On Monday, The Washington Post published an article by Casey Mulligan and Tomas Philipson attacking Lawrence Summers’s statement that “thousands” of individuals would die if the Republican tax bill became law. Summers reached his estimate after carefully reviewing the literature and consulting with health economists Jonathan Gruber and Mulligan and Philipson’s University of Chicago colleague Dean Kate Baicker, who has published a number of influential studies on the effect of health insurance on health.
While Summers claimed only that thousands would die, the Baicker et al. studies suggest that at least one person dies for every 1,000 people who lose health insurance. Even with very conservative assumptions about how many people will lose health insurance as a result of the repeal of the Affordable Care Act's individual mandate, this implies tens of thousands of premature deaths.
Mulligan and Philipson make errors in logic and engage in a highly selective use of evidence. After seeing the best challenge that the Trump Council of Economic Advisers can come up, we are more convinced of our initial conclusions. Five observations seem important.
The argument against the individual mandate is uncompelling.
Standard economics predicts that people are better off making choices voluntarily only when their choices have no external effects on others. This is patently not true in health-insurance markets once we have nondiscriminatory insurance pricing. Such “community rating” is supported by the vast majority of Americans and is the most popular part of the ACA. If insurers are required to charge the healthy the same as the sick, then when I, as a healthy person, decide not to buy insurance, it raises your rates as a sicker person. Such a case is the classic market-based scenario where a mandate can improve welfare, as Summers pointed out more than 25 years ago in the American Economic Review.
Moreover, a focus of recent research has been on inconsistencies in individual decision-making in health-insurance markets, which further strengthen his original argument. Many studies document the problems that individuals have in making appropriate decisions about their health-insurance plans. Perhaps most compelling is a recent study that showed that the majority of employees at a company chose “dominated” health-insurance plans, where their total costs would certainly have been lower if they had chosen other options offered.
Changes to the rules of the game, therefore, can have significant negative effects on consumers. A particularly telling case is what happened during welfare reform in 1996. It changed the rules so that legal immigrants who had been in the United States less than five years could no longer qualify for Medicaid. This rule caused a large decline in insurance coverage among immigrants who had been here for more than five years. That is, even though the coverage was valuable and free, individuals who were entitled to keep the coverage dropped it because they didn’t understand the rules.
The authors turn to what they regard as empirical evidence and are simply wrong.
They completely misinterpret the evidence from the RAND Health Insurance experiment of the 1970s. This experiment did not test whether insurance mattered (everyone in the experiment was insured), but rather whether the generosity of insurance coverage mattered (it didn’t). They wrongly cite a review paper by David Meltzer and Helen Levy that concluded that the evidence on health insurance and health outcomes is uncertain. And they ignore the bulk of the evidence in the 15 years since the Meltzer and Levy review that shows that health-insurance coverage does matter for health.
In particular, numerous studies over the past 15 years using strong research designs have found that expanding health-insurance coverage improves health care. Mulligan and Philipson instead choose to focus on the one article that does not show significant effects on physical health, without highlighting that (a) that same study found massive improvements in mental health, (b) that study had a short follow-up period and (c) that study’s results are sufficiently imprecise that they can’t rule out the mortality effects cited in Summers’s earlier article. Indeed, Summers's article used a highly conservative interpretation of some of the best work of the past 15 years. And Mulligan and Philipson accuse Summers of cherry-picking!
They attack the Congressional Budget Office's predicted premium hike as “speculative.”
In doing so, Mulligan and Philipson ignore that CBO's prediction of premiums in the exchanges was amazingly accurate — as of the 2017 rate increases, premiums were almost exactly where CBO predicted they would be. It is true that predicting effects on market premiums is hard — but the 10 percent estimate by CBO is as likely to be low as it is to be high.
They claim that expanding insurance coverage will worsen the opioid crisis.
Exactly the opposite is true. As Gruber and Angela Kilby recently showed, the expansion of Medicaid has led to much more intensive use of cost-effective treatment for opioid addiction. By cutting back on insurance coverage, we deny addicts treatments that can save their lives.
They argue that the tax bill will save lives by promoting economic growth.
This is perhaps the most disingenuous part of their argument. In a recent survey of renowned economists from across the political spectrum sponsored by the University of Chicago, only one said the bill would substantially increase economic growth, and he later recognized that he had misread the question. Moreover, there is no evidence in the modern U.S. economy that faster growth reduces mortality. Indeed, the best research in this area shows that downturns in the economy lower mortality, although that refers only to cyclical fluctuations and not long-term growth.
But there is universal consensus that these tax cuts will cause a massive increase in the nation's deficit — which ultimately will lead to offsetting tax increases, or, more likely, spending cuts. These cuts in spending will probably be focused on the very programs that save lives and provide valuable financial protection to our nation’s poorest and oldest. It seems implausible to argue that a tax break where 60 percent of the benefits go to the 1 percent of Americans, paid for by broad cuts to public spending focused on the needy, will improve public health.
The United States is facing a crisis of confidence in our academic institutions, and in the role of facts in general in important public policy debates. Ad hominem and poorly researched attacks on credible and thoughtful policy analysis only feed that crisis. We urge our colleagues to draw conclusions from a broad reading of the best available evidence. Doing so makes it obvious that this is a piece of legislation that will be bad for public health.