Working Paper: NBER ID: w11514
Authors: Ginger Zhe Jin; Alan T. Sorensen
Abstract: We use data on the enrollment decisions of federal annuitants to estimate the influence of publicized\nratings on health plan choice. We focus on the impact of ratings disseminated by the National\nCommittee for Quality Assurance (NCQA), and use our estimates to calculate the value of the\ninformation. Our approach exploits a novel feature of the data—the availability of nonpublic plan\nratings—to correct for a source of bias that is inherent in studies of consumer responsiveness to\ninformation on product quality: since publicized ratings are correlated with other quality signals\nknown to consumers (but unobserved by researchers), the estimated influence of ratings is likely to\nbe overstated. We control for this bias by comparing the estimated impact of publicized ratings to\nthe estimated impact of ratings that were never disclosed. The results indicate that NCQA's plan\nratings had a meaningful influence on individuals' choices, particularly for individuals choosing a\nplan for the first time. Although we estimate that a very small fraction of individual decisions were\nmaterially affected by the information, for those that were affected the implied utility gains are\nsubstantial.
Keywords: health plan ratings; consumer choice; publicized information; NCQA
JEL Codes: I11; L15
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| publicized health plan ratings (I13) | consumer choice (D10) |
| nonpublic ratings (G24) | consumer choice (D10) |
| publicized ratings (Y30) | utility gains for influenced consumers (D11) |
| consumer choice (D10) | health plan selection (I13) |
| familiarity with plan options (G52) | consumer choice (D10) |