Negative Income Tax
By Jodie Allen
In its purest form a NIT promised a revolution in American social policy. Gone would be the intrusive and costly welfare bureaucracy, the pernicious distinctions between “worthy” and “unworthy” recipients, the perverse disincentives for work effort and family formation. The needy would, like everyone else, simply file annual—or perhaps quarterly—income returns with the Internal Revenue Service. But unlike other filers who would make payments to the IRS, based on the amount by which their incomes exceeded the threshold for tax liability, NIT beneficiaries would receive payments (“negative taxes”) from the IRS, based on how far their incomes fell below the tax threshold.
The NIT would thus be a mirror image of the regular tax system. Instead of tax liabilities varying positively with income according to a tax rate schedule, benefits would vary inversely with income according to a negative tax rate (or benefit-reduction) schedule. If, for example, the threshold for positive tax liability for a family of four was, say, $10,000, a family with only $8,000 of annual income would, given a negative tax rate of 25 percent, receive a check from the Treasury worth $500 (25 percent of the $2,000 difference between its $8,000 income and the $10,000 threshold). A family with zero income would receive $2,500.
Very neat. So attractive that researchers in the Office of Economic Opportunity, the brain trust for Lyndon Johnson’s Great Society in the midsixties, began planning a large-scale field experiment of the idea. Several sites in New Jersey were ultimately selected for the test, which was launched in 1968 with the University of Wisconsin’s Institute for Research on Poverty in charge of the research and a Princeton-based firm, Mathematica Inc., in charge of field operations and data collection.
The primary purpose of the experiment was to address the concerns of labor-supply theorists. These labor economists worried that providing the working poor with a basic income guarantee if they quit work, and then reducing that guarantee at a fairly steep rate as income from work increased would damage work incentives and ultimately swell NIT costs. They feared that NIT would reduce work effort in two ways. First, by giving a family with no outside income a guaranteed minimum income, NIT might, at the extreme, discourage family members from working at all. And even if workers didn’t quit entirely, they might work less since they could satisfy their basic needs with less work effort. Economists call this an income effect. Second, by reducing benefits some fraction of a dollar for every dollar earned, NIT would, along with payroll and state and local income taxes, reduce the net value of wages and induce recipients to “substitute” leisure for work. Economists call this a substitution effect.
Using a complicated model intended to minimize program costs for a given sample design, the New Jersey experimenters set out to measure the strength of these two effects. Potential participants were assigned to a variety of “treatment” cells, with the treatment being a particular combination of basic guarantee and negative tax rate. The sample included a “control” group of families not eligible for any experimental payments. Thus, a treatment might offer a guarantee equal to half the poverty line, which was then about $8,000 for a family of four, with benefits being reduced by 50 percent of the family’s income. When income was zero, the family would receive the full $4,000 (50 percent of $8,000). When income reached $8,000, the benefit would be reduced to zero and the family would “break even” (i.e., neither receive negative taxes nor pay positive taxes).
The experimenters—and the planners in what was then called the Department of Health, Education, and Welfare (myself included) who drew upon their work in designing President Richard Nixon’s Family Assistance Plan of 1969 (FAP)—quickly encountered a host of problems, both conceptual and administrative. These continue to haunt negative tax advocates to this day.
The first and most basic problem is that it is currently fiscally—and perhaps administratively—impossible to construct an NIT that simultaneously
1. provides an income guarantee as generous as the cash and in-kind benefits already available to many welfare recipients in the United States,
2. provides an ostensible incentive to work (a far greater concern when benefits are to be extended beyond the traditional welfare population dominated by female-headed families), and
3. restricts coverage to any manageable proportion of the population—the so-called “break-even” problem.
These constraints are, in fact, irreconcilable as long as the median income remains within striking distance of the poverty line—a situation that has barely improved over the last two decades of slow average economic growth.
The McGovern plan proposed by the Democrats’ 1972 presidential candidate starkly illustrated this problem. With a guarantee determined by the candidate’s promise of $1,000 per person, and a benefit reduction rate limited to 33 1/3 percent at the behest of economic advisers worried about imposing work disincentives on a sizable proportion of the labor force, the plan had a break-even of $12,000 for a family of four—roughly the median income at the time. Thus, it would have converted roughly half the population into federal tax beneficiaries while the other half of the population would have paid for these transfers along with the cost of all other federal activities.
The second problem with a NIT is that the welfare system already provides a package of cash and in-kind benefits that, in many states, is worth considerably more than any likely NIT (though at the cost of excluding large groups of the poor—such as two-parent families—from eligibility). Political and humanitarian considerations prevent reducing these benefits, thus vitiating one of the NIT’s attractions—the possibility of abolishing the welfare system.
Competition from welfare was a severe problem for the New Jersey experimenters. Many of the families in the study were actually receiving welfare benefits worth more than the experimental payments. Therefore, some experts questioned the experimenters’ findings that the NIT had only a minimal effect on work incentives, and indeed questioned whether the experiment had really measured anything at all. HEW attempted to solve these problems by launching subsequent income-maintenance experiments in Seattle and Denver (SIME/DIME). These experiments more carefully integrated existing welfare programs and offered more generous NIT plans. But the generosity of most of the tested plans made them unlikely to be replicated on a national scale, and more complicated to analyze.
The Stanford Research Institute (SRI), which analyzed the SIME/DIME findings, found stronger work disincentive effects, ranging from an average 9 percent work reduction for husbands to an average 18 percent reduction for wives. This was not as scary as some NIT opponents had predicted. But it was large enough to suggest that as much as 50 to 60 percent of the transfers paid to two-parent families under a NIT might go to replace lost earnings. They also found an unexpected result: instead of promoting family stability (the presumed result of extending benefits to two-parent working families on an equal basis), the NITs seemed to increase family breakup.
The SRI researchers—Michael T. Hannah, Nancy B. Tuma, and Lyle P. Groeneveld—hypothesized that the availability of the income guarantee to some families reduced the pressure on the breadwinner to remain with the family, while the benefit-reduction rate also reduced the value to the family of keeping a wage earner in the unit. Other researchers, notably the University of Wisconsin’s Glen G. Cain, disputed the analytical strength of these findings. But at the very least the results were discouraging to those who promoted an NIT as a boon to family stability.
A third set of problems is administrative. One is the matter of the appropriate income-accounting period, which spawned a whole literature (to which I contributed my share). If income for NIT purposes is measured over a year, as in the positive tax system, families in great, but temporary, need may be denied benefits. If the accounting period is shorter, say a month, as in the welfare system, and income reporting procedures are lax, potential costs and caseloads might be as much as 70 percent higher than those predicted by the annual income-based models used to estimate the costs of FAP and its early successors.
The income accounting and reporting analysis—backed up by an HEW administrative experiment in cooperation with the Denver welfare department—also drew attention to the fact that many negative tax participants cheat on income reports. To be sure, many taxpayers cheat too. But the irregularity of income sources and shifting family arrangements at the lower end of the distribution make it unlikely that the IRS could prevent widespread fraud unless it converted itself into a facsimile of the much-detested local welfare offices.
Some indication of the difficulty that the IRS might experience in administering a large-scale NIT is provided by the Earned Income Credit (EIC), a hybrid version of an NIT that was slipped into the tax code in 1975 by Finance Committee Chairman Russell Long. Long saw it as a way to offset the regressive effect of payroll taxes on low-income earners. The EIC works like this (using 1993 rules for a family with one qualifying child): For every dollar of earned income, the family receives a refundable tax credit of 18.5 percent, up to a maximum credit of $1,434 (not including additional credits for young children and health insurance costs) for a family with earnings of $7,750. The credit is phased out at a rate of 13.21 cents for every dollar of adjusted gross income (AGI) above $12,200, until it reaches a zero value for a family with AGI of $23,050.
One attraction of the EIC is that because its benefits rise positively with earnings up to the phase-out point, it has a positive rather than negative effect on work incentives for workers earning a very low income. In the phase-out range, however, its substitution effects are identical to those of a comparable NIT. More important, a 1990 IRS study revealed that owing in part to the complexity of the EIC rules, almost 40 percent of EIC benefits were paid to families who were not eligible for them. Yet the finding of such a high error rate did not deter Congress from both enlarging and further complicating the EIC in the fall of 1990.
To these problems must be added the messy reality of the day-to-day stresses and strains of many of America’s poor. Data from another HEW experiment revealed that many of the low-income population’s problems are not readily addressed by the provision of an extra few hundred dollars in annual income. Indeed, recent revisionist social welfare thinking has questioned the central premise of the NIT planners—that a nonintrusive income-maintenance system is preferable. More recent policy has stressed direct interventions to improve family functioning and self-reliance, including eligibility rules that require work effort.
Nonetheless, the NIT is still popular among many researchers and politicians. Nixon’s FAP, which compromised with both the conceptual and administrative problems by essentially retaining the current welfare system and grafting on to it a low-level income-maintenance benefit for the working poor, pleased neither academics nor welfare advocates. Nor did it please the big-city mayors and governors, who looked at welfare reform as primarily a way to reduce their share of the system’s cost. (Sen. Daniel P. Moynihan’s Politics of a Guaranteed Income, is the definitive text on this point.) The Ford administration floated a reworked version of FAP, but it foundered on the same concerns as FAP.
President Jimmy Carter set out to develop a welfare reform that would have zero additional cost. Although Carter’s plan emphasized work alternatives to welfare, HEW’s analysts quickly shifted the emphasis to a more generous reworking of the FAP concept. Forced to compromise with analysts at the Department of Labor—myself, by then, included—who were promoting a low-wage guaranteed job approach, the reformers ended up with the unwieldy Better Jobs and Income Plan (BJIP), which would have cost many billions more and pleased neither liberals nor conservatives in Congress.
The Reagan administration put its efforts into cutting back, rather than extending, income benefits, though the eighties witnessed both an expansion of earned-income tax credits and the 1988 welfare act, which tried, at so-far modest cost and effect, to direct more welfare recipients into jobs. More recently, some “middle-income tax relief” proposals floated in Congress in 1991 would convert the current positive tax personal exemption into a “refundable” credit, paid out in cash to families with incomes below the tax-liability threshold. These proposals, thus, would incorporate a full-bodied NIT into the Internal Revenue Code. The appeal of a negative income tax lives on. And so do its many problems.
Jodie T. Allen is a senior writer with U.S. News and World Report and was previously editor of the Washington Post‘s “Sunday Outlook” section. Before joining the Post in 1980, she worked in several government departments and research institutions, analyzing costs and consequences of tax, transfer, and defense programs.
Allen, Jodie T. Designing Income Maintenance Systems: The Income Accounting Problem. 1973.
Browning, Edgar K. Redistribution and the Welfare System. 1975.
Friedman, Milton. Capitalism and Freedom. 1962.
Moynihan, Daniel P. Politics of a Guaranteed Income. 1973.
Munnell, Alicia H., ed. Lessons from the Income Maintenance Experiments. 1987.
Pechman, Joseph A., and P. Michael Timpane, eds. Work Incentives and Income Guarantees: The New Jersey Negative Income Tax Experiment. 1975.