The Future of Entitlement Reform

Remarks Presented at the American Enterprise Institute
Dean Douglas Elmendorf
June 9, 2016

Thank you. I am delighted to be here and grateful to Jim for inviting me to participate. The report that Jim and his colleagues are releasing today is a very interesting take on entitlement reform, and I am glad we will have a chance to discuss it. However, I needed to prepare these remarks before I saw the report, so I will just offer my own views and will save a discussion of similarities and differences with the report for the panel.

To preview my conclusion, I think that entitlement reform should be pursued but should protect lower-income and middle-income Americans from substantial cuts in benefits.

What Should Be the Goal of U.S. Economic Policy?

The philosopher Friedrich Nietzsche once said: “To forget one’s purpose is the commonest form of stupidity.” One of my colleagues at the Kennedy School uses that quote to remind people of the importance of setting goals before choosing tactics. So, I want to begin by being clear about my goals for economic policy, including entitlement reform and other aspects of policy.

I think the principal goal of U.S. economic policy should be to raise living standards for lower- and middle-income people. Between 1979 and 2011, real gross domestic product (GDP) per person in this country increased about 70 percent. The Congressional Budget Office (CBO) has published estimates of income growth for different groups of the population over that period. CBO estimated that incomes before taxes and benefits—so-called market incomes—rose 16 percent in the bottom quintile of the distribution, 9 percent in the middle quintile, and 77 percent in the top quintile. You should not put much weight on those specific figures because measuring income growth is difficult. In particular, the figures probably understate the improvement in living standards because they do not fully capture quality improvements or new goods and services. However, the estimates make clear that most people have benefited relatively little in direct ways from the growth of total output and income during the past few decades.

That outcome is clearly unfortunate for the affected people. The outcome is damaging to our social cohesion and political process, because many lower- and middle-income people feel a growing distance from higher-income people and a growing frustration with our economic policies. And the outcome is harmful to our country’s international leadership because it casts doubt on the desirability of our political and economic system and because frustration with past policy choices hampers our ability to adopt new policies that would enhance our position in the world.

Therefore, as I said, the primary goal of economic policy should be to raise living standards for people of modest means. I realize that my explicit focus on a distributional objective may be unsettling to some listeners. People often talk about overall economic growth and give less weight to distributional issues on the presumption that a rising tide will lift all boats. But the rising tide in this country in the past few decades has not lifted all boats to anything like the same degree, so we need to face distributional issues head-on.

I do not view economic policy as a zero-sum activity and I do not like setting people against each other. But in making policy, we face tradeoffs between maximizing the size of the total pie and generating slices of that pie of a minimum size for people in different circumstances. We should be explicit about that. If we are focused on raising living standards for people of modest means, then we should reject policies that would boost total income but reduce living standards of lower- and middle-income people, and we should support policies that would diminish total income but raise living standards of lower- and middle-income people. That perspective matters critically for entitlement reform.

Reducing Government Benefits for Lower- and Middle-Income People Would Generally Make Them Worse Off

We need to recognize that reducing government benefits for lower- and middle-income people would generally make them worse off. That statement may seem obvious to some listeners and false to others, so I will explain why I think it is not obvious but is nonetheless true.

In the simplest economic models, more consumption is better than less, and if people lose benefits, their consumption falls, so they are worse off. We might stop the story there, but the world is more complicated than those simple models, so let me offer four counterarguments. I will also say why I find those arguments unpersuasive and conclude that the simple, intuitive story is mostly right.

One possibility is that people who lose government benefits might work enough more to maintain the same total income and consumption.

Indeed, I think that people generally would work more if their benefits were cut. For example, when I was at CBO, we estimated that the Affordable Care Act’s expansion of subsidies for health insurance for lower-income people would reduce their labor supply—and therefore, the elimination of those subsidies would increase labor supply. I view that discouragement of work as an unfortunate side effect of those subsidies.

However, cuts in benefits would not always increase labor supply and, even when they did, would not usually increase it very much. As another example, CBO analyzed the effects of cutting benefits from SNAP—formerly known as food stamps. We found that work effort could increase or decrease, depending on the nature of the benefit cuts. Specifically, if policymakers focused benefits more on people with the lowest incomes by phasing out benefits more quickly as earnings increased, beneficiaries would face a higher implicit tax rate on extra work and might work less. We also found that the change in labor supply in either direction would probably be small, because an extensive review we had conducted of research on the responsiveness of labor supply to tax rates showed that the responsiveness is generally low.

In addition, when the demand for labor falls short of the potential supply, as it has throughout the past eight years, even people who want to work may not be able to do so. Moreover, people with limited skills tend to have significant trouble finding work, even when overall labor demand is strong. And even if people did work enough more to maintain their total income after a benefit cut, the extra work would reduce their time for childcare, elder care, or other activities. This crowding out of activities by additional work would make people worse off, at least in simple models.

That leads to a second way in which the simple logic that benefit cuts hurt recipients might be wrong. Suppose that work makes people better off through channels they do not fully recognize. Then, reducing government benefits and making people work more might be in their interest.

For example, people may work too little because they underestimate the long-term value of building skills and climbing a job ladder. Or because they have trouble following through on their plans to find jobs. Or because they underestimate the importance of their working for the future work habits of their children.

I think those stories apply to some people. But evidence that those stories are quantitatively important for significant numbers of people is lacking.

Meanwhile, there is a growing body of evidence that children in lower-income families that receive certain benefits do better in the labor market when they grow up than children in families that do not receive those benefits. Researchers have shown that children in lower-income families that receive certain federally subsidized housing opportunities, health care, high-quality preschool education, and other benefits earn significantly more as adults. So, if we want children in lower-income families to have a stronger start in their lives, the evidence suggests that reducing benefits received by those families would be exactly the wrong policy to pursue.

A third way in which the simple logic that benefit cuts hurt recipients might be wrong is that benefit cuts might increase overall output enough to raise recipients’ pre-tax-and-benefit incomes by more than the loss of government benefits.

I agree that cutting benefits might increase overall output and income in the long term, by reducing either budget deficits or taxes. For example, smaller deficits or lower tax rates might, over time, increase investment in business capital and thereby productivity and wages. Indeed, at the end of a paper I wrote last fall advocating the use of dynamic scoring for major legislation, I advocated the use of dynamic distributional analysis, in which the macroeconomic effects of policy changes would be included in distributional analyses. So, I appreciate the logic of this argument.

However, there is no reason to expect that pre-tax-and-benefit incomes of lower-income people would rise by enough to fully offset the lost benefits. First, cutting benefits might actually decrease overall output and income by reducing investment in human capital—namely, the health and education of lower-income children. Second, even if overall income rose, lower-income people might not receive a significant share of that gain. And third, even if overall income rose and lower-income people received a proportional share, their gains would be smaller than their loss of benefits. Specifically, CBO estimated that nonelderly households in the bottom quintile of the income distribution in 2006 received, on average, about $13,000 in market income and about the same in benefits less taxes; therefore, a 20 percent cut in their benefits would mean a cut in their income after benefits and taxes of more than 10 percent. But CBO also estimated that reducing budget deficits by $2 trillion over the coming decade and more in later years—which would be more than the 20 percent of all means-tested spending—would raise income per person by only 3 percent after 25 years.

A fourth line of counterargument is that the supposed failure of the War on Poverty shows that government benefits hurt lower-income people even if the precise mechanisms are unclear.

I think that assessment of the War on Poverty is not consistent with the evidence. The poverty rate according to the official measure has declined only slightly over time, but that measure excludes SNAP, Medicaid, housing vouchers, the EITC, and other programs that constitute most of the War on Poverty. Measures that include those benefits show a marked decline in the poverty rate during the past several decades.

It is true that the War on Poverty does not appear to have generated much increase in the earnings of lower-income people: The percentage of people whose pre-tax-and-benefit income is below the poverty threshold has not fallen notably over time. However, I do not think we could realistically expect much progress of those earnings in an economic environment in which technological change, globalization, and other forces have hit lower-income people so hard. Moreover, most entitlement spending goes to programs in which most beneficiaries are eligible through age rather than financial need, so the spending dedicated to lower-income working-age people has been less than some observers seem to think.

The crucial good news is that our system of taxes and benefits has ensured that living standards have diverged less than market incomes during the past several decades. According to CBO’s estimates, between 1979 and 2011, the bottom and middle quintiles of the distribution experienced growth in incomes including taxes and benefits that was faster than growth in market incomes and closer to (although still less than) growth in incomes including taxes and benefits for the top quintile of the distribution. In addition, the tax and benefit system provided a critical buffer during the recent economic downturn, when average market incomes in the bottom and middle quintiles fell markedly but average incomes including taxes and benefits were more stable.

In sum, I think the counterarguments to the simple economic logic are not supported by the evidence: Reducing government benefits for lower- and middle-income people would indeed make them worse off. That would work against what should be the primary goal of economic policy. So, where does that leave us for entitlement reform?

Recommendations for Entitlement Reform

It leaves me with three recommendations about entitlements—and one about tax policy.

First, entitlement reform that could truly enhance efficiency or equity—and not just provide a cover for cuts in benefits—should receive serious attention from policymakers.

Numerous sorts of policy changes would meet this standard for serious attention. For example, I have advocated the adoption in Medicare of a premium-support system, in which all beneficiaries would choose between competing insurers and would bear the full difference in cost between their choice and a benchmark. That system could be viewed as an expansion of the current Medicare Advantage program that would strengthen price competition between private insurers and strengthen competition between private insurers and fee-for-service Medicare. I favor premium support because I think it would enhance efficiency and could be designed to ensure that lower- and middle-income beneficiaries maintained access to reliable health care.

However, some proposals for premium support have combined the efficiency-enhancing aspects of that approach with large cuts in benefits, including for lower- and middle-income beneficiaries. I would not favor proposals of that sort, and calling them entitlement reform without being explicit about the benefit cuts they include hinders the discussion of entitlement reform we should be having. My hope is that premium support and other possible changes to Medicare, Social Security, and other entitlement programs can be considered and evaluated without becoming just cover stories for benefit cuts.

Second, entitlement reform should avoid significant cuts in benefits targeted at lower- or middle-income people—as well as significant across-the-board cuts in benefits.

Reducing benefits targeted at lower- or middle-income people would make those people worse off, as I have discussed at length. Making significant across-the-board cuts in benefits would also make lower- and middle-income people worse off. For example, nearly half of elderly Americans have incomes apart from Social Security benefits that would leave them below the official poverty line, while only about a tenth remain below the poverty line after accounting for those benefits. So, large, across-the-board cuts in programs like Social Security and Medicare that provide benefits to a broad group of Americans would undermine the goal of raising living standards for people of modest means.

Let me add that I think entitlement reform should avoid across-the-board increases in benefits as well. There may be specific groups for whom benefit increases would make sense as part of entitlement reform. However, given the pressures on the federal budget arising from the aging of the population and rising costs of health care, across-the-board benefit increases would not be the best way to raise living standards for lower- and middle-income people.

Third, entitlement reform should include cuts in benefits for people with higher incomes.

CBO projects that, under current law, federal debt would rise indefinitely relative to GDP. That path is not sustainable, so we will need to make policy changes to stabilize debt relative to GDP, and I think we should ultimately reduce debt relative to GDP. I think we should accomplish that goal in part by cutting benefits for higher-income people. And when I refer to higher-income people, I do not mean just the top few percent of the income distribution, because changing just their benefits would not affect the budget in a noticeable way; instead, I mean something like the top third or half of the distribution. For example, we could change the Social Security benefit formula to reduce benefits for higher-income retirees but not for others, and we could substantially expand the number of people who pay income-related premiums in Medicare.

I recognize that this approach does not offer a free lunch. Making Social Security and Medicare more progressive would weaken the connection between an individual’s taxes and benefits, which could undermine the earned-benefit character of the programs if taken too far. Also, the lower benefits received by people with higher incomes would reduce their incentives to work and save. But I think this is the best way to reduce entitlement spending without unduly burdening those Americans who have fared the worst in economic terms in recent years.

Lastly, my recommendation for tax policy is to increase revenue above what would be collected under current law.

Twenty-five years ago, the number of people age 65 or older was about 20 percent of the number of people between ages 20 and 64; twenty-five years from now, that fraction will be about 40 percent. This represents a doubling of the number of people for whom the federal government provides substantial support relative to the number of people who are working and paying much of the tax bill.

I think the best response to this massive demographic shift and to other forces we are experiencing would be a significant increase in tax revenue relative to the size of the economy. Some of that increase is built into current law: By CBO’s projections, revenue will rise to about 19½ percent of GDP in 25 years, compared with about 17½ percent 25 years ago. I think we should enact a further increase above 19½ percent.

Given my goal of raising living standards for lower- and middle-income people, I would concentrate the tax increases on people with higher incomes. Increasing marginal tax rates would generally reduce work and saving, which would lower GDP compared with what it would be otherwise. However, the empirical evidence suggests that we can raise significant additional revenue without large reductions in labor supply or the capital stock, especially if we focused on limiting deductions, credits, and other special provisions in the individual and corporate income taxes. I would also tax carbon emissions because pricing emissions is the most efficient way to reduce them, and I would use some of the receipts to offset part of the burden on people with lower incomes.

To summarize my views on entitlement reform with the sentence with which I started: I think that entitlement reform should be pursued but should protect lower-income and middle-income Americans from substantial cuts in benefits. Thank you.