Estimating Self-Selection in Medicare Advantage

Abstract

We explore the fundamental question of selection in Medicare Advantage by exploiting quasi-experimental variation from the Initial Enrollment Period for Medicare eligibility to evaluate and describe participation in Medicare Advantage. Using administrative claims data between 2007 and 2017, we investigate the transition from commercial insurance to Medicare Advantage for a comprehensive subset of commercially insured enrollees. We use the sharp cutoff at age 65 in one of the largest commercial and Medicare Advantage databases in the United States to implement a “positive correlation” test. Our findings using baseline characteristics at age 64 reveal that enrollees in Medicare Advantage are advantageously selected on multiple measures of health status such as Charlson Comorbidity Index (CCI) scores, out-of-pocket costs, utilization, while differentially selected on demographic characteristics. Furthermore, using predicted Medicare Advantage utilization based on observable characteristics, we find even stronger evidence of advantageous selection, suggesting that observable characteristics at age 64 systematically predict which enrollees would benefit most from Medicare Advantage's features, indicating forward-looking selection behavior.