With the passage of the welfare reform law of 1996, 60 years of federal welfare policy was abruptly reversed. The Temporary Assistance for Needy Families program (TANF) ended the legal entitlement to benefits, mandated that a large percentage of recipients work, and imposed a five-year time limit on the receipt of federally funded benefits. Since then, welfare caseloads have fallen sharply, and the percentage of single mothers working has risen dramatically. And despite the concerns of many--on the left and the right--there has not been a substantial increase in material hardship. In fact, for most single mothers, incomes have risen.

With this track record, one would think that TANF's reauthorization, due by September 2002, would have been uncontroversial. But it did not work out that way. Faced with fierce partisan disagreements, Congress adjourned without reauthorizing the law. Instead, it passed a short-term extension, so that the reauthorization battle will continue at least into 2003.

The stalemate in Congress was triggered by sharp differences over whether to expand work requirements or to increase aid to low-income families not on welfare in the form of child care and other benefits. These two issues have been the main poles of the welfare reform debate for decades and promise to be the continuing source of contention even after TANF is eventually reauthorized. Before turning to these matters, however, let us begin by assessing welfare reform's achievements and failures over the last several years.

A Short History

The Aid to Families with Dependent Children program (AFDC) was created in 1935, as part of the Social Security Act of that year. AFDC was aimed at families where the father was unavailable to provide support due to his death, disappearance, or disability. For over half a century, it seemed that caseloads could only grow. Through the 1960s and early 1970s, the focus of progressive social policy was to expand eligibility and to ensure that eligible families received benefits. Between 1960 and 1977, the size of the program more than quadrupled, from about 800,000 families to over 3.5 million families (see figure 1)

Even those opposed to expanding welfare coverage should appreciate that much of the increase came with the demise of Jim Crow laws and practices that wrongly denied government benefits to African-American and, to a lesser extent, Mexican-American families. In the summer of 1968, as a civil-rights worker in Mississippi, I saw black families given bus tickets to the North instead of welfare assistance, and black children denied admission to public hospitals though they were suffering from malnutrition and other serious illnesses.

In any event, by 1994, welfare caseloads had reached a historic high of 5.1 million families, representing about 15 percent of American families with children. But that year caseloads began a seven-year decline that eventually reduced the national rolls by about 60 percent. Caseloads dropped almost everywhere--in suburbs, rural areas, and even most inner cities--with substantial declines for whites, blacks, and Hispanics alike. The number of white families on welfare showed the steepest decline, falling by 63 percent, while the number of black families on welfare fell by 52 percent, and the number of Hispanic families by 44 percent, with immigration presumably countering what would have been a larger decline. No one predicted a decline of this size and scope, not even the strongest proponents of the original TANF legislation.

Around July of 2001, caseloads stopped declining in most states and started rising again, most likely because of the weakening economy. But although about 10 states experienced caseload increases of 10 to 20 percent, as of June 2002, the nationwide rise has been surprisingly modest.

Was It Welfare Reform?

Why the astounding decline in welfare caseloads? It suited the purposes of Bill Clinton and the Republican Congress, as well as most governors, to attribute the decline to welfare reform. That way, they could claim credit for it. But although welfare has certainly changed in many important ways, it has not changed nearly enough to account for even half of that 60 percent caseload decline.

One of the most significant achievements of welfare reform was to change the culture of welfare offices. Case workers went from simply giving checks to poor families to encouraging applicants and recipients to find jobs or other sources of support. In most welfare offices, this new emphasis is called "work first," and usually includes a mandatory "job search" by applicants as well as recipients, often combined with job clubs, classroom instruction on job-seeking skills, access to phone banks, and peer support. (These requirements are often backed up by benefit cuts against those not "playing by the rules," but relatively few families are actually thrown off welfare.) One-time cash payments to help individuals get or keep a job (for instance, money for car repair) are also available. A sign stenciled on a New York City Job Center summed up the pro-work atmosphere in most offices: "Be prepared to work, or be prepared to leave."

Many welfare agencies also offer work-preparation programs that provide short-term education, training, or work experience. However, participation in such programs is limited, and not much different than in the past. In an average month in 2000, only about 5 percent of TANF families included an adult receiving education or training.

These two programmatic elements of welfare reform--a "work first" policy and limited education and training--are unlikely to have caused the large caseload decline. Before TANF, even the most richly funded welfare-to-work experiments rarely achieved reductions of even 10 percent. (Such experiments cannot measure "entry effects," that is, the number of mothers deterred from applying for benefits because of program rules and procedures. But there is no reason to believe that such "entry effects" would have accounted for the remainder of the decline.)

Mandatory work, often called "workfare," also cannot explain the decline. Despite popular impression, few states operate large-scale mandatory work programs in which recipients perform public service activities in return for their welfare checks. In an average month in 2000, only about 3 percent of TANF families included an adult in a mandatory work program -- and most of these were in either New York City, Wisconsin, Ohio, or New Jersey.

The decline also cannot be explained by the increase in welfare recipients who are combining work and welfare. Under AFDC, the federal government established what is called an "earnings disregard," that is, a set amount of earnings that is not taken into consideration in determining the welfare grant. For all practical purposes, the AFDC earnings disregard was limited to the first $90 in earnings, after which benefits were reduced dollar-for-dollar as earnings increased--a 100 percent marginal tax rate.

Under TANF, however, individual states set the amount of their earnings disregards, and most greatly liberalized them. The typical earnings disregard is now on the order of 50 percent of earnings, and a few states even disregard all earnings up to the poverty line for a specified period of time. As a result, the proportion of recipient families in which an adult is combining work and welfare has more than doubled, up from about 8 percent in 1994 to about 16 percent in 2000.

Some people assume that this increase in those combining work and welfare was an important factor in reducing caseloads--that recipients became more employable after being introduced to the world of work and building their confidence and job skills. However, careful studies of the impact of "income supplement programs"--that is, programs that allow recipients to earn money while staying on welfare--usually show an increase in caseloads. Even when such programs have been combined with time limits or work requirements, they have rarely resulted in a caseload reduction of more than 5 percentage points.

In the face of these realities, most experts conclude that welfare reform reduced caseloads because it changed the expectations and thus the "atmospherics" of the program--that current and potential recipients had responded to the new, pro-work message. That is the most plausible explanation available, but, even so, it is unlikely that an emphasis on work alone has reduced national caseloads by 60 percent--especially since some of the jurisdictions with very large declines made relatively few policy or programmatic changes.

The Economy of Work

Two other factors--a strong economy and sharply increased aid to low-income, working families--almost certainly contributed as much (if not more) to the caseload decline as welfare reform. Academic and government researchers have used national data sets to estimate the relative impacts of these factors. Looking across all the studies, and discounting the weakest ones, the most reasonable conclusion is that, although welfare reform was an important factor in reducing caseloads (accounting for 25 to 35 percent of the decline), the economy was probably more important (35 to 45 percent of the decline), and expanded aid to low-income, working families (primarily through the Earned Income Tax Credit, or EITC) was almost as important as welfare reform (20 to 30 percent).

Defenders of welfare reform argue that such studies are highly variable and depend on various subjective assumptions (which is true), and that historically, strong economies have not led to reductions in welfare rolls. This latter point, however, is not quite true.

First, during earlier economic expansions, divorces and nonmarital births were both rising sharply, while at the same time barriers preventing single mothers (especially African-American mothers) from going on welfare were falling. For example, until the 1960s, many states had "suitable home" or "unfit mother" rules that excluded unwed mothers (and sometimes divorced mothers) from welfare. Since these mothers were disproportionately African American, the three-decade liberalization of these and other rules allowed hundreds of thousands of black families to receive the benefits to which they were legally entitled.

Second, according to some measures, welfare recipiency actually did decline during the economic expansion of the 1980s, just not by a lot. For example, between 1982 and 1990 (the beginning and end points of that decade's economic expansion), the percentage of single mothers on welfare fell from 56 percent to 52 percent. (Population increases raised the total caseload, thus obscuring declines in the rate of welfare recipiency.) The strong economy of the 1980s did not draw more single mothers off welfare and into the workforce because employers first hired married mothers, who tended to have more job skills. Over two million additional married mothers started working full time during the 1980s expansion, raising the percent of married, working mothers from about 34 percent to about 44 percent.

Third, and most important, the 1990s economy was much more favorable to low-skilled, single mothers than was the 1980s economy. In the 1990s, employers essentially ran out of married mothers to hire (even though an additional 1.3 million married mothers started working full time), and so employers then turned to single mothers. Between 1994 and 2001, the increases were striking: 31 percent more divorced and separated mothers and 55 percent more never-married mothers were working full time.

The increase in work among divorced and separated mothers, and especially among never-married mothers, had a tremendous impact on welfare caseloads. In 1994, these two groups of single mothers made up 82 percent of the national AFDC caseload. (Moreover, never-married mothers are the mothers most likely to be long-term welfare recipients, because they tend to be the youngest and least educated of the three groups.)

Based on wage data for entry-level or low-skilled workers, it appears that the strong economy drew these mothers into work, rather than welfare reform pushing them out. For example, even as over a million more single mothers were going to work, the average hourly wages for those single mothers in the lowest wage quartile were rising, suggesting unmet demand in the low-skills job market. Between the fourth quarter of 1996 and the second quarter of 2001, their hourly wages climbed 17 percent, from $6.57 to $7.71 (in 2001 dollars). These wage gains came on top of a 213 percent increase in the EITC between 1990 and 1996, from $1,276 to $3,997 (in 2001 dollars). Quite a double-barreled incentive to get a job.

Life after Welfare

How did these three factors--"work first," a strong economy, and increased aid to low-income families--combine to reduce caseloads by 60 percent? Many experts use a "push/pull" metaphor to describe the dynamic: Welfare reform pushed single mothers off assistance while the economy pulled them into jobs. That is an incomplete explanation, however, because it does not account for alternative sources of support available from other government programs, as well as from relatives and boyfriends, that enabled many mothers to leave welfare without working.

Contrary to popular impression, only about 40 to 50 percent of mothers who have left welfare are working regularly, that is, have steady, full-time employment. According to surveys in various states, these mothers are earning between $6 to $8 an hour. Assuming a national average of $7 per hour, that's about $14,000 a year for full-time, full-year employment.

An income of $14,000 is not a lot of money, of course, especially for a woman with two children (about the average number of children for those leaving welfare). But various government programs help make it pay for her to work. In 2001, a single mother with two children earning this salary would be entitled to about $3,800 from the EITC, up to $2,700 in food stamps, and about $1,200 in subsidized school meals for both children. After taxes, that comes to a total of almost $21,000. If she also receives subsidized housing, add another $3,000 or more. In addition, her children would remain eligible for Medicaid (an average of $2,428). Moreover, although by working she would have added costs, her biggest cost--child care--would generally be covered, and most states would also provide transportation assistance.

This combination of earnings and government benefits is not enough for a life of middle-class comfort, but it is much better than being on welfare. And this is probably why the majority of those surveyed invariably say that they are better off after having left welfare. Moreover, most studies show that their earnings rise the longer they stay off welfare, presumably because they are gaining job skills and developing a positive work history.

What about the other 50 to 60 percent of former welfare mothers who are not working regularly? About 10 to 15 percent appear to be working part time, while about 20 to 35 percent are not working at all. Very few families in either group end up moving into shelters or becoming homeless. (Families with children are unlikely to live on the streets and, when they do, it is for a very short time before the authorities find them and place them in a shelter.) Instead, mothers who work irregularly are assisted through other government programs (mainly food stamps, housing, and Supplemental Security Income, or SSI) or by friends and relatives. Often these women live with others who support them, although this is one of the least acknowledged or understood aspects of low-income life. While the data are incomplete, it appears that, on average, these mothers lived in households that saw a modest rise in income (probably because others in the household were earning more thanks to the strong economy).

The remaining mothers who have left welfare, ranging between 15 to 30 percent, cycle on and off welfare as they alternately find a job or someone to support them, and then lose the job or support. These mothers rightly concern us the most because they often have no one to turn to for help when they suffer setbacks. Although there are only limited data about this group, it appears that they have substantially lower total household incomes than those who leave welfare and stay off -- as well as lower incomes (including other benefits) than they had before they left welfare. In a Milwaukee study, the difference was $3,000 per year less: $7,450 versus $10,460 per year. Although it is probably not correct to characterize their lives as desperate, grim would not be an exaggeration.

Why did these divergent groups leave welfare? Very roughly, for those who left for work, it appears that the strong economy combined with increased nonwelfare aid made work possible, while "work first" policies encouraged them in that direction. Those who left without employment (or at least without full-time employment) probably found the added hassle of the new welfare policies not worth the trouble because they had other sources of support.

Conservatives and Liberals

TANF was passed only after an acrimonious debate in Congress about the nature of welfare dependency and the likely impact of ending the welfare entitlement and emphasizing work instead. Taking a broad view of what happened, one would have to say that both conservatives and liberals were proven right--and wrong.

Conservatives were proven right that the welfare system itself encouraged dependence and that many welfare mothers could instead hold down jobs or rely on others for assistance. A pro-work welfare system of limited, short-term assistance successfully pushed many mothers off welfare.

Furthermore, welfare reform did not produce a social catastrophe, even in the inner cities. Despite intensive investigation by liberal policy analysts and journalists, no substantial increase in hardship has been documented. The one worrisome exception is the group of mothers who seem to be cycling on and off welfare. They appear to be having a difficult time under welfare reform, and their numbers could easily increase if the economy continues to weaken. Already, this group seems to have grown significantly. Caseload churning, an indirect indicator, has been rising rapidly. In 2001, for example, the national TANF caseload almost completely turned over, meaning that the number of families that entered welfare and left was almost equal to the total number of recipients. This high rate of churning--about 50 percent higher than the historic rate--signals great instability in the caseload. If exit rates decline or entry rates rise even a little, caseloads could rise quickly.

Many of the arguments of liberals have also been vindicated, especially regarding the central importance of a strong economy. The high demand for entry-level workers has been a major cause of the decline in caseloads--the only question being whether its impact was greater than that of welfare reform.

Liberals were also right about the difficulty that most mothers leaving welfare would have in becoming financially self-sufficient. With average wages of $6 to $8 an hour, even those working full time have relatively modest incomes. Only the expanded aid now available to low-income, working families (especially child care and the EITC) makes it worthwhile for them to leave welfare. And, of course, some of these women, especially those cycling on and off welfare, seem to be doing worse than before.

More fundamentally, the huge drop in welfare dependency and the rise in work does not seem to have substantially changed the lives of these families. Almost all the social indicators that started heading in the right direction in the early and mid 1990s seem to have stopped improving, and some have started to retreat. Since these improvements began before welfare reform, it is difficult to attribute them to reform. For example, although some commentators have pointed to a declining rate of nonmarital births as one of the most striking results of welfare reform, the picture is actually quite mixed. Nonmarital births started declining in 1994, about the time that the first welfare-reform waivers took effect and at least two years before TANF's implementation -- and they declined only through 1997, after which they started climbing again. Similarly, there has been no detectable increase in marriage rates, although since 1994 cohabitation among couples with children seems to be up by at least a third.

Of course, it may be too early to see the impact of reduced welfare dependency on conditions that took root over many decades. Working at a regular job may change a mother's outlook on life, but earning enough to make ends meet would also surely help--and that may take time as she accrues work experience. A successfully employed mother may be an important role model for her children, but it may take a decade or more before a change in their behavior is evident.

Thus if caseload decline is the measure, then welfare reform, in contributing to this decline, can be counted a success. But if self-sufficiency and stronger families are the test, then, at best, the jury is still out. Indeed, by all appearances, much more needs to be done to reach these fundamental goals.

Mandating Work?

Mandatory work, or "workfare," has long been controversial. Supporters see workfare as a way to build the job skills of recipients who have not had much work experience, while fulfilling the reciprocal obligation between recipient and government. It does not hurt that workfare "smokes out" recipients who are working (under the table) unbeknownst to the agency. Workfare also encourages recipients who can find a job to do so, since their choice is either workfare or a "real job," as they put it. And even if they do not succeed in getting jobs, purposeful activity rather than inactivity may be better for most welfare mothers.

Until TANF, opponents of mandatory work were able to water down the work requirements that since the 1960s have regularly been proposed in Congress. In the 1970s, they succeeded in labeling workfare as "slavefare." It was not just liberals that stymied workfare programs. State leaders, even very conservative ones, also shied away from workfare programs because of the costs and administrative challenges involved and the political controversy they generated. Invariably, unions opposed placements where real work is performed--fearing that they would displace existing state or municipal union employees.

TANF was supposed to have finally settled the issue, by imposing seemingly tough "participation standards" on recipients. By 2002, 50 percent of all welfare families that included an adult were required to be in work activities for at least 30 hours a week (20 hours per week for a single parent with a child under age six). The requirements were even stiffer for two-parent families, with 90 percent of parents required to participate for at least 35 hours a week (or 55 hours per week if the family is receiving a child-care subsidy). But as mentioned above, in 2000 only about 3 percent of the national caseload was in fact in mandatory work programs, and only 16 percent was participating in any work-related activity.

What happened is that because of the massive caseload decline, various TANF provisions excused states from having to establish mandatory work programs. A state's caseload decline is credited toward its TANF participation requirements, thereby lowering them. Since combining work and welfare also counts toward participation requirements, the large number of people in this category satisfied whatever was left of the TANF requirement regarding single mothers on welfare. (In many states, however, these factors were not sufficient to meet the higher participation requirements for two-parent families. But instead of creating mandatory work programs for them, many states simply transferred their two-parent welfare families to what are called "separate state programs" that had no participation requirements.)

Few states have been interested in taking on the challenge of a mandatory work program--especially since welfare caseloads seemed to melt away without their doing so. Unfortunately, the absence of true mandatory work threatens to undermine the new pro-work atmosphere that contributed to the caseload decline. The pro-work message rests on the enforceability of two related rules: First, recipients must pursue employment in good faith or face real consequences; and second, recipients hitting the five-year limit on benefits will either be dropped from welfare or suffer other repercussions. Both rules are vulnerable to being seen as bluffs if they are not enforced. And, as of now, they are not--unless the agency determines that the recipient has other options available.

Welfare reform has been accompanied by a lot of dramatic rhetoric among policy advocates about throwing families off welfare for not complying with work requirements or when the time limit is reached. But actual practice is much different. No state seems to have the political or moral stomach (rightly, I would add) to deny benefits to a single mother with children if she seems to be playing by the rules and looking for a job in good faith.

As TANF has taken hold, welfare agencies have grown more reluctant to decide that recipients have options other than welfare. First, those mothers still on welfare appear to be less employable than those who have left, making it increasingly difficult for welfare workers to determine whether recipients are really looking for a job. The tendency is to give mothers the benefit of the doubt, and not enforce strict work requirements. The economic downturn has made agencies even more accepting when recipients say they cannot find a job. Second, the five-year time limit on benefits is beginning to bite. Tens of thousands of families have reached their limit, and many more will do so in the next year or two. Some states have exempted all those hitting their limit or provided an alternate source of assistance, sometimes on the grounds of the weak economy. If enough recipients are simply exempted from mandatory work-related activities and time limits, welfare reform may come to be seen as a grand bluff. Imposing mandatory work is the only humane alternative to the termination of benefits when a single mother claims that she cannot find a job.

Because so little work or even work-related activity has been required under TANF, President Bush's bill reauthorizing TANF proposed major increases in both the required participation rate and in the required hours of work. Most congressional Democrats opposed the increases, and many apparently even wanted to water down the current hollow requirements. But surely some substantial work requirement is necessary, if only as an alternative to a time limit (preferably under an explicit federal authorization).

Support for "Working Families"

Bill Clinton popularized the phrase "make work pay." His goal was to "ensure that no one with a family who works full time has to raise their children in poverty." Most liberals and, apparently, many congressional Republicans have taken up his cause by expanding eligibility and benefits for programs not limited to welfare families. This attractive goal may make sense for single mothers with only one or two children. But as the number of children in a low-skilled household rises, an income above the poverty line becomes impossible to achieve without creating even larger disincentives to work and marriage--or spending tens of billions of dollars more a year to extend and slow phase-out rates to moderate-income families.

Work-related assistance to welfare families and those leaving welfare can mean the difference between a successful and an unsuccessful exit from welfare. But the expansions have already gone far beyond what would be needed to help families leave welfare (or avoid it in the first place).

Aid to working families not receiving cash welfare is at its highest level since the Great Depression. Between 1984 and 1999, total means-tested spending on families with children under 18 that are not part of AFDC/TANF rose fivefold, from about $20 billion to about $100 billion. (Compare that to the $60 billion or so spent on welfare families.) By 2001, total means-tested spending on families with children reached about $175 billion. The main programs are the EITC, food stamps, Medicaid, housing assistance, SSI, and subsidized school meals.

This increase was not primarily caused by a rise in population or a rise in poverty; rather, it was policy-driven. New programs were created and eligibility and benefits were expanded in many existing programs--and more are being proposed. Half of this increase came after the Republicans gained control of Congress in 1994.

Over 40 million people in families with children now receive these non-TANF benefits, compared to only around 16.5 million in 1984. (These figures exclude other benefit programs for the disabled and the elderly.) That is nearly three times the number of people on welfare at its highest point in 1994 (14.2 million). Indeed, although there are ambiguities in the data, it appears that from 1996 to 1999 the total number of Americans in families with children not on TANF that received some other means-tested benefit rose by about 5.9 million (or 25 percent)--about as many as left welfare.

These figures show that Bill Clinton's promise to "end welfare as we know it" deserved more careful parsing than it received. Old welfare is largely gone, but new welfare is bigger and, for its supporters, better than ever. It is billions of dollars and millions of recipients larger--and it enjoys much broader public support because it is tied to "working families."

As an example of the political armor these programs for working families now enjoy, consider how in 1995, when the Republicans wanted to scale back some of the recent expansions in the EITC, Democrats successfully labeled the proposal a "tax increase." Never mind that the families affected did not pay any taxes.

Such expansions in social welfare spending attract many supporters. Low-income, working families are more sympathetic beneficiaries of aid than are nonworking welfare recipients. Most of us would like them to live more comfortable lives and have their children succeed. Unfortunately, as with welfare, the expansion of this non-AFDC/TANF aid often comes with unavoidable increases in work and marriage disincentives--and sets in motion a political dynamic of ever-increasing expansions of eligibility and benefits.

These programs are all means-tested, which means that they terminate at a particular income level. They are either phased out over a range of incomes or end suddenly. In either case, they in effect reduce recipient incomes as they reach the eligibility cut off. This is social welfare's version of the tax code's marginal tax rate--but it can be much steeper.

The cumulative marginal tax rates of means-tested programs often range from 55 percent to over 100 percent and can affect families with incomes as high as $40,000 or more (for a family of three), particularly for families that receive benefits from subsidized housing, child care, or the State Children's Health Insurance Program (SCHIP). Such marginal tax rates are much larger than those in the tax code, and can have a greater impact on behavior--whether work or marriage--because the loss in benefits represents a much larger percentage of family income.

Ironically, past efforts to encourage work have now increased the marriage penalties for low-income working mothers. Between 1986 and 1998, the marriage penalty associated with the loss in welfare benefits and increased taxes for a nonworking mother marrying a man earning $15,000 per year fell by 75 percent, from about $6,000 to about $1,500. If, however, she was working and earning $10,000 per year, the marriage penalty would increase by 70 percent, from about $3,000 to about $5,000.

Given the complexities of the interactions involved, it is doubtful that most low-income couples fully understand or respond to these incentives. However, some groups, such as cohabiting couples, will see firsthand the difference marriage might make. Thus the debate over TANF's role in strengthening families and promoting marriage must take these incentives into account. Liberals and conservatives have made numerous proposals to reduce disincentives by slowing the phase-out rate of benefits. Unfortunately, such efforts require either reducing benefit and income-eligibility levels (which is unlikely to happen) or raising eligibility to yet higher incomes (and into the heart of the middle class), where they impose work disincentives and marriage penalties on far greater numbers of people, as well as cost much more.

Already, there are proposals from well-respected economists to extend the phase-out of benefits to incomes as high as $40,000 or $50,000. Some have suggested that married couples file separately, so that tax liability would depend on each individual's earnings. But this would cost tens of billions of dollars and would allow families with earnings above $50,000 to qualify. (That would take the subsidy well into the middle class, since, in 2001, median family income was about $51,000.)

Toward a European Welfare State?

This is not the place to attempt complex social accounting, but these new middle-class recipients will not necessarily gain from such massive income transfers. Unlike the poor, they pay taxes, and unless the taxes of only the very rich are raised to pay for these new benefits, the middle classes will eventually be taxed more--as happened in European countries through, for example, their Value Added Taxes.

Moreover, every increase in benefits and eligibility is likely to trigger a subsequent expansion. The economists and others who propose extensions seem to think that, once these initial correctives are made, eligibility and benefit levels will remain unchanged. This process, however, could easily take on a life of its own, just as it did in Europe. As benefits go to more middle-class families, the middle class will become the chief advocate for further benefit expansions--encouraged by the provider industries that benefit from the government subsidies. Already, advertisements on TV and in magazines and newspapers urge voters to contact their elected representatives to support expansions of Medicaid for long-term care for the elderly and Medicare reimbursement.

Many on the left welcome this gradual expansion of benefits, seeing it as a way for America to develop a European-style social welfare state. And that, I think, is the danger. Continued expansions could trigger a political process that, as in western Europe, would saddle the economy with immense transfer systems that create troubling distributional inequities, stifle business investment, create huge work and marriage disincentives, and lock political parties into bidding wars to win over the middle class with yet more government subsidies.

I am not naive enough to advocate a major rollback of these programs. But surely this startling expansion of the welfare state should be debated, and alternative approaches considered. One priority might be to get rid of as many "cliffs"--abrupt ends to a subsidy--as possible, since they often seem to have the greatest negative impact on behavior. In addition, the phase-outs of various programs should be coordinated, with consideration given to their interaction as well as the needs of recipients. Some programs might even be merged. There should also be a conscious decision about the income levels at which work and marriage disincentives should be greatest. One reason for keeping the disincentives at lower incomes would be that other tools can be used to mitigate them. For example, a TANF-like work requirement might be attached to non-TANF benefits, thus effectively lowering the value of the benefit without reducing the amount of assistance provided to a low-income family. Or, the benefits might be time limited, thus retaining the incentive to leave welfare without creating long-term dependency on a new form of welfare.

The Challenge Ahead

The success of welfare reform demonstrates that even the most seemingly intransigent social problems are not beyond our reach. Just as the wrong policies or the wrong programs encouraged dependency, the right ones seem to have encouraged work. Welfare reform helped to reduce dependency and started many hundreds of thousands of single mothers on the path to self-sufficiency.

But the process of reforming welfare is far from complete, and more complex challenges remain ahead. Establishing true mandatory work programs may be the best way to protect the pro-work culture of contemporary welfare. But few states have ever wanted them. Minimizing the work and marriage disincentives of aid programs to low-income, working families is also an urgent goal, but that will require rationalizing and capping the growth of benefits to moderate-income families that vote. Expect the debate to be long and heated.

Douglas J. Besharov is the Joseph J. and Violet Jacobs Scholar in Social Welfare Studies at AEI.