Responsible researcher: Eduarda Miller de Figueiredo
Authors: Jason Abaluck, Mauricio Caceres Bravo, Peter Hull, Amanda Starc
Intervention Location: United States
Sample Size: 186,603,694 beneficiary-years
Sector: Healthcare
Variable of Main Interest: Mortality
Type of Intervention: Differentiation between health plans
Methodology: OLS and IV
Summary
The quality of a product is an important point when consumers make their decisions and, knowing this, policy makers are concerned with better informing consumers. The objective of this article was to estimate the effects of different private health insurance plans on mortality, investigating why some plans have better quality. Furthermore, there is an important assessment of consumer demand and responses to mortality effects. Using OLS and IV methodologies, the authors demonstrated that there may be benefits to directing consumers to utilize lower observational mortality plans.
The quality of a product is an important point for consumers to make their decisions and investments. If consumers cannot determine whether certain plans are more likely to improve their health, competition is unlikely to encourage insurers to invest in quality. To better inform consumers, policymakers disseminate supplier and plan quality measures. But there is little evidence of how well existing quality measures predict causal impacts, much less whether consumers are aware of quality differences between plans.
The objective of this article was to estimate the effects of different private health insurance plans on mortality, investigating why some plans have better quality and evaluating whether consumer demand responds to the plans' mortality effects.
The analysis was conducted by looking at the Medicare Advantage (MA) market, in which beneficiaries choose from a wide range of private managed care plans that are subsidized by the government. The MA Program is large and growing, covering more than a third of Medicare (KFF, 2019). Furthermore, it is important to highlight that annual mortality in the elderly population benefiting from MA is high, a rate of 4.7%.
The Medicare was created in 1965 to provide insurance coverage for Americans aged 65 and over. Where Parts A and B of the program are called “ Traditional Medicare ” (TM) and covers hospitalizations and medical services for most beneficiaries. “ Medicare Advantage ” (MA) has a large and growing share of beneficiaries who have chosen to have coverage through a range of different private health care plans. Medicare beneficiaries must choose between TM and MA at their workplaces, where MA plans must provide all required TM insurance benefits in exchange for a monthly payment.
Competitive plans may charge lower premiums or offer supplemental benefits to attract certain consumers. Where MA plans tend to vary significantly in their insurance networks, offering more generous financial coverage or better cost-sharing.
The MA program has historically had two conflicting goals: expanding consumer choice and reducing Medicare (Commission, 2001, 1998). However, even as policymakers recognize the need for health insurance beneficiaries to make informed decisions in the marketplace, what is less discussed is the role of competition among MA plans in increasing product quality.
Public plan quality ratings have existed since 1999, with current quality rankings, the famous rating stars. These stars began playing a major role in policymaking with the Affordable Care Act of 2009 [1] , which provided bonus payments for top-tier MA plans. Unlike other programs, MA plans are not currently rated or rewarded for achieving low mortality rates per enrollment.
Medicare beneficiaries aged 65 and older in one of the 50 US states or the District of Columbia from 2006 to 2011. The sample consists of 186,603,694 beneficiary-years with enrollment, demographic, and mortality information. For the analysis using the instrumental variable method, the authors restrict the sample to the period 2008-2011, with beneficiaries who ended up in an MA plan the previous year.
The initial analysis is computing observational differences in mortality rates between Medicare operating in the same county using ordinary least squares (OLS) regressions. Afterwards, the authors use the instrumental variables approach, using plan terminations to measure the validity of observational mortality differences in predicting differences in causal plan mortality effects. With this approach, it is possible to estimate the expected impact on mortality of relocating beneficiaries to different plans.
The authors also estimated the extent to which higher quality plans tend to attract greater market share. Through an additional extension of structure IV, we estimated the implicit weight that consumers place on the mortality effects of the plan and estimated the implicit willingness to pay for the quality of the plan. Thus, it was possible to estimate latent demand based on the market share of a plan after accounting for price differences and, through the IV structure used in this article, it was possible to relate demand to the unobserved quality of the plan and recover the implicit willingness to pay of this relationship.
The results demonstrated by the OLS estimations show that beneficiaries enrolled in high or low mortality plans that are terminated in year t-1 tend to choose plans in year t that are more typical in terms of observational mortality, in relation to beneficiaries more inertial in non-terminated planes.
When analyzing the average predicted mortality between terminated and non-terminated plans in different mortality deciles, the 1st stage results demonstrated that there is no differential trend in predicted mortality for terminated compared to non-terminated plans. Therefore, according to the authors, any differential trend in the actual mortality of beneficiaries in terminated versus non-terminated plans would be unlikely to be due to preexisting differences in their health.
Lagged observational mortality strongly predicts subsequent mortality for beneficiaries previously enrolled in non-terminated plans, but this relationship is effectively flat for beneficiaries previously enrolled in terminated plans. That is, beneficiaries in terminated high-mortality and low-mortality plans appear similar to those in corresponding nonterminated plans until they are induced by terminations to choose more medium plans.
Examined together, the results suggest that a large proportion of the considerable variation in observational mortality among MA plans reflects the causal impact of plan enrollment. In which the authors emphasize that the findings do not rule out selection bias in observational mortality, for example, it can be expected that sicker beneficiaries (not observed) would systematically prefer certain plans with greater coverage.
When analyzing the demand for higher quality plans, the authors estimated an upper limit on the implicit willingness to pay that varied between 275 and 476 dollars, that is, consumers are willing to pay no more than this amount to compensate for an increase of 1 percentage point in 1-year mortality.
In this article, the authors demonstrated, through a robust methodology, that the effects of mortality are critical for evaluating consumer choices, not only observing the financial consequences, but the quality of the plan. Furthermore, the findings suggest, according to the authors, that plans with higher premiums, more generous drug coverage and higher expenses tend to reduce consumer mortality.
Therefore, the results demonstrate that there may be major benefits to directing consumers to utilize lower observational mortality plans. Furthermore, they suggest that insurers face weak incentives to invest in improving consumer health, which could be strengthened by new contractual or organizational forms.
References
KFF (2019). A dozen facts about Medicare Advantage in 2019. https://www.kff.org/medicare/
issue-brief/a-dozen-facts-about-medicare-advantage-in-2019/. Accessed: 2021-03-05.
[1] 2009 Affordable Care Act
Check the Institution's registration in the e-MEC System here
© 2025 All Rights Reserved
