Next Steps for Federal Health Care Policy

The Marshall J. Seidman Lecture in Health Policy at Harvard University
Douglas W. Elmendorf
November 2, 2015

Thank you. I am truly honored to be presenting the 15th annual Marshall J. Seidman Lecture, following in the steps of a number of my mentors and friends. I am also delighted to be here with many people with whom I have had the chance to work on health care issues. And I want to thank my former colleagues at the Congressional Budget Office (CBO), who have not seen these remarks but who taught me so much during the six years I was director.

I will use this opportunity to offer my recommendations on the next steps for federal health care policy. Discussions of health care policy generally involve two broad topics—insurance coverage and costs. I will talk about both today, but in rather different ways. For insurance coverage, I think federal policy has established the right framework, and rather than describe specific next steps within that framework, I will explain why I think we should keep that framework. For health care costs, I think federal policy has made less progress, and I will propose a set of significant changes.

Health Insurance Coverage

Let me begin with insurance coverage. As you know, the Affordable Care Act (ACA) significantly expanded federal subsidies for health insurance for lower-income people, and it made important changes to the rules governing insurance markets. In my view, the system of rules and subsidies established by the ACA is the right system in its fundamental elements, although we could discuss the pros and cons of specific changes within that system. My view is based on two judgments. The first is a value judgment—I think we should bear the cost of achieving nearly universal health insurance in this country—and the second is an analytic judgment—there are no alternatives to the ACA framework that would achieve that goal at significantly lower cost. Let me explain.

A Value Judgment

Because I am an economist, not an ethicist, I will not spend long on the value judgment. There are other valid value judgments; I am just trying to be explicit about mine. But I will note two points that influence my value judgment.

One is that incomes of people across most of the income distribution have benefited very little from the growth of total output and income in this country in the past few decades. To me, that pattern increases the importance of focusing economic policy on helping people of modest means. I have written recently about how that focus affects the policies that one should support in encouraging economic growth and in helping older Americans.

The other point that influences my value judgment is that health insurance is valuable for people to have. Studies of the Oregon Medicaid experiment show some clear benefits of insurance but also show ways in which insurance seems to be less valuable than one might expect. Indeed, one recent paper argued that the benefits received by new Medicaid recipients were less than the government’s costs. However, the key driver of that result was the finding that poor, uninsured people pay only about a fifth of the cost of their care. To the extent the rest is paid by extended families, those other people are likely to have low incomes as well, and to the extent the rest is borne by health care providers, subsidizing insurance explicitly makes more sense. So, those results do not demonstrate that expanding insurance coverage has low value on the whole.

What Are the ACA’s Coverage Provisions Doing?

That brings us to the analytic judgment that the ACA framework is the most effective way to achieve nearly universal health insurance. Let us review what the ACA is doing and then consider alternatives.

In 2015, about 10 million people are insured through exchanges and about 10 million more are covered by Medicaid because of the ACA; the great majority of those people would otherwise be uninsured. In future years, the number of people who will be insured because of the ACA will reach 20 million or more. Whether we achieve nearly universal coverage depends on one’s definition of the term and on future enrollment patterns. In CBO’s projections, the ACA ultimately raises the share of nonelderly people who have health insurance from about 80 percent to about 90 percent, and almost all of those who remain uninsured are either unauthorized immigrants and thus ineligible for coverage under the ACA, or eligible for federal subsidies through Medicaid, exchanges, or their employers but choosing not to enroll. Providing and paying for health care for the people who remain uninsured is an important challenge, but a challenge that would exist under any feasible alternative as well. Even with that challenge, and even if CBO’s projections are too optimistic, the increase in coverage under the ACA is striking.

The costs of that increase are about $6,000 of federal subsidy per newly insured person in 2017 (as an example) and a collection of significant changes in insurance rules. That dollar cost—which includes both the direct subsidies and the tax revenue lost indirectly from a reduction in labor supply—is substantial, but it is not surprisingly high given the cost of health care. It also bears emphasis that the ACA subsidies will be only about 10 percent of total federal subsidies for health insurance over the next decade: About 45 percent of those subsidies will go to Medicare, 25 percent to Medicaid apart from the ACA, and 20 percent to tax expenditures for employer-sponsored insurance. Essentially, the ACA is providing subsidies to the largest group that was not getting subsidies before, namely lower-income people who were not eligible for Medicaid or employer-sponsored insurance—and that group is small and healthy enough that the cost of subsidies for them is much less than the cost of the subsidies that were already in place.

The changes in insurance rules have generated no major disruption to coverage for the great majority of people who had health insurance before the ACA. The changes have forced some people to buy broader and therefore more expensive insurance coverage than they would like, and the changes have enabled some people to buy broader insurance coverage they prefer but could not buy previously because of adverse selection. Those groups are much smaller, though, than the group of people whose health insurance has continued with no notable changes.

What Are Alternatives to the ACA’s Coverage Provisions?

That is what the ACA is doing. What about alternative ways to expand health insurance coverage relative to what it was before the ACA?

One sort of alternative involves a larger federal role. For example, some people object to providing subsidies for private health insurance and instead would expand the fee-for-service portion of Medicare to cover everyone in the country. But that is not the sort of alternative envisioned by most critics of the ACA. Most critics support a smaller federal role, with much less money spent on subsidies and much looser rules imposed on insurance markets. We have seen proposals of this sort from some analysts, some Members of Congress, and some Presidential candidates.

That approach of smaller subsidies and looser rules might have various advantages, but there is no analytic basis for expecting it to achieve the same level of insurance coverage as the ACA. Because health care is expensive, raising coverage without imposing an excessive burden on lower-income people requires substantial subsidies. Because health care spending varies widely among people in ways that can be partly predicted and because society provides emergency care for those who need it, health insurance markets cannot function entirely like markets for most goods and services. In my time at CBO, we explored many changes in health insurance subsidies and rules, often working with the majority or minority staffs of key committees. Nothing in that process suggested there are easy ways to significantly expand coverage. That point was not so clear to me, by the way, when the ACA was being designed and enacted, but it has become more clear over time.

Certainly, the advocates of smaller subsidies and looser rules claim great things for their proposals. However, I am not aware of any alternative to the ACA for which we have a reliable, independent estimate of its effects on insurance coverage, the federal budget, and insurance markets that could be compared with the effects of the ACA. The lack of such an estimate after five years of heated debate about repealing and replacing the ACA is itself revealing. If the chairman of any Congressional committee had publicly endorsed a specific replacement and requested an estimate of its effect, CBO would have produced and published that estimate—and if there were alternatives to the ACA that would cover roughly as many people at much lower cost, someone would be eager to publicize it. But there are no published CBO estimates for replacement proposals.

Based on the evidence I have seen, achieving the substantial increase in insurance coverage that is occurring under the ACA requires subsidies and rules similar to those of the ACA—or an even larger federal role. Of course, within the ACA framework, we could make many specific changes, and we should have a vigorous discussion about those, just as we discuss changes in other public programs. But unless we are willing to reverse the sharp gains in health insurance coverage that we have seen in the past few years, we need to keep the ACA framework.

Health Care Spending

Let me turn now to the second broad topic of health care policy, which is costs—or, more accurately, spending, because the issue is not just the cost of individual medical services but the total amount we spend on health care and what we spend it on. As you know, health care spending in this country is very high relative to spending in other countries, even adjusting for our higher income, and it has grown much faster here in recent decades. In the past several years, though, the growth rate of spending has slowed considerably. Those patterns provide important context for policymaking, so I will talk briefly about them before turning to policy recommendations.

Why Are We Concerned About High and Rising Health Care Spending?

The high level and faster growth of health care spending in this country do not prove that we are doing something wrong: We might simply have a greater demand for medical care and be rationally allocating more of our resources to it. However, some other indicators suggest that we are, in fact, doing something wrong.

One indicator is that we are providing mostly unbounded public subsidies for health care. For people with private health insurance—the majority of people under age 65—the federal government provides an open-ended subsidy by excluding employer payments from taxable income. For people in the fee-for-service part of Medicare—the majority of people age 65 and over—the federal government sets prices but leaves quantities to the discretion of providers and patients.

A second indicator we are erring is that, despite high spending, Americans appear to be less healthy, on average, than people in other advanced countries. To be sure, health outcomes are affected not only by health care but also by diet, exercise, family structure, and many other factors. In addition, other countries are free-riding on our health care system in some ways, with the development of drugs as one prominent example, and there is research showing that treatment for certain health conditions is more effective in this country than in others. Still, given the huge differences in the resources devoted to health care across countries, one might expect that health outcomes would be clearly better here, and they are not.

A third indicator we are doing something wrong is that health care spending varies significantly across the country, even after controlling for regional characteristics, but health outcomes do not vary correspondingly. Those differences in spending probably stem partly from patient characteristics and partly from providers’ practice patterns. If differences in practice patterns matter for spending but not for outcomes, then the additional spending seems to have low value.

An important development in the past several years is that growth in health care spending has slowed a lot. That slowdown has been very broad-based: We have seen it in private insurance and in federal health programs, and in Medicare, we have seen it in every major type of service and region of the country. The slowdown has also been substantial in magnitude and persistent, beginning roughly a decade ago and continuing through the present. The causes of the slowdown are not entirely clear. The weak economy, the spread of high-deductible insurance plans, and changes in the behavior of providers and beneficiaries for other reasons have probably all helped. The ACA is having mixed effects—raising spending by increasing care for the newly insured, but restraining spending through cuts in the growth of payments to Medicare providers, changes in the ways providers are paid, and perhaps a broader signal that payments for health care will be more constrained in the future.

The slowdown in spending growth has had important effects on the outlook for health care spending. For example, CBO expects that, under current law, inflation-adjusted Medicare spending per beneficiary will increase at an average annual rate of 1¼ percent over the next decade, compared with 4 percent in the two decades ending in 2007. That difference reflects extrapolation of the slow growth in recent years as well as constraints on payment rates and an influx of younger beneficiaries. For federal health care programs as a whole, CBO projects that growth in spending as a percentage of GDP over the next 25 years will be driven as much by population aging as growth per beneficiary, which is quite different from the projections several years ago. And CBO’s projections of national health care spending have also declined in the past few years.

Still, with a significant share of our health care paid by insurance on a fee-for-service basis, the incentives that have pushed up spending in the past remain largely in place, so health care spending per person will probably continue to rise notably faster than income. That matters a lot because the resources we are devoting to medical care are already so large. National health expenditures exceed 15 percent of GDP today and are probably heading to 25 percent over the next 25 years without changes in policies. The federal government devoted more than 1¼ trillion dollars to health care subsidies last year; measured as a percentage of GDP, federal subsidies for health care are currently about 7 percent and rising.

The rising cost of federal subsidies for health care is a key factor behind the unsustainable trajectory of federal debt, and those subsidies are crowding out crucial spending for other purposes. Moreover, the rising costs for health care paid by households, businesses, and state and local governments are crowding out other important spending.

What Strategies Could the Federal Government Pursue?

Therefore, it is very important that we make good choices about how much to spend on health care and how to allocate that spending. Making better choices than we are making today will require a lot of smart and hard work by health care providers, health insurers, private purchasers, and state and local governments. My focus, though, is on the federal government. There are two broad strategies the government could pursue: One is to increase the role of the federal government in centrally managing the health care system; the other is to increase the role of market forces in decisions about health care.

Part of the case for greater centralized management would be that most advanced economies have a larger role for the government, spend less for health care, and have better health outcomes—although, as I noted, interpreting those facts is complicated. The case would continue that health care costs per beneficiary have increased slightly less rapidly in Medicare and Medicaid than in the rest of our health care system—although that comparison also is complicated to interpret.

Notwithstanding those and other arguments, however, I do not think a strategy of greater federal management of health care is feasible in this country. Advocates of a larger role for market forces would argue that competition by private firms facing price-sensitive consumers generates tremendous gains in well-being in many markets. They would continue that government insurers inevitably have less flexibility than private insurers to adjust benefits, change coinsurance, follow different treatment approaches, manage utilization, determine provider networks, and make other changes to reflect both preferences and cost pressures. Moreover, many Americans have a significant distrust of the federal government. Given the strong negative reaction to the ACA by many people, increasing federal control of health care is a nonstarter.

Then, to make better choices about health care spending, we need to increase the role of market forces. We need to change incentives and foster competitive markets so that people make choices about health care that are more cost-effective—that is, choices that reward care of high quality and low cost but penalize care of low quality and high cost. And we need to accomplish this in ways that protect individuals who need medical care from excessive costs at the time of service and that protect people of modest means from excessive insurance premiums. That is a tall order. Generating appropriate incentives and market structures is inherently difficult because of patients’ lack of understanding of the sorts of care they need and the quality of care they receive; many providers’ natural tendency to do too much rather than too little; the extreme skewness of the distribution of health care spending; and other factors.

As a result, it is unclear how much our choices about health care would be improved by any particular federal policies. Still, we can make real progress. I think we should strengthen the role of market forces in four ways.

Do Not Provide Public Subsidies for Health Insurance That Is Unduly Expansive

First, we should not provide public subsidies for health insurance that is unduly expansive.

In the ACA’s insurance exchanges, plans must cover a certain scope of health care services and have an actuarial value (the share of spending for covered services that is paid by insurers) above a threshold; a person’s subsidy depends on his or her income and the premium of a plan with an actuarial value of 70 percent. If someone chooses a plan with a broader scope or higher actuarial value, he or she pays all of the premium difference with after-tax dollars. Thus, the subsidies are based on the cost of insurance that meets minimum standards; they are not open-ended. By contrast, our income and payroll tax systems provide an open-ended subsidy for employer-sponsored insurance: However broad the scope or high the actuarial value, employers’ payments are fully excluded from taxable income. The ACA effectively limited that subsidy beginning in 2018 through the backdoor method of imposing an excise tax—the so-called Cadillac tax—on plans with premiums above a threshold, with some adjustments and exceptions.

That limitation is an important advance, and repealing the Cadillac tax would be a significant mistake—not primarily because of the lost revenue but because of the lost incentive to restrain health care spending. Restoring the open-ended subsidy would encourage insurers to reduce deductibles and co-payments and to reduce their efforts to manage utilization and negotiate aggressively with providers. Those responses would increase health care spending and insurance premiums, which would lower other forms of compensation, such as cash wages. In addition, repealing the tax would increase budget deficits, mostly to the benefit of higher-income people.

To be sure, some people pay more for insurance not because they have more expansive plans but because they are in insurance pools with less healthy people. To the extent that regulations implementing the tax can adjust for that effect, so much the better. However, the tax code has many nominal thresholds that correspond to different amounts of real spending or standards of living for people in different areas of the country, and we accept those thresholds as second-best policies. That same logic applies here. Other tax policies could also eliminate the open-ended subsidy provided by the current tax exclusion, but those policies would have disadvantages similar to those of the Cadillac tax.

Another step in eliminating subsidies for insurance that is unduly expansive is to prevent medigap policies from covering all of Medicare’s cost-sharing requirements. For example, we could bar medigap policies from paying the first several hundred dollars of an enrollee’s cost-sharing obligations and limit their coverage for the next several thousand dollars of cost-sharing to only a fraction of those amounts. Those changes would make beneficiaries more sensitive to the cost of medical services.

Move to a Premium Support System in Medicare

Second, we should increase competitive pressure in Medicare by moving to a system of premium support, in which all beneficiaries would choose between competing insurers and bear the entire difference in cost between their choice and a benchmark.

That system would be much like the subsidies in the ACA’s insurance exchanges and could be viewed as an expansion of the current Medicare Advantage program. About 30 percent of Medicare beneficiaries are currently enrolled in private plans through Medicare Advantage, and CBO projects that share to rise to about 40 percent during the next decade under current law. However, the structure of Medicare Advantage does not create the most effective competition among private insurers or between private insurers and fee-for-service Medicare.

We should move to a premium support system in two steps. The first step is to change Medicare Advantage to strengthen competition between private plans: Specifically, to base the benchmarks for federal subsidies on insurers’ bids rather than being preset amounts, and to make beneficiaries pay all of the premium amounts above the benchmarks and receive all of the savings relative to them. The second step is to include fee-for-service in the bidding system. Moving in two steps rather than one would provide a chance to address problems that may arise in competitive bidding before the fee-for-service option is incorporated. It would also offer additional time to address the challenge of risk adjustment. Risk adjustment has worked fairly well in Medicare so far, in part because price competition has not been that strong. But premium support is designed to boost price competition, and some of the differences in prices—in particular, between private insurers and fee-for-service—will be very large relative to previous Medicare experience. As a result, the pressures on risk adjustment will be much greater. Although CBO concluded in a report a few years ago that risk adjustment probably would be good enough to maintain a stable insurance market, reinsurance or some other protective mechanism might be needed.

Maintaining fee-for-service as one of the competing insurance options would be important, both because some beneficiaries are more comfortable with it and because maintaining it would lower overall costs (for reasons described by CBO). In addition, setting benchmarks based on the average bid in each region—rather than a lower bid, as is sometimes suggested—would sensibly limit the shift of costs onto beneficiaries. Those and other design choices would have important effects on how the arrangement worked.

A premium support system would have significant benefits. As we have seen with Medicare Part D and the ACA’s insurance exchanges, competition restrains costs. According to CBO’s estimates, the approach outlined here would reduce both federal costs and beneficiaries’ costs (premiums plus out-of-pocket payments) by about 5 percent. However, that reduction in beneficiaries’ costs is an average figure that masks considerable variation among people: Some would pay significantly more than under current law, with the biggest burden borne by those who stayed in the fee-for-service program in regions with high fee-for-service costs. The savings for both the federal government and beneficiaries would not be larger in part because fee-for-service payments in Medicare are scheduled to increase slowly, and private insurers will have difficulty lowering their relative costs. Still, the estimated savings are notable. And in addition to the figures I just mentioned, the heightened competition would probably lower long-term spending growth a bit by shifting demand toward less costly treatments. Moreover, beneficiaries’ choices in a premium support system would provide a useful signal about how much they value the benefits that come from additional spending for health care.

Even under a premium support system, however, a substantial share of Medicare beneficiaries would enroll in the fee-for-service option because of the combination of price and other characteristics it would offer. That brings me to the third item on my agenda for increasing the role of market forces in health care.

Restructure Payments to Medicare’s Fee-for-Service Providers to Increase High-Value Care

Third, we should restructure payments to Medicare’s fee-for-service providers to create stronger incentives for high-value health care.

In their basic form, fee-for-service payments represent incentives for providers to deliver more care and more complex care—even when that care does little to improve people’s health—and the payments provide no incentive to deliver higher-quality care. We know that we need to change those incentives by shifting from fee-for-service to alternative payment models. The ACA lowered the growth rate of fee-for-service payments, which will make providers more willing to consider alternatives, and it set a number of payment reforms in motion. The Department of Health and Human Services is pursuing those reforms now.

This effort should proceed with as much vigor and urgency as possible: Significant changes to the payment structure in Medicare would lead to higher-value health care not only for Medicare patients but also for patients with private insurance, which often follows policies adopted by Medicare. But we need to recognize that this process of change is very difficult: Many experiments with alternative payment models have had disappointing results, at least in part because providers have been unable or unwilling to change their behavior enough to deliver significant reductions in cost or improvements in quality. That history should not discourage us from continuing to experiment, but it should temper our expectations.

I will not try to offer a full list of desirable changes in payment models, but let me highlight two directions for policy that are important. One is expanding the bundling of payments, in which Medicare makes one payment for all of the care associated with a particular health episode rather than paying each provider separately. Such bundling can both reduce Medicare’s costs and improve the coordination of care because providers that are being paid together will tend to work together more closely. Medicare is already expanding its use of bundled payments, and it should keep going as quickly as possible.

Another important policy direction is to stop paying the full cost of every new treatment that is developed. The most important driver of health care spending in the long term is the adoption of new treatments; sometimes those treatments have significant effects on people’s health at low cost, sometimes they have significant effects at high cost, and sometimes they have insignificant effects. We currently pay for medical care in ways that do not differentiate between those categories, and that should stop. Certainly, moving in this direction needs to be a deliberate process, because there are considerable risks and difficulties involved in reference pricing and similar policies. But not moving at all in this direction is difficult to justify.

Make the Underlying Markets for Health Care and Health Insurance More Competitive

Fourth, we should aim to make the underlying markets for health care and health insurance more competitive. Let me briefly describe four specific types of policies that fall in this category.

One type is to limit the extent of further consolidation by health insurers and health care providers. For competitive market forces to work, we need competitors. If a significant market segment in any region of the country has a monopoly health care provider or monopoly health insurer, that monopoly would effectively be a public utility, and we would need to regulate it as such. That tactic would fit a strategy of greater government control, but not one of greater use of market forces. Even when consolidation stops well short of creating a monopoly, it often increases the market power of providers or insurers, which leads to higher prices. Again, that is not consistent with restraining spending through market forces. Limiting consolidation of providers does mean forgoing some gains in efficiency and care coordination, but those are costs we need to bear. Limiting consolidation of insurers means forgoing some countervailing power vis-a-vis providers, but that countervailing power does not seem to hold down premiums; in addition, the effects of insurer consolidation cannot be fully offset by the ACA’s rules regarding medical loss ratios.

A second type of policy that would strengthen the role of market forces is to increase the number of competing providers in other ways. We should explore relaxing scope-of-practice laws, increasing the number of positions in medical schools, and allowing more people trained outside the country to practice medicine here.

A third type of policy to enhance competition is to help people become more-effective buyers of health insurance and users of health care. That means providing them with more information about the price, quality, and other characteristics of insurance plans and medical services they are considering. It also means providing them with tools to help them make good choices and perhaps setting default choices that reduce the harms visited upon inattentive or uninformed buyers.

And the fourth type of policy in this category is to provide strong support for the work of the Center for Medicare & Medicaid Innovation, the Agency for Healthcare Research and Quality, and the Patient-Centered Outcomes Research Institute. Without the data collection, research, and experimentation being done by those organizations, our ability to strengthen the role of market forces in health care would be greatly limited.


Let me conclude. The ability of our elected leaders to enact forward-looking health care policy during the past five years has been greatly limited by their preoccupation with two issues—the Affordable Care Act and the Sustainable Growth Rate (SGR) mechanism for Medicare’s payments to doctors. At long last, the SGR has been eliminated, and the huge amount of time that policymakers and their staffs spent crafting and debating short-term fixes to the problems created by the SGR can be devoted now to more constructive policymaking. And unless those who object to the ACA are willing to endorse an alternative with comparable information about its effects on coverage and costs, then the huge amount of time that has been devoted to the ACA debate also should be devoted now to more constructive policymaking. We have a lot to do, and we should get going. Thank you.