Healthcare and the Moral Hazard Problem

There are plenty of contexts in medicine in which there’s a cheap option and an expensive option. Sometimes, the cheap option is just as good as the expensive one. For instance, an upper respiratory infection can be treated in an emergency room (expensive) or in an ordinary doctor’s office (cheap). Some conditions can be treated with generic drugs (cheap) or branded drugs (expensive). If the consumer pays the same price for either option, why not choose the expensive option? And if all consumers face the same incentives and behave in the same way, healthcare spending, overall, might rise in ways that don’t actually improve health.

There’s a technical term for this issue: it’s called a “moral hazard problem.” In general, moral hazard problems are situations in which there are two parties in a transaction and one party cannot control the actions of another. In the case of health insurance, there’s the insurer and the consumer, and the insurer bears the costs of the consumer’s healthcare decisions. The consumer can choose the cheap option or the expensive option, and the insurer pays either way.

Now, to be clear, that’s not to say that generous health insurance is a bad thing. Generous health insurance also protects consumers from risk. The issue here is that there’s a trade-off. On the one hand, we want health insurance to be generous so that people are protected from risk. On the other hand, we don’t want health insurance to be generous, because of moral hazard. The generosity of health insurance—how much the consumer has to pay for healthcare—has to balance those two forces.

But all of that is theory, words on a page that describe how some people think the world works. Theory needs to be tested. Next, let’s turn to real-world evidence on moral hazard in health insurance.

What do copayments do?

Health economists first examined moral hazard in health insurance with a field experiment. In the late 1970s, a team of health economists sat down at the RAND Corporation, a think tank in Santa Monica, California. Joseph P. Newhouse, now at Harvard, was a young economist just starting out at RAND. He asked his colleagues a simple question: How do people respond to the price of healthcare? If ordinary people have to pay more for healthcare, do they consume less of it? Newhouse found that lots of economists had opinions on that question, but no one had any good evidence.

Some argued that healthcare was different than other goods, that healthcare is always a matter of “your money or your life.” Therefore, they argued, people would pay whatever price they faced for healthcare—the price didn’t matter, because healthcare was so important. And, by extension, those people were not concerned with moral hazard: since healthcare is different from other goods, it doesn’t matter, they argued, that health insurance makes healthcare cheap.

Economists knew that prices matter for ordinary goods: coffee, wheat, motorcycles. If the price of coffee goes up, people buy less coffee. That is, as we say in Econ 101, a matter of the demand curve. Is the same true for healthcare? No one knew.

The RAND researchers, led by Newhouse, decided to run an experiment. They took 2,750 American families and randomized them into two groups. The study included both urban and rural households and spanned a broad range of income levels. One group of families was put on a free-care plan: for several years, all of the healthcare they needed would be free. Every doctor visit, every dentist visit, every medication: they would pay nothing.

Other families were put on a high-deductible healthcare plan. They would be responsible for all of their healthcare costs up to a thousand dollars. After $1,000, their health-insurance plan would kick in and cover everything. But until then, they had to foot the bill on their own.

Remember: randomization makes experiments valid—it means that both groups began the experiment with the same health, on average. They had the same average income, level of education, number of children, number of televisions, and, most importantly, the same average health. As a result, any differences in outcomes in the years following the RAND experiment can be interpreted as the impact of the health-insurance plans themselves.

The study became known as the RAND Health Insurance Experiment and it lasted from 1976 until 1982. For health economists, the experiment amounts to a combination of NASA launching a space shuttle, Bill Gates starting Microsoft, and Ayatollah Khomeini returning to Iran. It happened in the late 1970s, and it’s a big deal to us. The RAND experiment is one of the most expensive experiments ever performed by social scientists.

For years, the families participating in the experiment led their ordinary lives; the only thing out of the ordinary was that a team of researchers at RAND was handling their health insurance. Then, after years of being on either the free-care plan or the high-deductible plan, the participants were given a final physical exam, and the experiment was over.

Newhouse and his colleagues spent years poring over the data, trying to understand how having to pay for healthcare affected families. The researchers studied the results from every possible angle, slicing and dicing the data every which way. The results of the experiment filled hundreds of academic papers and also a 516-page book. But, decades later, the most relevant discoveries from the experiment boil down to three main conclusions:

1. High prices really do cut how much care people choose to consume. First, the researchers compared the amount of healthcare consumed by families that were put on the free-care plan with that of families randomly assigned to the high-deductible plan. Those put on the free-care plan consumed an average of almost $2,000 (in 2021 dollars) as compared with about $1,600 in healthcare for those on the high-deductible plan. That’s a roughly 20 percent difference—a large difference.

In other words, incentives matter, even for healthcare. People who face a higher price for healthcare consume less healthcare. In the language of Econ 101, demand curves slope down, even for healthcare. Yes, healthcare is important, and people treat it as important, but, at the end of the day, the price still matters.

2. Deductibles lead families to cut back on all healthcare, regardless of whether it’s effective or ineffective. The second conclusion of the experiment arose as the researchers tried to figure out which healthcare the families on the deductible cut out. The researchers assembled a panel of physicians and gave them all of the medical charts associated with the experiment. They asked the physicians to categorize all of the healthcare as “highly effective” care or “rarely effective” care. Going to the ER for a runny nose: that’s rarely effective care. Going to the ER for a heart attack: that’s highly effective care.

The panel of physicians worked through the stack of charts, methodically categorizing all visits as highly effective or rarely effective. Then they studied how the high-deductible plan affected those two categories.

Families facing a deductible cut back on highly effective care by about 30 percent relative to those on the free-care plan. And then the researchers found roughly the same effect for rarely effective care: a roughly 30 percent drop in utilization. The lack of contrast between those two findings is the second conclusion of the experiment.

And that finding alone is kind of disappointing. Health policy would be much simpler if people behaved like shrewd medical experts whenever they faced a deductible. Unfortunately, that’s not how it works. Patients are not physicians themselves—they don’t know what is effective and what is ineffective. When faced with a high price, they just cut back on all of it, both care that really matters and also care that is probably wasteful.

3. A high-deductible plan takes people from bad shape to worse shape. Lastly, the researchers studied what deductibles did to people’s health. Remember that families were randomized to the two health-insurance plans, so any differences in health outcomes during the RAND experiment years were probably a result of the impact of those plans.

After several years on the insurance plan that they were assigned to, everyone’s health was evaluated. Overall, there was no difference: people who spent 3–5 years on the high-deductible plan finished the experiment in roughly the same health as people who spent that time on the free-care plan, according to a paper by Newhouse and a team of researchers.

Things were different, however, for one group of participants. The researchers focused on what they called “elevated-risk participants.” That group consisted of people who were in poor health at the start of the experiment. Maybe they already had a chronic condition or maybe they were obese. For that group, the researchers found that the deductible plan harmed their health. A few years on a high-deductible plan took people from bad shape to worse shape.

That finding is, perhaps, intuitive. If you’re in good health, a deductible will induce you to consume less healthcare, and that’s going to have a very small impact on your health. After all, you’re in good health, so a bit more or a bit less healthcare is not going to have a big effect, at least on average. But if you’re already at elevated risk, a deductible leads you to consume less healthcare, and for you, that really matters.

The three conclusions of the RAND experiment paint a confusing picture of what deductibles do to people. On the one hand, deductibles lead people to cut back on healthcare a lot but they do not hurt people, on average. On the other hand, the participants who were assigned a high-deductible plan cut back on all healthcare, not just ineffective care. And the most vulnerable among them ended up worse off.

Matthew J. Notowidigdo is the David McDaniel Keller Professor of Economics and Business and Public Policy Fellow at Chicago Booth. Tal Gross is a professor in the Department of Markets, Public Policy & Law at Boston University. This is an edited excerpt from their book, Better Health Economics. Reprinted with permission from the University of Chicago Press. © 2024 by the University of Chicago. All rights reserved.