Venue: The Fuqua School of Business, Duke University, 1 Towerview Drive, Durham, NC 27708-0120
Presentation
A Comparison of Treatment Effects Estimators Using a Structural Model of AMI treatment choices and Severity of Illness Information from Hospital Charts
A common problem in economics is to estimate the effects of an intervention on individual outcomes in the presence of unobserved heterogeneity. We compare the performance of various estimators that have been proposed to address this problem using a novel approach in the absence of a measure of the true treatment effect. We estimate a structural model of treatment and hospital choices for individuals with acute myocardial infarction (AMI) using a unique data set that combines administrative Medicare claims data with hospital chart data on patient severity of illness. Using the estimated structural model we simulate data for which the treatment effect is known and use this to evaluate the performance of the treatment effects estimators. We find that the estimators do a poor job in recovering the true treatment effect in the presence of unobserved heterogeneity. As measures of heterogeneity are added to the regression specification the results improve in general. Our results also suggest that the linear regression specification typically adopted in using these estimators also contributes to poor estimates. Overall, our findings provide a tale of caution for usersof these estimators when they suspect that there is a high degree of unobserved heterogeneity in the sample or that the data generating process is highly non linear.