a negative incidence effect is not sufficient to indicate existence of selection behavior. Moreover, the traditional single difference-in-differences estimate cannot capture the report cards' long run welfare effect. More broadly, my study can be extended and applied to other industries with experts, such as automobile service, law, and education.

For more information, contact Philipp Kircher.