We exploit a brief period of asymmetric information during the implementation of Pennsylvania's ‘report card’ scheme for coronary artery bypass graft surgery to test for improvements in quality of care and selection of patients by healthcare providers. During the first 3 years of the 1990s, providers in Pennsylvania had an incentive to bias report cards by selecting patients strategically, with patients having no access to the report cards. This dichotomy enables us to separate providers' selection of patients from patients' selection of providers.
Using data from the Nationwide Inpatient Sample, we estimate a nonlinear difference-in-differences model and derive asymptotic standard errors. The mortality rate for bypass patient decreases by only 0.05 percentage points because of the report cards, which we interpret as evidence that quality of bypass surgery did not improve (at least in the short-term) nor did patient selection by providers occur. Our timing, estimation, and asymptotics are readily applicable to many other report card schemes. Copyright © 2012 John Wiley & Sons, Ltd.