Paying Beneficiaries, Not Providers: Transforming Medicaid and Medicare into poverty-fighting programs would greatly improve Americans' health.

• Author: Hyman, David A.
• Position: HEALTH & MEDICINE

Over the past few decades, retirement savings has shifted from a defined-benefit model to a defined-contribution model. If we look at the major social welfare programs operated by the federal government, we see a very different pattern. Two of those programs, Medicaid and Medicare (both enacted in the mid-1960s), are defined-benefit programs, which pay health care providers for specific goods and services. A third, substantially older, major federal social welfare program, Social Security, is a defined-contribution/cash-transfer program, which gives beneficiaries a fixed amount of money and lets them decide how to spend it.

Our continued reliance on a defined-benefit approach for Medicaid and Medicare when we simultaneously use a defined-contribution approach for Social Security raises obvious questions. What do we know about the effects of Medicaid and Medicare? Is there a case for transforming them into programs more like Social Security? Which approach better ensures good health?

MEDICAID AND MEDICARE'S HEALTH EFFECTS

We begin by focusing on the effects of Medicaid and Medicare on health. Medicaid financially assists impoverished Americans by paying for their medical treatments. Medicare does the same for senior citizens. Not surprisingly, health care providers are enthusiastic supporters of both programs because they relieve them of much of the burden of providing charity care. Provider preferences aside, popular support for both programs has more to do with the collective belief that poor people and the elderly suffer from unmet medical needs and will enjoy better health when their access to providers is improved.

Unfortunately, it has proven to be quite difficult to document health-improving effects from Medicaid. In 2017, the Kaiser Family Foundation--a staunch supporter of Medicaid--summarized the findings of the Oregon Health Insurance Experiment, which randomly sorted applicants into a group that received Medicaid and a group that did not. Evidence of an effect on physical health was mixed at best. The main positive effects were increased detection of diabetes and use of diabetes medication. Although members of the recipient group reported improved mental health, that effect was likely due (at least in significant part) to a reduction in beneficiary concern with financial insecurity related to the risk of health care-related expenditures.

Studies of the Medicaid expansion under the 2010 Patient Protection and Affordable Care Act (ACA) have also had difficulty identifying material improvements in health. A 2018 systematic review/ meta-analysis found that the main effects of the ACA's Medicaid expansion on health outcomes were improvements in self-reported mental health and general health. As in Oregon, documented improvements in beneficiaries' physical health were few and far between. A 2019 Kaiser Family Foundation literature review reached similar conclusions, although another 2019 study found evidence of a reduction in mortality in states that expanded Medicaid.

One important reason for the paucity of positive findings is that health status is a function of a wide array of nonmedical factors including social, behavioral, environmental, and genetic components. The standard estimate is that nonmedical factors account for at least 80% of overall health outcomes, meaning that even substantial investments in health care are unlikely to result in dramatic improvements.

Medicaid is also unlikely to improve population health because a lot of Medicaid's budget is spent on things that are unlikely to generate quantifiable health improvements. For example, a substantial fraction of Medicaid's budget is used to house, feed, and care for poor, elderly beneficiaries. The services that nursing homes and other businesses provide for people who are elderly or disabled are certainly valuable, but their effects on health, longevity, and mortality are likely small and hard to assess.

Similar difficulties have beset efforts to measure the effects of Medicare on health. Because all Americans over age 65 are eligible for the program, there is no control group. Researchers Amy Finkelstein and Robin McKnight studied the health effects of Medicare and were "unable to reject the null hypothesis that, in its first 10 years, Medicare had no effect on elderly mortality." A more recent study by David Card, Carlos Dobkin, and Nicole Maestas found that the all-cause mortality rates and self-reported health status for people immediately before and after the Medicare eligibility threshold (age 65) are quite similar, but there was evidence of a Medicare-associated reduction in mortality for a subset of hospital admissions.

What about the effects of health insurance on populations other than Medicaid and Medicare beneficiaries? Robert Weathers and Michelle Stegman studied previously uninsured applicants for Social Security Disability Insurance who were randomly assigned to three groups. One group was excluded from health insurance coverage, while the other two received health insurance policies with varying benefits. They found evidence of improvements in mental health for the treatment groups, but no statistically significant reduction in mortality. A study of the near-elderly by Bernard Black, Jose-Antonio Espin-Sanchez, Eric French, and Kate Litvak similarly found little evidence that the insured have lower mortality or improved health in either the short or long run.

In interpreting these findings, it is important to note that many uninsured individuals do receive some health care. So, these studies are measuring the incremental (marginal) effect of having health insurance compared to a population that lacks insurance but has some access to health care. Consistent with this observation, Finkelstein and McKnight note that part of the explanation for their (null) findings is that "prior to Medicare, elderly individuals with life-threatening, treatable health conditions sought care even if they lacked insurance, as long as they had legal access to hospitals." Similarly, Black et al. note that given the same dynamic, "much of the additional care the [uninsured] would receive if insured could provide limited marginal benefit."

Where does that leave us? Most of the coverage...