FROM ISSUE NUMBER 21 ~ FALL 2014 GO TO TABLE OF CONTENTS

TEXT SIZE A A A

Overselling the Earned Income Tax Credit

LAWRENCE M. MEAD

Since the welfare reforms of the 1990s, the federal government and the states have conditioned much of the aid they provide to the poor on employment, or at least the active pursuit of employment. Linking welfare to work began as a conservative proposal that liberals opposed, but by the end of the Clinton era the idea had become reasonably bipartisan. And although it is now criticized more broadly and openly on the left than at any point in the last two decades, the idea remains essential to the way most Americans of all political stripes think about welfare.

Even among advocates of work requirements, however, confusion reigns regarding how public policy can best encourage lower-income Americans to seek work. Increasingly, many on the left contend that we need not enforce work through the heavy-handed means that characterized the '90s welfare reform — requiring welfare mothers to enter the work force and empowering case managers to make sure they did. Many liberals believe that it is enough to subsidize wages for those of meager means, and that the resulting incentive will motivate them to seek work on their own.

The chief evidence for this view is the supposed success of the Earned Income Tax Credit in lifting work levels among welfare mothers during welfare reform. The EITC subsidizes the earnings of low-income workers, thus effectively raising their wages. But its success in "making work pay" and combatting poverty among Americans who are employed has too easily been confused with success in causing more Americans to go to work. And this confusion is evident not only among opponents of strict work requirements. Even some on the right, like Senator Marco Rubio and Congressman Paul Ryan — who each have championed both work requirements in anti-poverty programs and an expansion of the EITC (or similar wage subsidy) in recent years — have at times confounded the potential benefits of the two. This confusion does not advance the cause of raising work levels in low-income America.

The EITC is an effective weapon in America's anti-poverty arsenal, but it is crucial to understand what it does and does not achieve. The credit rewards those who work and helps keep them working, but it does little to get them to go to work in the first place. To raise work levels, we still need the sorts of administrative work requirements seen in the last major welfare reform.

HELPING WORK PAY

The Earned Income Tax Credit emerged from the decades-long struggle to reform Aid to Families with Dependent Children, the controversial cash-assistance program for poor families. Conservatives initially invented the EITC, first enacted in 1975, as a way to condition some aid to poor families on employment, which liberal welfare-reform proposals of that era did not do. But over time, as it became clear that unconditional cash aid under AFDC was doomed, liberals embraced the EITC as a way to make work-based aid more generous. Some states even enacted their own wage subsidies on top of the EITC.

The credit was expanded in 1986 as part of that year's comprehensive tax reform, and then was expanded again in 1990 and 1993. In 1996, welfare reform replaced AFDC with a new program called Temporary Assistance for Needy Families, which conditioned cash support on far stronger work tests than before. Some states, though not all, enforced these rules with tough oversight by case managers. As traditional cash-welfare rolls plummeted, the EITC was left standing as the nation's largest cash subsidy for poor families.

The fall in welfare rolls was clearly correlated with rising work levels. The share of poor single mothers (with children under age 18) who worked jumped from 44% in 1993 to 64% in 1999, and the share working full year and full time almost doubled, from 9% to 17%. Analysts agree that the rise was due essentially to three factors: welfare reform proper (the tougher work tests for cash aid legislated as part of TANF), the good economic conditions of the late 1990s, and the EITC.

The EITC was popular with policymakers because it appeared to raise work levels through incentives and without TANF's punitive measures. The credit was reformed and somewhat expanded in 2001 (with changes intended to help mitigate its "marriage penalty") and was then expanded further for families with three or more children as part of the 2009 stimulus package.

The subsidy that the EITC now provides is therefore quite substantial. In 2013, workers with children and with family incomes up to $51,567 could receive a credit amounting to as much as 45% of their earnings, up to $6,044, depending on income and the number of children in the family. Under EITC's structure, as earnings rise, the subsidy builds to its maximum value, stays flat for an interval if earnings rise further, and then at still higher earnings is phased out until it disappears. Single workers without dependents receive a much smaller benefit. The credit is refundable, meaning that if workers have insufficient tax obligations to exhaust the credit, they receive the difference in cash. For many low-income workers, "tax day" in April is now a time when they receive income from Uncle Sam rather than having to pay him more.

Thanks to this generosity, the EITC has become a large program. Twenty-seven million families received the credit in 2012. That year, the EITC cost $63 billion, or twice the cost of TANF and associated state programs.

But in their enthusiasm for the credit, some policymakers have not been careful to distinguish the credit's role in increasing employment among the poor in the 1990s from the roles played by TANF and by the improved economy. Indeed, in recent years some on the left have downplayed the role of work requirements and have asserted that the EITC was the biggest driver of the work increase. This assertion implies that the main reason that relatively few poor mothers worked before the mid-1990s was not that welfare failed to require work, as welfare reform presumed, but that low-skilled wages were insufficient. The key constraint on entry into the work force for the poor, in this view, is the low-wage labor market. Give the poor a better deal and work levels will rise, without a need to enforce work. Much the same case is made for raising the minimum wage.

These claims are based on several statistical studies that credit the EITC with much, even most, of the rise in work levels among welfare mothers in the 1990s. The most noted of these studies, published in 2001 by Bruce Meyer of Northwestern University and Dan Rosenbaum of the University of North Carolina, Greensboro, found that the EITC explains as much as 62% of the increase in the employment rates of single mothers when comparing 1984 to 1996 and as much as 35% of the increase between 1992 and 1996. Another study, published in 2003 by Jeffrey Grogger of the University of California, Los Angeles, found that the EITC explains 34% of the rise in employment among single mothers from 1993 to 1999. Similarly positive findings come from studies by Nadda Eissa and Jeffrey Liebman (in 1996); David Ellwood (in 2000); Joseph Hotz, Charles Mullin, and John Karl Scholz (in 2006); and Michael Keane and Robert Moffitt (in 1998).

These studies assume that the EITC operates as a work incentive, moving people into work due to what economists call "the substitution effect" — raising the value of jobs compared to other activities, so that more people choose to work. Meyer and Rosenbaum think that the EITC was the "dominant" cause of work increases by single mothers between 1984 and 1996, and they question the need for "punitive measures, such as time limits and work requirements." Grogger remarks that "the EITC may be the single most important policy measure for explaining the decrease in welfare and the rise in work and earnings among female-headed families in recent years."

In earlier decades, most experts believed that the EITC actually reduced incentives to work, for reasons explained below. But in light of these and similar recent studies, a new consensus has formed that the credit promotes work while also subsidizing the poor, a rare combination not previously achieved in social policy. Liberal experts now commonly assert that, as Peter Edelman and Harry Holzer put it last year, the EITC has "demonstrated the potential of 'make work pay'...to raise employment levels while improving earnings and income among the poor." A Brookings Institution review of the EITC a few years ago found its effects generally mixed, but claimed a "consistent" finding that the credit raises employment among single mothers, again citing the same set of statistical studies. According to a 2013 study by the Center for Budget and Policy Priorities, a liberal advocacy group, the EITC expansions were "the most important reason why employment rose among single mothers with children during the 1990s." Specifically, it was "more effective in encouraging work than either welfare reform or the strong economy."

But this new consensus misses crucial parts of the story. The theory behind work incentives, the nation's actual experience with them, long-established patterns in the psychology of poverty, and on-the-ground testimony all suggest that the EITC in fact played a minor role in welfare reform and had only modest effects on decisions by welfare recipients to seek work. Economic studies fail to perceive this because they lack any such hands-on reference. Field researchers say that politics, not economics, largely drove the welfare revolution. Administrative work enforcement was the main reason work levels rose, and similar methods must be the main basis of policies aimed at raising work levels among the poor (especially poor men), as many would now like to do.

INCOME AND SUBSTITUTION

It might seem obvious that subsidizing wages, as the EITC does, would tend to raise work levels, since the credit means that those not working will earn more by taking a job than they could before. But as economists have long known, the subsidy also raises the wages of those already working. They can now cover their needs with fewer working hours than before and may cut back work effort. This incentive, called the income effect, offsets the substitution effect of the higher wage. As a way of promoting employment, work incentives could thus turn out to be self-defeating. It is unclear a priori whether instituting a wage subsidy will raise or lower work levels overall.

Some think that the EITC escapes this problem because the credit requires employment: It subsidizes only low earnings, not low income in general, so only the substitution effect should operate. But families receiving the EITC could nonetheless feel an income effect at the top end of the credit's income-eligibility range. That is, among the employed there could be a temptation to work less so as to remain eligible for the credit. Since over three-quarters of EITC earners have incomes in the upper ranges of the credit, this income effect could be important. EITC does reduce work effort by wives in two-parent families, apparently due to income effects, as Nada Eissa and Hillary Hoynes have shown. So as with work incentives in general, it is unclear a priori whether EITC raises work levels overall.

Indeed, work incentives have been added to income-maintenance programs and experiments since the 1960s, and — until the relatively recent EITC studies — no one ever claimed that they much affected employment. In welfare, work incentives consist of reducing benefits by less than the full amount of earnings if a recipient takes a job. In AFDC, such incentives varied widely among states and over time, yet they had next to no effect on the work levels of mothers on welfare, which remained low throughout the life of the program. Under TANF, work incentives offered by states also have had no palpable effect on employment, even though on average these incentives are much stronger than those under AFDC, as Jordan Matsudaira and Rebecca Blank showed in a paper earlier this year.

One particular set of studies demonstrates the danger of the income effect. Between 1968 and 1982, the federal government conducted four experiments to see how work levels would change if low-income families were guaranteed an income coupled with work incentives of varying strengths. The results showed clear income effects, in that work levels varied inversely with the grant level. But the work incentives showed less effect than statistical studies had anticipated. That is, varying the benefit-reduction rate in response to earnings and thus the effective wage had little impact on whether recipients worked.

Some say, again, that the problem here was that neither AFDC nor the income-maintenance experiments strictly required work in order to qualify for a benefit. So work incentives should be more effective in the case of the EITC, since only workers are subsidized. But in recent decades, several experiments have tried to motivate welfare mothers to work by linking various benefits to employment, and the effects have been uneven at best. These projects included the New York State Child Assistance Program, the New Hope Project, the Minnesota Family Investment Program, Canada's Self-Sufficiency Project, JOBS-Plus, and the Opportunity NYC – Family Rewards Demonstration. All of these efforts conditioned the benefits they offered in some way on employment. And some had sizable effects — MFIP and the Self-Sufficiency Project raised the employment of welfare mothers by a third.

But a pure substitution effect would operate by non-workers finding out about work subsidies, calculating that work now pays more, and entering jobs on their own. All the recent projects relied on a lot more than that to raise work levels. Except for CAP and JOBS-Plus, each of the programs required that eligible people work at least 30 hours a week to receive the benefits. Each program also used staff or other activities alongside the incentive to persuade eligible individuals to put in those hours and helped them to do so. MFIP even enforced participation for long-term recipients, on pain of cuts in their welfare payments. Some projects — CAP, New Hope, and Opportunity NYC — recorded no or small work increases even with a 30-hour rule and a staff structure. In no case did a work-incentive experiment achieve what the recent studies claim regarding the EITC: a large work increase based on the substitution effect alone.

Doubts about the effects of the EITC are also raised by the fact that many of the eligible individuals in these experiments declined the benefits offered. In CAP, only a sixth of mothers eligible for the program took up its offer to earn higher incomes through working. In SSP, only 36% of eligible people participated at any time, and only 25% or less in any given month. Most notably, New Hope offered a generous package of jobs paying at least poverty-level wages, plus child and health care, in return for working 30 hours a week. And it did so in two Milwaukee neighborhoods where 12,000 to 30,000 adults were eligible. Yet it took the program 17 months to enroll only 1,362 applicants. This appears to contradict the premise underlying work incentives, which is that non-working adults actively seek to work but are barred from doing so by adverse incentives.

Another serious doubt about the studies concerns the causal direction of the effects they imply. The assumption behind the idea of an EITC effect is that, as the credit was expanded in 1986, 1990, and 1993, non-working poor mothers heard about it and went to work in higher numbers than before, producing the noted increases in employment levels. But a precondition of getting the EITC is that one is working. So at either the individual or aggregate level, one would expect work levels to precede and predict rising EITC claims rather than the other way around. Of course, one might hear of the EITC and then go to work and claim the credit, as the substitution effect supposes. But in terms of what is measurable, work levels should still determine EITC levels rather than vice versa.

The statistical studies would also lead us to assume that claims for the EITC grew as more mothers went to work. But that did not occur. According to research by Joseph Hotz and John Karl Scholz, the number of EITC claimants reached 19 million in 1994 (from under 12 million just five years earlier) and then did not grow much further during the 1990s, even though labor-force participation by single mothers soared. The expansions of the credit in 1990 and 1993 apparently had little effect on EITC participation. How then could the credit explain rising work levels? If newly employed low-income women did not claim the credit in large numbers, how can the promise of the credit be what drew them into the work force?

The effects attributed to the EITC by these studies are by far the largest ever claimed for a work incentive. But they are a poor fit not only with EITC claim levels in the '90s but also (as we shall see below) with the experience of those involved in welfare reform. One explanation might be that the studies are all statistical, and therefore less able to separate out the EITC from other influences on work levels than were the policy experiments noted above. Since the EITC's benefits and substitution effects, as described by these studies, strengthened in the mid-1990s just when welfare reform and a good economy were also promoting higher work levels, the EITC effect could easily be overestimated. Also, the effect of welfare reform on work levels is tougher to measure than that of the EITC or economic conditions.

All the studies have to gauge reform pressures indirectly, for instance by looking at the "waivers" that states obtained from Washington to run various work experiments prior to welfare reform. Due to these measurement problems, the studies could easily overestimate the effect of the EITC on work levels while underestimating that of welfare reform. The question of whether the EITC drove welfare recipients to work in the '90s is ultimately a question about human motivation, and on this matter economic analysis alone cannot be trusted to provide a complete answer. Fortunately, these analyses are not our only source of insight regarding the psychology involved.

MOTIVATIONS FOR WORK

The core assumption underlying the claims of the new EITC consensus is that non-working welfare recipients learned of the work incentive offered by the credit and then responded to it. That is, they acted rationally so as to optimize their incomes, as economics assumes most people do. But direct observations of the long-term poor raise serious questions about whether this claim is plausible.

The first question is whether eligible people even knew about the EITC. The studies noted above claim that, by 1993, a work increase driven by the EITC was well underway. Yet in 1993 and 1994, interviewers of low-income populations in Cambridge, Massachusetts, and Chicago found "virtually no awareness of the credit," according to Nada Eissa and Jeffrey Liebman. Later, awareness of the EITC grew somewhat, but that was partly because many low-income workers heard about it from the tax preparers they consulted to calculate their tax returns once they were working, according to several contemporaneous studies. Few apparently responded to the credit purely on their own, as the substitution effect presumes.

Of course, for the substitution effect to operate, recipients need not understand EITC in detail. It is enough to know that going to work will somehow generate added income through the tax system. It is difficult for beneficiaries to perceive this, however: Though the program permits employers to distribute the benefit to employees on an advance basis (which would allow recipients to see rewards regularly), very few employers actually do this. Instead, the vast majority of recipients of the EITC see the benefit only once a year, when they file their tax return. It defies common sense to believe that a population that showed so little response in incentive experiments in which benefits were paid monthly could be strongly moved to work by a credit to be paid only months in the future.

Standard economic psychology also assumes that the poor are highly sensitive to economic payoffs, but that too is in doubt. Many welfare mothers fail to work despite an apparent need for income, but others do work despite punishing losses of welfare and other benefits that appear to discourage work. Few, whether working or not, can be shown to understand the incentives that welfare, the EITC, or other programs generate for them.

Further, work incentives assume that non-working adults want to work and will do so when the payoffs improve. First-hand accounts of low-income life, however, suggest that regular employment is usually not an immediate priority. Poor non-workers typically want to work, yet their lives are not organized around employment, as is true for the better-off. Single mothers are absorbed with family problems, while many men make money off the books without holding steady, legal jobs. Everyone struggles with immediate difficulties, mostly in private life, rather than working regularly to earn income.

The economic language that surrounds work incentives assumes that non-workers reason about jobs the way trade unionists do. If they do not work, the reason must be that jobs are lacking or are unacceptable due to low pay or other conditions. But in ethnographic accounts of life in poverty, hardly anyone talks like this. Poor adults typically treat employment as a moral issue, not an economic one. They assume jobs exist on acceptable terms, and they reproach themselves and others for not taking and holding them. If economics does not govern work behavior, there is little chance that "making work pay" will move non-workers into the labor market, as recent claims for the EITC suggest.

During my own research on welfare reform in Wisconsin, I interviewed 60 welfare officials at the county and state levels. This was in 1995, when the state's welfare caseload was plummeting, a fall closely linked to rising work levels. I asked respondents to explain this decline. They mentioned a wide range of factors, especially welfare reform, good economic conditions, and child-support enforcement. Not a single one cited the EITC or the state's own wage incentive.

In 2005, I asked several more senior welfare administrators in Wisconsin and New York whether they had seen or heard of an EITC effect on recipients going to work. The officials included Jean Rogers, who had headed welfare reform in Wisconsin; Jason Turner, one of her leading deputies in Wisconsin who later led welfare reform in New York City; Mark Hoover, who was a senior aide to both Rogers and Turner; and Swati Desai, longtime chief of welfare policy and evaluation in New York City. The city, like Wisconsin, underwent a radical reform, causing an extreme decline in caseloads. And New York state, like Wisconsin, provided its own low-income wage subsidies on top of the EITC. If the credit had any effect on going to work, it should have shown it here.

But Rogers reported that "[t]here was no indication that EITC was any sort of motivating factor in people's decision to go to work." Recipients became interested in the credit only after they had been "working for a while" and saw what they could get from it. Turner said that front-line supervisors he had talked to in work programs emphasized a duty to work rather than the rewards. But after recipients left welfare, EITC might operate as "an incentive for keeping people in the labor force, as opposed to a motivating factor getting them there." Mark Hoover told me that he had tried and failed to get more welfare recipients to apply for the EITC in several states. In none of these states had he ever "had anyone claim EITC had an effect in the initial decision to go to work." But after getting off welfare, the credit might help some recipients stay off.

Swati Desai said that the EITC was "one of the factors that influence[d] welfare recipients' decision to go to work" because the contractors who run the welfare work programs in New York City used the credit to help persuade recipients that they would be better off working than remaining on welfare. So again, just as with staff in the work-incentive experiments or paid tax preparers, administrators informed beneficiaries about the credit — they seldom learn of it and respond on their own.

Staff use of the EITC as an inducement to work, moreover, was probably stronger in New York City than elsewhere. Due to the city's high TANF payments and other benefits, persuading recipients to leave aid for work might be more difficult there than in less generous localities. Elsewhere, though, according to field research by Irene Lurie, "financial incentives to work received little attention in the welfare and workforce offices." Staff concentrated on telling recipients that they had to work as a condition of TANF, leaving them to find out about the EITC on their own.

Thus, on-the-ground research on welfare reform offers essentially no reason to think that the EITC deserves much credit for driving recipients off of AFDC or TANF and into jobs.

WHERE CREDIT IS DUE

The EITC is certainly an important component of America's anti-poverty efforts. By stabilizing employment and forestalling returns to welfare, the credit appears to bolster work levels slightly, if to far lesser degree than claimed by some of the statistical studies noted above.

The real value of the credit comes in integrating the poor after they enter the work force. Employment gives them a legitimacy in the eyes of the wider society that they never had on welfare, while the combination of employment and the EITC gives them higher incomes and some ability to plan for the future. The poor — at least if they work — are brought closer to mainstream America.

Jason DeParle was the leading journalist covering welfare reform in the years surrounding its passage. He closely observed the welfare revolution in Wisconsin. In 2006, I asked him the same question I put to the senior welfare officials: Had he seen or heard of the EITC's affecting recipients' decisions to go to work? He replied:

I know that David Ellwood and others have established social science correlations between EITC and work effort, and I have [no] reason to doubt them. But no, I haven't heard people on welfare say the credit pushed them to go to work. My sense is that the hassle factor of welfare is much more powerful in pushing them off the rolls (and consequently into jobs) than the vaguer promise of later wage subsidies.

On the other hand, that's not to say that they don't think about the EITC. They are hugely aware of it, especially after they start working. It's a big part of their survival strategy. It no-doubt reinforces the desirability of work (or, phrased the opposite way — it plays a huge role in blunting the harshness of the low-wage economy). In enabling people to buy cars, keep up with the rent, etc., it may even make work possible. Without it, many fewer people might be willing to stay in the workforce....

So that's why I say I accept the point that it raises work rates.... [M]y impression is that it's not the thing that initially gets welfare recipients working. The first is the hassle of the welfare system. My sense is the EITC then plays an important secondary role, in stabilizing them economically and rewarding them psychologically.

Did the EITC have more effect on raising work-force participation levels among the wider low-income population than it did among welfare mothers? That seems unlikely as well. The decline in welfare rolls mostly reflected diversion — poor mothers who went to work in larger numbers without going on welfare at all. Was diversion driven by EITC?

In 2008, the American Enterprise Institute held a conference on diversion, entitled "Why Did Welfare Caseloads Collapse? The Mystery of Diversion." Attendees included most of the leading field researchers who had produced the key studies of welfare reform at the state and local levels, and yet the EITC was mentioned only in passing. In the view of many of these researchers, diversion had had little to do with incentives of any kind. Rather, states told poor mothers that they were supposed to go to work in available jobs and avoid cash welfare except as a last resort. That message emanated not only from the welfare system but from political leaders, other opinion makers, and even from the business community — which, in the strong economy of the late 1990s, hired most of the mothers leaving the welfare rolls. This message was so pervasive and effective that welfare rolls fell drastically in every state, even those that did not in fact change welfare very much. Mothers might also claim remaining benefits such as EITC and food stamps, but this was after they had gone to work.

Much of the effect the statistical studies find may be due to the EITC keeping people working rather than causing them to go to work. The studies appear to err, in part, because they take an exclusively statistical approach. As is common in economics, these studies are based entirely on large databases. There is no hands-on inquiry to clarify the behaviors underlying the numbers. It is simply assumed that individuals are acting to optimize their incomes, and using all available resources to do so. In this case, these assumptions lead to unrealistic conclusions.

To economists the important thing is the correlation they find between the EITC and work. Whether the credit operates through promoting work or retention in work, it is raising employment in some way. That is true, as far as it goes. But if such studies are to inform social policy, how the credit operates matters and cannot be ignored. If the EITC causes many poor adults to go to work on their own, as the recent studies assert, then other steps to promote work are superfluous and the idea that the poor just need better incentives and opportunities to work is confirmed. If, on the other hand, the credit only bolsters income after employment, then going to work must still be encouraged and enforced by administrative means.

This difference has even broader implications for how we think about poverty. If one assumes that the response to work incentives reflects straightforward income-maximizing behavior, one can downplay the differences between the poor and the middle class. But if, as the observational accounts suggest, the poor are far less responsive to incentives or far less focused on work, then, again, work must be enforced.

THE NEXT WELFARE REFORM

The Earned Income Tax Credit clearly does reduce poverty, but it raises work levels far less than some of the statistical studies of the past decade claim, and it appears to do so by encouraging working people to keep working, rather than driving the non-working poor toward jobs. If we wish the credit to promote work as well as raise incomes, we cannot count on the substitution effect alone. We must add other suasions to promote and enforce work, such as those found in the more successful work-incentive experiments reviewed earlier. These include mandating participation in work programs and setting some threshold of working hours that claimants have to achieve to get benefits.

If we want the EITC itself to play this role, there are ways we might reform it. The credit as currently structured has no hour threshold, for instance, and that could be changed. Keeping track of hours might require shifting the administration of the credit from the IRS to the welfare or employment-insurance systems, which are better set up to monitor how much claimants are working. Another way to help the credit promote work would be to support state-level staff to reach out to eligible people and persuade them to put in the working hours needed to claim the benefit. Such an approach would take account of the role that administrators played in the more successful experiments of recent decades. Again, such an approach might require moving the EITC into the welfare or employment systems, though doing this would link the program more closely to "welfare," which is unpopular with recipients and taxpayers alike.

The recent claims made on behalf of the EITC have stoked false hopes that work incentives alone might bring low-income men into the work force in large numbers. Several scholars think that expanded wage subsidies for these men would raise their work levels as well as incomes. One proposal is to increase the small subsidy that EITC now gives to single men without child-care responsibilities.

But poor men, like poor women, show only a weak response to work incentives alone, as research on the New Hope Project by Greg Duncan, Aletha Huston, and Thomas Eisner has shown. To help move poor men into work, reformers would have to take steps like establishing a 30-hour threshold for benefits and creating a staff structure to reach out to nonworking men, as was done in the New Hope experiment. And they should enforce work for men who are already supposed to be working — those on parole from prison or in arrears on child-support payments. Such a regime would amount to welfare reform for men, and it would be a very good idea. But it cannot consist of an expansion of the existing EITC alone.

The idea that "making work pay" can by itself raise work levels significantly has obvious appeal. It suggests that, if society just gives poor adults a better deal, they will choose to work on their own, without the unpleasantness of work enforcement. Experience shows, however, that this kind of incentive is very unlikely to suffice. Welfare reform succeeded in large part by establishing clear requirements and engaging directly and intensely with those subject to them, and the same would likely be true in any effort to raise men's work levels.

The vast majority of Americans surely share the sentiment once voiced by President Bill Clinton: "If you work, you shouldn't be poor." But satisfying the "if" is the hard part, and work incentives alone will not accomplish that.

Lawrence M. Mead is a professor of politics and public policy at New York University and an adjunct scholar at the American Enterprise Institute.