Better Health Economics: An Introduction for Everyone

Better Health Economics

Better Health Economics

An Introduction for Everyone

Tal Gross and Matthew J. Notowidigdo

The University of Chicago Press

Chicago and London

The University of Chicago Press, Chicago 60637

The University of Chicago Press, Ltd., London

© 2024 by The University of Chicago

All rights reserved. No part of this book may be used or reproduced in any manner whatsoever without written permission, except in the case of brief quotations in critical articles and reviews. For more information, contact the University of Chicago Press, 1427 E. 60th St., Chicago, IL 60637.

Published 2024

Printed in the United States of America

33 32 31 30 29 28 27 26 25 24     1 2 3 4 5

ISBN-13: 978-0-226-82029-3 (cloth)

ISBN-13: 978-0-226-82033-0 (paper)

ISBN-13: 978-0-226-82030-9 (e-book)

DOI: https://doi.org/10.7208/chicago/9780226820309.001.0001

Library of Congress Cataloging-in-Publication Data

Names: Gross, Tal, author. | Notowidigdo, Matt, 1981– author.

Title: Better health economics : an introduction for everyone / Tal Gross and Matthew J. Notowidigdo.

Description: Chicago : The University of Chicago Press, 2024. | Includes bibliographical references and index.

Identifiers: LCCN 2023017134 | ISBN 9780226820293 (cloth) | ISBN 9780226820330 (paperback) | ISBN 9780226820309 (ebook)

Subjects: LCSH: Medical economics—United States. | Medical care, Cost of—United States.

Classification: LCC RA410.53 .G748 2024 | DDC 338.4/73621—dc23/eng/20230527

LC record available at https://lccn.loc.gov/2023017134

This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).

Contents

Introduction

PART I  Demand

1  What Does Health Insurance Do?

2  Health Insurance versus Broccoli

3  Free Care Is Not Free: Who Pays for the Uninsured?

4  Moral Hazard

5  Behavioral Economics

PART II  Supply

6  How Much Should Physicians Be Paid?

7  Doctors and Hospitals Respond to Financial Incentives (Just Like Everybody Else)

8  Payment Reform

9  Horizontal Mergers

10  Vertical Integration

11  Quality

12  Drugs

PART III  Other Determinants of Health

13  Contagion

14  Health Gradients

15  Social Determinants of Health

Conclusion

Acknowledgments

Notes

Index

Introduction

What Do Health Economists Do?

At cocktail parties, we introduce ourselves as health economists. Most people we meet smile and then find someone else to talk to.

After all, what even is a health economist? And why does the world need an entire species of economists dedicated to health? What’s next—shoelace economists? Pavement economists?

The world needs health economists because healthcare is different from every other part of the economy. Most transactions outside of healthcare are relatively straightforward. For example, a farmer sells heirloom tomatoes to a customer at a green market. Everyone knows what they’re getting—there are no surprises, no complications. The customer can reach out and feel the tomatoes—she might even ask for a sample.

In contrast, imagine a doctor selling annual physicals. Suddenly, things are much more complicated.

•  The doctor can be paid a fee for each annual physical she provides, or a lump-sum fee for each patient she covers each year, or the doctor might be salaried. It’s not obvious which option is best. (Chapter 8)

•  The customer is no longer the one paying—typically a health insurance company is footing most of the bill. (Chapter 4)

•  Customers do not even know how much healthcare they need. Everyone knows how many tomatoes they want to buy, but if a doctor recommends an extra blood test or two, most patients just have to trust them. (Chapter 7)

•  A farmer doesn’t care who buys his tomatoes. By contrast, doctors (and insurers) care deeply about which people they treat. Some people will always be unprofitable for insurers and sometimes doctors face incentives to avoid certain types of patients. (Chapter 2)

•  No farmer is required to give away his tomatoes, but healthcare organizations are often required by law to provide free care. (Chapter 3)

•  Whether my neighbor buys tomatoes or not is his business—it doesn’t affect me. But in healthcare, my neighbor’s decisions can sometimes affect my health through contagion. (Chapter 13)

All of which is to say that healthcare really is different. And that is why health economists exist. The tools that economists have developed to understand agriculture—or telecommunications, or real estate—are different from the tools necessary to understand healthcare.

Dolphins and Tuna

One of our thesis advisers in graduate school, Jon Gruber, used a dolphins-and-tuna metaphor as a way to talk about the broader issues in health policy.

Dolphins and tuna often swim together in the ocean. The two are very different: dolphins are mammals that hold their breath to swim underwater, and tuna are fish. For reasons that are not completely clear, eating dolphin is seen as deeply offensive, while tuna is one of the world’s most commonly eaten fish.¹

In the 1980s, people became concerned that tuna fishermen were accidentally killing dolphins. The dolphins would get caught in the nets meant for tuna and then drown. The technical term is that the dolphin had become a bycatch. Through the 1990s, regulators developed standards that were meant to prevent dolphins from dying this way. Americans adopted the Dolphin Safe label to reassure concerned consumers that no dolphins were harmed in the production of their tuna.

So: we want to catch the tuna but leave the dolphins in the ocean. We often face the same problems in healthcare. We want to prevent wasteful healthcare spending but preserve important healthcare. When it comes to the pharmaceutical industry, for instance, we want to pay much less for the drugs that exist, but we also want the drug industry to continue to develop new drugs. Or, in another context, we want to reward high-quality hospitals but also help low-quality hospitals improve. All of these are examples of the tuna-dolphin challenge: we want to cut out bad healthcare—that is, yank the tuna out of the water—but we don’t want to limit good healthcare—we want the dolphins to happily swim away.

This book is something of a tour of health economics, and maybe by the end you’ll see how regulating the healthcare sector is an endless series of dolphin-tuna challenges. What we hope to do in this book is explain why.

Who We Are

For nearly twenty years, ever since we met as graduate students at MIT, we’ve worked as health economists. We’ve written half-a-dozen papers together and a couple dozen papers with other coauthors. We’ve taught courses on health economics. One of us, Tal Gross (hereafter Tal), taught health economics to students getting master’s degrees in public health at the Mailman School of Public Health at Columbia University, and then to students getting a master’s in business administration at the Questrom School of Business at Boston University. The other, Matt Notowidigdo (hereafter Noto), taught health economics to undergraduate students at Northwestern and later to students getting their master’s in business administration at the Booth School of Business at the University of Chicago.

Those courses are a tour of health economics—how to think about health insurance, healthcare delivery, and the life-sciences industry. Our notes for these courses would collectively fill a giant three-ring binder: materials for class, summaries of research, exhibits.

If we could, we would drag that three-ring binder to every cocktail party to which we are invited. And then, when people ask us what it is that we do for a living, what health economists actually do, we would pull out that binder and walk them through it.

But it’s a party, not a deposition. And while some eccentricities are socially acceptable, pulling out a binder in the middle of a party is not.

And so, instead, we have turned that three-ring binder into this book. This book is an answer to all of the people who wonder what it is that health economists do. It’s a summary of our health economics courses, a summary of what we believe our field has accomplished in the fifty-odd years that it’s existed.²

Part I

Demand

1

What Does Health Insurance Do?

To many people, health insurance is boring. It’s a matter of faxed paperwork and deductibles and your call is very important to us, please stay on the line. Health insurance is a financial product: it determines who pays for your healthcare. For most people, that’s about as thrilling as filing their taxes.

But in the United States, unlike every other wealthy country, not everyone is lucky enough to have health insurance. About 50 million people were uninsured a few years before the Affordable Care Act—also known as Obamacare—was implemented in 2014, and about 25 million people were still uninsured a few years later. In other words, let’s ask the most American of questions: What does health insurance do?

The Fundamental Purpose of Health Insurance: A Red Sox Parable

In the fall of 2007, we were graduate students at MIT. Jenny, Noto’s then girlfriend and now wife, bought a dining room table. The table was solid cherry and expensive: it cost a thousand dollars. We, being graduate students, couldn’t afford a $1000 table, but there were a couple factors that brought that $1000 table into Noto and Jenny’s apartment. At the time, Jenny was a software engineer and she spent her weekends decorating her and Noto’s Cambridge apartment. Even if Noto couldn’t afford the table, she could.

But there was something else that made the table more affordable that fall. Jenny bought the table from a large Boston chain called Jordan’s Furniture. And that fall, Jordan’s Furniture ran a promotion. If the Red Sox won the World Series, then all of the furniture they sold would be free.

So Jenny bought the table and then happily explained to Noto the price. Yes, it cost a thousand dollars. But if the Red Sox won, they would get the money back. The Red Sox had previously won the World Series in 2004, the first time they’d done so since 1918. In the minds of Boston’s fundamentalist sports fans, the 2004 victory ushered in a new era in which the Red Sox would probably win the World Series every year. By 2007, when Noto and Jenny bought the table, the team seemed downright overdue.

Noto recognized right away that he and Jenny faced a very particular form of uncertainty. In one timeline they would get a thousand dollars and a free table—in the other, they would get nothing and would have bought a very expensive piece of midcentury-modern furniture (table 1.1).

Table 1.1. Uncertainty Jenny Faced

Most people detest uncertainty, and because Noto is an economist, he detests it even more. So Noto came up with a scheme that would eliminate the uncertainty. At the time, a popular website, Tradesports.com, made it easy to bet on sports teams. Noto used the website to place a $500 bet against the Red Sox.¹ If the Red Sox lost the World Series, the bet would pay out $1,000.

Why did Noto do this? Well, think about the two possible outcomes here as two different timelines: one in which the Red Sox win the World Series, the other in which the Red Sox do not win the World Series (table 1.2).

Table 1.2. Jenny’s Situation after the Bet

Noto used the bet as a way to transfer resources across two hypothetical timelines. The bet let him move $500 from the timeline in which the Red Sox won to the timeline in which the Red Sox lost. And that transfer of resources eliminated the uncertainty that he and Jenny faced: once Noto made the bet, she would be up $500 regardless of who won the World Series.² But to enjoy that certainty, Noto had to pay $500 for the bet.

Insurance, at its core, offers the same service. Insurance allows people to transfer resources across timelines.³ After all, think about car insurance. One day, you park your car and run inside a store for an errand. While your beloved car is just sitting in the parking lot, another driver plows into it, causing thousands of dollars’ worth of damage. That’s one timeline, and a pretty bad one. The other timeline is realized all other days, when your car is not damaged. Without insurance, you would face uncertainty: either you’re going to pay a mechanic thousands of dollars tomorrow or you’ll have no need for a mechanic. Either you’ll spend a lot of money, or you’ll spend nothing. Car insurance reduces that uncertainty: each month you pay a set premium rather than facing the risk of thousands of dollars in repairs.

Health insurance is similar. If we become ill, health insurance allows us to receive thousands of dollars in medical care for a fraction of the cost. The primary purpose of health insurance is to resolve that uncertainty, to be there in case we get ill or injured, by committing to the fixed cost of a health insurance plan. This leads to a textbook answer to the question What does health insurance do?

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.⁴

Risk averse individuals—people who prefer to avoid risk—is key. Without aversion to risk, there would be no demand for insurance.

For decades, policymakers have struggled to get all Americans covered by health insurance. President Roosevelt tried and failed in 1939, President Nixon tried and failed in 1974, and President Clinton tried and failed in 1993. Why, for so long, have American politicians spent scarce political capital in the hopes of getting all Americans on a health insurance plan?

Sometimes, politicians would link the effort directly to risk aversion. When signing Medicare into law in 1965, President Johnson declared, No longer will illness crush and destroy the savings that [Americans] have so carefully put away over a lifetime. President Obama also emphasized the financial security provided by the Affordable Care Act.

Given that definition, you might assume that economists have been studying the financial consequences of health insurance for decades. But when we started graduate school, there had been very little work on health insurance’s financial effects. Most of the research we read studied the effect of health insurance on healthcare and health-related outcomes: Did having health insurance lead to more doctor’s visits? More preventive care? Those are certainly interesting questions, but we start first by focusing on the financial effects of health insurance. We do so because health insurance is, first and foremost, a tool for handling financial risk.

The Effect of Health Insurance on Financial Health

Noto used online gambling as a financial product to reduce the uncertainty he and Jenny faced.⁵ Health insurance is also a financial product: it primarily determines who pays for your medical care. And so the first effect of health insurance we examine is its financial impact.

We have been interested in the financial effects of health insurance since 2006. That year, we were both graduate students in the economics department at MIT. And we were struggling graduate students, struggling with a fundamental challenge that faces graduate students in economics. Doctoral programs in economics require a couple years of classes—exams, problem sets—and then several years of research. The transition between those two pieces is often difficult. For a couple years, a graduate student is handed well-defined, concrete problems to solve. And then, all of a sudden, the student is told: you are a researcher now. Come up with a paper on your own. Find a way to impress the entire field of economics: we’re done teaching you—now you teach us.

And so, for months, we stared at blank sheets of paper. Occasionally, we would come up with an idea to research.⁶ We would bring the idea back to our thesis advisers, only to be told that it wasn’t good enough, had already been published, or was too ambitious. This was a slog.

We would occasionally meet for lunch and commiserate. For us, one image sums up the student-researcher transition: the two of us eating overpriced sandwiches at a depressing chain restaurant in Kendall Square. Noto, for some reason, would always buy a VitaminWater. Tal would order a tuna salad sandwich. And we would show each other notebooks filled with failed ideas, potential research topics that our advisers told us were not good enough.

Eventually, during one of those lunches, the sun started to shine. We started talking about bankruptcy. Bankruptcy is a legal procedure that allows Americans who are overwhelmed by their debts to get a fresh start.⁷ Those considering bankruptcy are typically in dire straits. Perhaps their small business has failed, or perhaps they’ve recently been laid off. Bill collectors are constantly calling. Their credit cards are maxed out. And then they find a bankruptcy attorney. For a fee, the attorney files paperwork on their behalf, and a judge reviews the case. Sometimes, the judge will require some assets to be sold and the proceeds distributed to creditors. But, usually, after a lot of paperwork, all of the person’s debt is wiped away, and they get a fresh start. (At the same time, their credit scores plummet and don’t recover for years.)

We started to wonder whether health insurance might play a role in all of this. In fact, there had been some discussion among law scholars as to how much healthcare costs might drive Americans to bankruptcy.⁸ Over lunch, we discussed those papers, and then we had an idea. What if we could somehow test whether health insurance coverage affects the risk of bankruptcy?

That day in Kendall Square, we discussed a particular thought experiment. Imagine that we took thousands of Americans and randomly selected some to be covered by Medicaid. Then we took thousands of others and made them uninsured. Those covered by Medicaid would face few medical bills. Medicaid in the United States is publicly provided health insurance for low-income households. There are few copayments and no deductibles. Meanwhile, the uninsured face the full price of healthcare. If they visit an emergency room, the hospital will often send them a bill for thousands of dollars.

We discussed the idea at length. And then we asked our thesis advisers what they thought. To our surprise, they thought we were on the right track. Jon Gruber, an MIT health economist, was especially encouraging. He sent us an e-mail. There are about 1 million bankruptcies per year. So you would need a pretty big change in medical expenditure risk to see a measurable effect on bankruptcies. The only really good candidate here is Medicaid expansions. You could see whether Medicaid expansions lowered bankruptcy.

And with that encouragement, we started to poke around. Medicaid programs began in the 1960s and originally only provided coverage to very young children and pregnant mothers. Over time, though, states gradually expanded who was eligible. In the 1990s, states started to allow low-income parents onto Medicaid and also older children. We collected data on state-level expansions of Medicaid in the 1990s. Some states were especially aggressive in expanding Medicaid: California and Missouri, for instance, made large shares of their populations eligible for Medicaid during that time. Meanwhile, other states—for instance, Texas and South Carolina—were much slower to expand coverage.

We then compared bankruptcy rates across states. Nearly all states expanded Medicaid in the 1990s, and nearly all states saw an increase in bankruptcy rates during that time due to broader trends in the economy. But the states that expanded Medicaid the most saw the smallest increases in bankruptcy rates. We then dug in and analyzed the relationship between Medicaid expansions and bankruptcies in various ways. 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.⁹

We wrote this all up and eventually published it in the Journal of Public Economics as one of our first papers. As of this writing, the paper has been cited over 250 times, which is not a bad outcome for two depressed graduate students sadly sipping VitaminWater and eating overpriced tuna salad sandwiches.

What’s more, this basic finding has been replicated and demonstrated by other researchers. Bhash Mazumder and Sarah Miller studied the Massachusetts health insurance expansion of 2006. In that year, Massachusetts enacted healthcare reform with the goal of achieving universal coverage. The policy was, eventually, the inspiration for Obamacare. All Massachusetts residents were compelled to purchase health insurance—those who did not would face tax penalties. And, indeed, over just a few years, the uninsured all but disappeared in Massachusetts.

Mazumder and Miller studied the effects of the Massachusetts reform in credit bureau data. They found a very clear relationship: the more Massachusetts residents who gained coverage, the fewer bankruptcies they observed. What’s more, Mazumder and Miller looked at all of the other markers of financial distress beyond bankruptcy. As more people gained coverage there were reductions in the amount of debt that was past due, higher credit scores, and fewer third-party collections.¹⁰ Gains in insurance coverage led to drops in all of those bad financial outcomes. That study by Mazumder and Miller, and our own prior work on bankruptcy, demonstrated a relationship between expansions of health insurance and a reduction in financial distress.

Researchers have also studied how health insurance helps those dealing with severe bouts of illness or injury. Noto wrote a paper with three coauthors—Carlos Dobkin, Amy Finkelstein, and Ray Kluender—that studied the effect of health insurance on people in crisis. The team linked data covering almost every hospitalization in California over four years to each patient’s credit report. Every American who has ever taken out a loan has a credit report—it lists their balances on all accounts and also their credit score, a measure of the chance that they’ll default on another loan. Then they focused on California residents who suddenly appear in the hospital for the first time. Most of those people were dealing with a first heart attack or stroke: they were suddenly grappling with a new health crisis. What did that crisis do to their credit report?

For those who had health insurance, the sudden hospitalizations barely affected their credit records. But for those who came into the hospital uninsured, things were much different. Figure 1.1 shows that, on average, a hospitalization leads to about $6,000 in unpaid medical bills for people who are uninsured. And this average, as one might expect, conceals a lot of variation at the extremes. About 10 percent of the uninsured who are hospitalized end up with over $18,000 in unpaid medical bills. By contrast, for those with health insurance, a hospitalization leads to a tiny increase in unpaid medical bills.¹¹

Figure 1.1. The impact of a hospitalization on unpaid bills for uninsured and insured adults. The data link consumer credit reports to all non-pregnancy-related hospitalizations in California between 2003 and 2007, limited to adults ages 25–64. The horizontal axis plots the months relative to the hospital admission.

To date, many other researchers have written similar studies.¹² At this point, the evidence is very clear. Health insurance leads to lower out-of-pocket medical costs. Those lower out-of-pocket medical costs lead, in turn, to reduced financial distress. In other words, more health insurance means fewer bankruptcies, defaults, and delinquencies. Health insurance is good for people’s finances.

What Does Health Insurance Do to Healthcare?

Around the time that we were sitting down for lunch in Kendall Square, another research project was just getting off the ground. Finkelstein, at the time, was an assistant professor at MIT and just making her name as an up-and-coming star in economics. One day, she was driving to work and, like all professors, listening to NPR on her car radio. That day, the reporters were discussing a story out of Oregon. The state of Oregon had decided to expand Medicaid, but it had a problem. In 2008, funding was sufficient to add 10,000 new Oregon residents to Medicaid, but many more than 10,000 people were interested.

After some debate, the state decided to run a lottery. Of the 90,000 Oregon residents who signed up for the lottery, 35,000 were selected at random to be invited to enroll in Medicaid. About a third of those lottery winners filed paperwork properly and were actually enrolled onto Oregon’s Medicaid program.

When the NPR reporter started talking about the lottery, Finkelstein immediately pulled over and took out a notebook. The word lottery got her excited. That was a big deal. A lottery meant that the state of Oregon was actually running a randomized clinical trial, even if no one there had realized it. Researchers could compare Oregon residents who won the lottery to Oregon residents who did not. Even better, since both these groups were large and selected at random, they were likely to have similar characteristics: people who won the lottery and people who lost the lottery likely had the same health factors (obesity, smoking, etc.) and demographics (age, race, etc.). And so Finkelstein could compare lottery winners to lottery losers to understand how health insurance affects behavior.

When the NPR story was over, Finkelstein finished scribbling in her notebook and drove the rest of the way to work. As soon as she got to her office, she started making calls. Would the state of Oregon allow her to study the effects of the lottery?

It took years, but Finkelstein eventually teamed up with a group of researchers who were as excited about the lottery as she was. She joined forces with Heidi Allen, Kate Baicker, Sarah Taubman, and Bill Wright. The team matched Oregon’s records on lottery numbers to records from local hospitals on ER visits. And that allowed Finkelstein and her coauthors to use the lottery as a way to test how Medicaid coverage affects the degree to which Americans use the healthcare system.

What did they find? Oregon residents who won the lottery were more likely to consume healthcare—all types of healthcare. They 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 (fig. 1.2). The results were clear: having health insurance leads to more healthcare.

Figure 1.2. The effect of Medicaid coverage on emergency room (ER) visits using data from the Oregon Health Insurance Experiment. The first panel shows the effect of gaining Medicaid coverage on the number of ER visits by comparing the average number of ER visits for those who gained Medicaid coverage through Oregon’s Medicaid lottery to the average number of ER visits for those who did not win the lottery and ended up in the control group. It shows an increase in ER visits of about 0.4 visits per year, or about a 40 percent increase over the average number of visits for the lottery losers. The second panel reports results for the share of the population with an ER visit (rather than the average number of visits). Gaining Medicaid coverage through the lottery increased the share with an ER visit by about 7 percentage points.

Before the study was complete, some experts expected a different outcome.¹³ There’s a stereotype about the uninsured in the