Jason Abaluck, Mauricio Caceres Bravo, Peter Hull, and Amanda Starc* – July 2020

Abstract

Competition in health insurance markets may fail to improve health outcomes if consumers are not willing to pay for high quality plans. We document large differences in the mortality rates of Medicare Advantage (MA) plans within local markets. We then show that when high (low) mortality plans exit these markets, enrollees tend to switch to more typical plans and subsequently experience lower (higher) mortality. We develop a framework that uses this variation to estimate the relationship between observed mortality rates and causal mortality effects; we find a tight link. We then extend the framework to study other predictors of mortality effects and estimate consumer willingness to pay. Higher spending plans tend to reduce enrollee mortality, but existing quality ratings are uncorrelated with plan mortality effects. Consumers place little weight on mortality effects when choosing plans. Moving beneficiaries out of
the bottom 5% of plans could save tens of thousands of elderly lives each year.

7 Conclusions

We find large within-market differences in mortality rates across MA plans after adjusting for observable differences in enrollee characteristics and statistical noise. We then show that this variation is a highly reliable predictor of true plan mortality effects with a novel quasi-experimental design. Publicly available quality measures are uncorrelated with true mortality effects. Perhaps as a result, consumer demand is under-responsive to this dimension of plan quality. In partial equilibrium simulations we show that one-year mortality would fall significantly if beneficiaries were reassigned to lower observational mortality plans, suggesting broad scope for policy interventions based on these measures.

5.1 Proxies for Plan Quality

We start by considering whether existing plan quality measures (star ratings) or prices (premiums)
proxy for observational mortality and true plan effects. To help beneficiaries select plans, CMS
produces star ratings on a 1-5 scale, with 5-stars indicating the highest quality. Star ratings depend on consumer satisfaction surveys and measures of clinical quality, but they explicitly do not condition on outcome data like mortality. In addition to making these ratings available to consumers, the government now pays “bonuses” to highly rated (4- and 5-star) plans.

6 Plan Choice and Mortality

Our forecast coefficient estimates in Section 4.2 suggest that MA plan mortality effects are enormously variable within a market and can be reliably predicted by observational mortality differences. At the same time, our WTP estimates in Section 5.3 suggest that consumers place little weight on this dimension of plan quality when making enrollment decisions. Together, these findings imply that redirecting consumers from observably low-quality plans to plans with better observational mortality could substantially improve beneficiary health.

READ MORE