Skip to main content Skip to secondary navigation
Main content start

Unraveling the key drivers behind health care costs

SIEPR’s Matthew Gentzkow talks about his new study showing why health care costs vary so widely across the country.

Economists and policymakers have long examined why health care costs dramatically vary across regions. The average amount that the government spends per Medicare beneficiary can be twice as high from one state to another.

The reason is usually explained in terms of supply and demand. Many studies have found that the vast geographic differences in costs are largely determined by supply-side factors related to providers — where practitioner styles, quality of care, hospital market structures and financial incentives come into play.

But now, there’s evidence that patient preferences and patient health – the demand side of the equation – may play a larger role, according to a study co-authored by SIEPR Senior Fellow Matthew Gentzkow.

In employing a unique empirical strategy – using Medicare patients as the centerpiece and examining their health care usage as they lived in different places – Gentzkow and MIT economists Amy Finkelstein and Heidi Williams have shed new light on why health care costs vary so widely from place to place. Their analysis of millions of Medicare patients boils down health care spending differences to roughly half-and-half between patient- and supply-side factors.

A paper detailing the study, “Sources of Geographic Variation in Health Care: Evidence from Patient Migration,” appears in the November issue of Quarterly Journal of Economics.

Gentzkow discussed the study in an interview.

Medicare accounts for about 20 percent of the nation’s total spending on health care. How is that distributed across the country?

America’s health care costs are higher in the U.S. than in other countries, and we don't seem to get much better outcomes as a result. And what people have observed is that it’s not only true that the U.S. spends much more, but how much we spend varies a lot across different places in the U.S. In some places, we spend way more and in some places, we spend much less. So if you want to understand why costs are so high, understanding why they vary within the U.S. would be very helpful.

There's been a lot of literature in the past trying to do that, and I think a key question that you would ask is, why might some places be spending a lot per person on health care. One obvious category of reasons is because people in that place are sicker. So they just need more health care.

So you think of Florida.

Yes, you think of Florida — people are older, people are sicker, and so we are going to spend more per person there. Or, some place where someone's behavior and lifestyle is less healthy, and they're going to have more complications, more diabetes.

There are also some places where we spend more because the doctors, the providers or the hospitals either have different incentives or different beliefs about how patients should be treated.

So we have stories that are really about demand and patients, and we have stories that are about supply — the doctors.

The idea with our paper is to say: if you want to understand what’s behind the difference in costs between two places, let’s see what happens when a patient moves from one place to another.

What’s new and different about your technique?

The approach in prior literature has been to observe whether any of the overall characteristics of patients in different places explain the cost differences — Are the patients older? Are there places where people smoke more? But when you do that, you don’t really explain much of the differences, and the conclusion has been that most of the geographic (cost) differences seem to be about something on the supply side.

We take a new approach: we focus on people who migrate from one place to another and see what their spending is like in different places.

So think of a high-spending place like Miami, and a low-spending place like Minneapolis, and imagine that we're going to see people who move from Miami to Minneapolis. If the difference between the two places is all about the doctors — the doctors in Miami just like to prescribe lots of expensive tests, whereas the doctors in Minneapolis are kind of frugal and don’t do that — then the day that somebody shows up in Minneapolis, their spending should drop because they're not getting all those tests.

If, on the other hand, the difference is that the people in Miami are sicker, have higher rates of complications and need more care, then they're still going to be sicker after we move them to Minneapolis and they're going to continue to spend a lot.

What were the key findings weighing patient health versus the supply-side factors?

What we find at the end of the day is that it's about half-and-half. It turns out about half of the differences between high- and low-spending places are due to patients, and what that means in our data is that (those differences) remain regardless of where people move.

Does the research point to anything glaring that the government should do or change?

There's at least some chunk of these differences that we are not going to be able to affect with policy, at least in the short run, because it's about the health status of patients, which is not going to change quickly.

On the other hand, we're also saying there's a big chunk of it which is about the doctors and the hospitals, so policies which were motivated by trying to affect that part — of trying to give doctors incentives to spend less or increase quality in hospitals and those sorts of things — there's still plenty of room for those to make a difference.

And what we find is that the effects of the supply side happen very quickly.  So if we can identify what those supply-side differences are and change them, we could have very immediate impacts.

More News Topics