Tal Gross and Matthew Notowidigdo have been collaborating since they were in graduate school together at MIT. Today, they are economists specializing in healthcare, Gross at Boston University and Notowidigdo at the University of Chicago. In their new book Better Health Economics, they proclaim, “The world needs health economists because healthcare is different from every other part of the economy.”

They list a bunch of these differences:

  • The most effective way of paying a physician is an open question.
  • An insurance company negotiates and pays on behalf of the patient.
  • Patients lack a good understanding of healthcare.
  • The market for healthcare is a matching market in which physicians, patients, and insurers select one another.
  • In some cases, government mandates the provision of healthcare without compensating the provider.
  • The choices that one individual makes with respect to healthcare may harm or benefit others.

“This book,” the authors tell the reader, “is an answer to all of the people who wonder what it is that health economists do.”

Insurance’s effects / Gross and Notowidigdo begin by introducing health insurance as “a financial product” that is “first and foremost, a tool for handling financial risk.” They quote from a 2018 Annual Review of Economics article by Amy Finkelstein, Neale Mahoney, and Notowidigdo: “Health insurance allows risk-averse individuals to transfer resources across states and thus smooth their consumption in the face of unanticipated, out-of-pocket medical expenses.”

Having health insurance entails “financial consequences.” As grad students, Gross and Notowidigdo measured the effect of rising enrollments in Medicaid programs on bankruptcy rates. They write, “nearly all states saw an increase in bankruptcy rates,” but “states that expanded Medicaid the most saw the smallest increases in bankruptcy rates.” Then they write:

In all cases, we found a robust negative relationship: Medicaid expansion seemed to lower the number of bankruptcies. If a state expanded Medicaid eligibility to cover an additional 10 percent of the population, our results suggested this would reduce the consumer bankruptcy rate by 8 percent.

There is an apparent conflict in these passages. On the one hand they write that Medicaid expansion lowered the bankruptcy rate, but on the other hand they claim that all states experienced an increase in bankruptcies. So, there must be a secular trend of increasing consumer bankruptcy, and Medicaid expansion was not enough to fully offset it. Nevertheless, they summarize the financial consequences of health insurance as follows: “More health insurance means fewer bankruptcies, defaults, and delinquencies.”

Health insurance not only has financial effects but also consumption effects, write Gross and Notowidigdo. Years ago, when Oregon randomly selected residents for the state’s Medicaid program, a group of researchers led by Finkelstein used it to study the effects of health insurance on health-related outcomes. They determined that Oregonians who “won the [coverage] lottery” and obtained Medicaid coverage were “more likely to visit the emergency room, more likely to have an outpatient visit, more likely to be hospitalized, and more likely to pick up prescription medications” than those who were not randomly selected.

Insurance also appears to have health effects, Gross and Notowidigdo write. Another study examined US taxpayers who were penalized for failing to have health insurance in the early years of the 2010 Affordable Care Act (ACA). IRS employees mailed randomly selected taxpayers “letters suggesting that they sign up for health insurance” so they might avoid paying a tax penalty the next time they filed a tax return. A greater share of taxpayers who were reminded to obtain health insurance did so than the share of taxpayers who were not reminded. The result: “Americans who received one of these letters faced a risk of death that was about half-a-percent lower than those who did not.” Individuals who obtain health insurance not only get more healthcare; their health improves.

Adverse selection / Having demonstrated that health insurance is good for people, Gross and Notowidigdo argue that government should require people to buy it. Their argument rests on adverse selection, the concept that people with higher health risk are more likely to buy insurance.

To illustrate, they imagine a market with 26 individuals on the demand side (Aaron to Zack) whose “expected healthcare costs” range from $1,000 per year (Aaron’s) to $26,000 per year (Zack’s). Their insurer aims to break even. If the insurer sets the premium at the average cost of $13,500, it will fail to break even because of non-medical costs. If the insurer raises the premium, healthier individuals with lower expected healthcare costs will decline to buy a policy, while sicker individuals with higher expected costs will remain. The insurer still fails to break even. “This process,” the authors state, “is sometimes called ‘unraveling.’”

“Throughout history and across countries,” they maintain, “there has never been a robust market for individual health insurance.” Competition does not help. Gross and Notowidigdo imagine an incumbent insurer that suddenly is faced with a competitor whose “business strategy is ‘cream skimming.’” This competitor refuses to cover customers with higher costs, and thus can charge a lower premium than the incumbent. The newcomer then attracts healthier, less expensive customers, while the incumbent is stuck with sicker, more expensive customers. The incumbent then raises its premiums further, incentivizing its remaining lower-cost insureds to switch to the newcomer, or joins the newcomer in refusing to cover higher-cost customers, or else goes bankrupt. To correct the problem of adverse selection and prevent the market for health insurance from unraveling, the authors recommend “an individual mandate and subsidies” for high-cost customers.

There are some problems with their story, however. Although there are more healthy and less healthy individuals, one reason the less healthy Zack spends an average of $26,000 per year is that he incurs “regular, predictable costs” such as “test strips, insulin” if he is diabetic. The purpose of insurance, the authors told us, is to cover “unanticipated” expenditures. An insurer cannot break even covering expenditures that are certain to occur. Put differently, if Zack is trying to find a way to pay for medical expenses he knows he’ll incur, he is not in the market for insurance, but for healthcare generally. So, it is not necessarily the market for insurance that is malfunctioning.

This difference explains why American healthcare is regulated the way it is. In addition to recommending “mandates and subsidies” to counter the selection problem, Gross and Notowidigdo reason that “regulators also need to dictate what services insurers cover.” Their view of government regulation is too rosy. Recall the legal controversies following the ACA requirement that employer-provided insurance cover contraceptive services. In this sense, regulation is not a fix; it is politicization that leads to courtroom battles.

Moral hazard / If health insurance providers and consumers overcome the selection problem and get together, another problem will surface: moral hazard. Gross and Notowidigdo explain it this way: “Healthcare becomes so cheap that people consume healthcare that they barely value.” The authors cite a RAND experiment in which some households paid for their healthcare while others did not. Paying households used significantly less care than those that did not pay. Furthermore, paying households used less care “regardless of whether it was effective or ineffective,” but that did not adversely affect a person’s health “on average.” That qualification is necessary because members of households who paid for healthcare that were less healthy to begin with became even less healthy.

The authors describe a new approach to dealing with moral hazard: value-based insurance design (VBID). “The VBID crowd,” they explain, “argues that people should face no deductibles at all but rather a different price depending on their health and the healthcare they are seeking.” Price setting by VBID might be difficult to accomplish. The authors admit that VBID is no panacea, not only because “it’s not so easy to single out which care is worthy of either zero cost or a high price,” but also because consumers might resist.

Supply side / Gross and Notowidigdo focus on the supply side of healthcare markets as well as the demand side. They grapple with the issue of doctors’ pay. The mean of all doctors’ incomes was $340,000 in 2017. Primary care doctors earned below the mean: $243,000. Surgeons earned above: $500,000.

Both the market process and politics explain doctors’ incomes. The authors report research based on “a unique data set with nearly all American physicians’ salaries.” Specialties that require more training fetch higher salaries than specialties that require less training. Also, specialists who work longer hours earn higher salaries than specialists who work fewer hours. The authors attribute those results to the market.

Politics matter because doctors lobby government officials to erect and maintain barriers to entry. For example, the American Medical Association lobbies to prevent nurse practitioners from competing against doctors. Similarly, foreign doctors face legal barriers to practicing in the United States. “The goals of organized physicians,” Gross and Notowidigdo write delicately, “do not perfectly align with what is best for the American public.”

Although the authors appreciate what doctors do, they have an “uncomfortable conversation” about how to compensate doctors. It is uncomfortable for two reasons: doctors’ compensation is about a tenth of all healthcare expenditures, and their desire to find a way to compensate doctors so that healthcare outcomes will improve.

Traditionally, doctors were paid on a “fee-for-service” basis. In that case, doctors might schedule patients for more appointments than necessary or perform more tests or procedures than are necessary. One of several “alternative payment schemes” is to pay bonuses to doctors when their patients achieve favorable health outcomes. For instance, if a doctor’s patients reduce their blood pressure, the doctor would receive a bonus. Aligning a doctor’s pay with the health of his or her patients sounds promising, but there are problems. For one, doctors might decline to assist the least healthy patients.

The authors expect the reader to ask, “What payment method is best?” Their response is, “They’re all terrible.” Each way of paying doctors involves tradeoffs. The authors resolve to continue searching for a better way to compensate doctors.

Care quality / There is much applied microeconomics in the book. Readers may learn, for example, who bears the burden of covering the cost of healthcare for those without insurance, how hospital administrators and doctors respond to incentives, and the effects of horizontal and vertical mergers in the healthcare sector.

Issues surrounding the quality of healthcare were especially interesting. Gross and Notowidigdo warn that “it’s hard to measure quality.” One group of researchers sought to determine whether hospitals with higher patient survival rates, lower readmission rates, and other favorable indicators attracted more patients. In fact, they did. Gross and Notowidigdo insist that anyone measuring the quality of healthcare must randomly assign patients to hospitals or doctors. This is to assure that whoever is studying quality will not be fooled into thinking that hospitals or doctors who accept the least healthy patients are providing lower quality care.

Measuring quality leads to unintended consequences. The authors describe a New York State initiative to measure the performance of cardiologists. The evidence showed the cardiologists referred less healthy patients to the Cleveland Clinic rather than keep them in their practices. In other words, measuring performance led to “cream skimming.”

Conclusion / Gross and Notowidigdo tend to be critical of markets and confident that experts or government officials can improve upon less-than-ideal market outcomes. Take this passage from their chapter on prescription drugs:

The private market, left to its own devices, is not going to work, because the R&D involved is a public good. There is a role for the government in developing new medicines, because free markets can’t sustain public goods. Individuals, on their own, won’t build lighthouses, because they can’t force users to pay for the lighthouse. The same is true of pharmaceutical R&D, and therein lies the problem that only a government can solve.

The authors do not mention Ronald Coase’s insight that private individuals did in fact operate profitable lighthouses in 19th century Britain, suggesting that public goods problems are not always intractable. Perhaps Gross and Notowidigdo should lessen their enthusiasm for “interventions” and “nudges.”

Overall, they accomplish their goal of providing “a tour of health economics,” though I was disappointed they did not include a discussion of preexisting conditions. As tour guides, they are at their best when citing empirical literature, warning of unintended consequences, and identifying tradeoffs. They would be better if they recognized the limits of expertise and government regulation.