The Marginal Tax Rate Mess

Nancy Folbre is professor emerita of economics at the University of Massachusetts, Amherst.

After years of partisan debate over marginal tax rates on the rich, it seems we are now destined for even more acrimony over implicit marginal tax rates on the poor. When families receiving such means-tested benefits as food stamps or housing subsidies earn more income, their benefits are reduced. That’s what means-testing means.

The reduction in benefits is accurately described as an implicit tax. The only way to avoid such an implicit tax is either to provide universal benefits or no benefits at all. On a fundamental level, means-tested programs represent an uncomfortable compromise between those who want governments to help their citizens, those who don’t, and all those in between.

Almost anyone anywhere on the political spectrum would, relieved of opportunities for strategic maneuver, agree that the current configuration of means-tested programs (including the Affordable Care Act) is not nearly as equitable or efficient as it could be. The same criticism should be leveled at the current state of debate over means-tested programs, seldom characterized by open discussion of conceptual differences or clear articulation of basic assumptions.

The conceptual mess is, hopefully, easier to clean up than the policy mess. Pursuing this hope, I compare my views with those of my fellow Economix blogger Casey Mulligan, whose views on most matters economic are diametrically opposed to mine.

I advocate increased public investments in health, education and employment through largely universal programs. Professor Mulligan, both in his recent posts and his book “The Redistribution Recession,” attributes both unemployment and sluggish economic growth to excessively generous public assistance.

But Professor Mulligan and I share an interest in labor supply that distinguishes us from Keynesian economists more preoccupied with labor demand. I am going to start by building on that shared interest, by explaining our differences on one introductory question: What are implicit marginal tax rates, and why should we worry about them?

In subsequent posts, I’ll address some related issues, including the definition of labor supply, levels of spending on means-tested programs, impacts on incentives to paid employment and the relative importance of labor supply compared with labor demand. (If these are the issues you care about most, please hold your fire until I get to them.)

As a result of losing eligibility for means-tested benefits, low-income and middle-income families sometimes experience much higher marginal effective tax rates (sometimes exceeding 90 percent) than those at the top of the income distribution. Phase-outs for any one program may not be large, but participation in several programs creates a cumulative effect.

The most recent estimates from the Congressional Budget Office conclude that more than 20 percent of low- and moderate-income taxpayers face marginal tax rates of 40 percent or more, based on the effect of their earnings on federal and state individual income taxes, federal payroll taxes and the phasing out of benefits from the Supplemental Nutrition Assistance Program.

Geographical variation is huge. For instance, researchers from the Urban Institute and Brookings Institution estimate that in 2008, a single parent with two children participating in Temporary Assistance for Needy Families and the Supplemental Nutrition Assistance Program (food stamps) moving from no employment to poverty-level earnings, would actually receive more benefits in New Jersey, but experience a significant reduction in benefits in Hawaii.

High effective marginal tax rates rightfully distress the families most directly affected by them. The complexity of means-tested public assistance programs makes it difficult to calculate the gains from additional hours of employment accurately, which is inherently discouraging.

But families are probably able to identify points where they “fall off a cliff,” as when a small increase in income renders them ineligible for any Medicaid assistance.

Means-tested benefits are also politically divisive, fostering resentment among those who believe they will never personally benefit from them.

These general criticisms, however, neither rely on nor complement the assumptions that Professor Mulligan makes in his quantitative estimates. As he sets out to examine the impact of marginal effective tax rates on a variety of different employment outcomes, Professor Mulligan relies on measures of the combined impact of taxes and benefit reductions, following the common practice of including Social Security taxes paid by both employers and employees.

But as a recent C.B.O. report on marginal effective tax rates notes, inclusion of Social Security taxes is problematic, because the payments made by both workers and employers offer future benefits in the form of retirement income and survivors insurance. We don’t consider expenditures on private insurance as a tax payment, so why should expenditures on public insurance be considered as such?

Nor are these explicit taxes the only ones that purchase public insurance or help finance direct benefits to taxpayers. Safety-net programs such as Temporary Assistance to Needy Families and the Supplemental Nutritional Assistance Program offer potential benefits to those who may need them in the future. State income taxes help finance education for many taxpayers’ families. The only tax payments that become “unavailable” to the worker are those that will be spent entirely on other people whom the taxpayer cares nothing about.

Of course, it would also be very difficult to actually measure marginal taxes net of marginal benefits. But the assumption that taxpayers derive no marginal benefits even from programs such as Social Security seems implausible to me.

If people don’t derive any utility from the taxes they pay, it is difficult to explain why many have voted for the politicians who put those taxes into place.

This reasoning suggests that benefit reductions might actually reduce utility more than tax payments, but there are other reasons that they may reduce it less. The marginal federal and state income tax rate goes up as families earn more and enter a higher tax bracket, but the marginal implicit tax rate goes down once taxpayers have exceeded the earnings at which they are no longer eligible for means-tested benefits.

In other words, low- or middle-income families may see their tax rates go up as they lose eligibility for benefits, but if they continue to work and earn more they may well reach a point at which this rate will decline.

Any worker who can estimate her own net effective marginal tax rate (a detailed calculation that exceeds the capacity and curiosity of many economists with Ph.D.’s) can also figure out that the labor market often rewards effort and experience more generously in the long run than in the short run. As any college student seeking an internship can explain, it is economically rational to work many hours for a zero wage if that effort will improve future job market opportunities.

Workers’ perceptions of their future opportunities in the labor market may affect their labor supply as much as, if not more than, their current marginal effective tax rate.

In making his argument that means-tested benefits have discouraged paid employment, deepened recession and prolonged unemployment, Professor Mulligan often describes his model as a simple “textbook analysis of labor supply.” For reasons I’ll lay out next week, that is a big reason that I believe his model is incorrect.