Who Benefits from Quality Payments? Evidence from Medicare Advantage

How do insurers respond to quality payments? In the private Medicare market, the Quality Bonus Payment (QBP) demonstration beginning in 2012 increased the benchmark for higher quality contracts; conditional on bid and benchmark, rebate also increases with quality. Previous studies on payment reforms in this setting typically use geographic variation in benchmark on plan generosity, also averaged by regions. Specifically in the QBP case, higher benchmark rate to double bonus county is associated with more entry in these counties. However, since rebate is calculated at the insurer level, whereas contracts serve multiple regions, understanding how rebate is distributed across service areas is important for understanding policy incidence and the effect on consumer surplus.

Although all contracts are potentially affected by the quality payment, we focus on the extreme tails of the quality distribution: contracts with rating 4.5 stars and above in the baseline (2009-2010) are most likely to receive payment increase from the 2012 reform, and form our treatment group. Contracts with baseline rating below 2.5 form the control. We link contracts longitudinally, and use within-contract variation around the reform for identification. In our distributional analysis, we further look at within-contract, cross-region variation over year, or a triple difference design.

We first show that rebate increased mechanically for high quality contracts due to the policy: raw difference between bid and benchmark did not change. Second, we find no reduction in average premium or drug deductible offered by high quality contracts, suggesting generosity may have increased for some enrollees but in fact decreased for others. This prompts us to investigate the within-contract cross-region variation in plan offering, pricing, and enrollment, or the distributional incidence of rebate payments. We find that high quality contracts significantly decreased enrollment in worse risk regions (relative to own market composition, which we find is stable over the study period), and increased market share in better risk regions. Further, they are more likely to offer high-premium-high-deductible plans in worse risk regions. On the margin of entry, high quality contracts are more likely to deploy new plans in better risk regions, and these plans tend to have higher premium than continuing plans. The insurer selection has meaningful impact on enrollee risk pool: high quality contracts have significantly smaller increase in risk score than control contracts.

Finally, we show that the risk selection is plausibly attributable to the design of the quality rating system itself: among the set of high contracts, higher risk score over the quality assessment period (2 years before enrollment) significantly reduces the likelihood of maintaining high status, and predicts lower star rating. Because of this empirical correlation between enrollee risk and contract quality, there is incentive for high contracts to risk-select better enrollees through pricing and entry, raising the question of the distributional incidence of rebate. Furthermore, risk adjustment in its current form cannot countervail the selection incentive working through the quality payment.