During the1995-1996 Debate on welfare reform, the greatest danger to the Republican legislation was not Democratic intransigence, although that certainly was a considerable threat. Rather, the major threat was that the Republican coalition would break apart on the rock of family formation. The majority of Republicans, especially in the House of Representatives, had been working for years on two parts of an agenda to transform welfare into a work program: first, to make work mandatory for a high percentage of welfare recipients; and, second, to improve the incentives for work, both by restricting welfare use and by ensuring that people who worked at low-wage jobs could make a decent living.

Many conservatives, however, were not convinced that the work agenda was ready for the challenge of reversing the nation's problems with poverty and-even more importantly from their perspective-the culture of indolence, non-marriage, single parenthood, and violence that had grown up around poverty. The conservative position on work is perhaps best articulated by Larry Mead1 of New York University, while the conservative position on illegitimacy is best articulated by Charles Murray2 of the American Enterprise Institute, both of whom have long histories of provocative thinking and writing about welfare. Mead argues that, although the poor have said that they want to work, in increasing numbers they have failed to do so. The conservative solution of several years ago, therefore, focused on rejecting the permissive welfare programs created by liberals and insisting on work by those seeking welfare. Ironically, the liberal writer and political analyst Mickey Kaus supported the work agenda because he believed Americans could not be accepted as full members of society unless they earned their own way.3 Indeed, Kaus's view of the work agenda was much more radical than that of conservatives because he believed that work requirements would transform the underclass culture and, over time, even promote marriage and improve child rearing.

Though Murray and a substantial number of other conservatives had nothing against the work agenda, they felt-in contrast to Mead and Kaus-that it was simply not of sufficient magnitude to address the nation's problems with poverty, violence, and other urban ills. For them, the real problem was that many Americans had rejected marriage and family. Illegitimacy was, to Murray and others, like sin to a Baptist preacher; it was the root of all evil. Unless something could be done to reverse the trend toward rearing an increasing number of the nation's children in female-headed families, they believed welfare dependency, school failure, violence, and intergenerational poverty would continue unchecked.

After a great deal of intra-party fighting-and not all of it in good spirit-Republicans in the House of Representatives agreed among themselves and with a range of conservative groups outside Congress that they would create a welfare reform bill addressing both the work agenda and the family agenda. That bill, in modified version, was signed into law on August 22, 1996. The purpose of this essay is threefold: to review the provisions of the 1996 legislation intended to promote work and reduce illegitimacy; to examine evidence for the effects of both types of provisions; and to discuss a strategy for the next phase of welfare reform, beginning with reauthorization of the 1996 legislation in 2002.

Overview of the 1996 Welfare Reform Law

Work Provisions

The heart of the 1996 reform legislation was the replacement of Aid to Families with Dependent Children (AFDC) with Temporary Assistance for Needy Families (TANF). The TANF program can be summarized in five provisions. First, the entitlement to cash benefits that characterized the old AFDC program was discontinued. For many liberals, such as Senator Daniel Patrick Moynihan, ending this entitlement was the most important reform and the one they detested the most because, in their view, it signaled the end of the nation's guarantee to help destitute families. Instead of an absolute promise of benefits, TANF required adults to make serious efforts to end their dependence. If they did not, their benefits could be reduced or even eliminated. As Larry Mead argued for many years, for Republicans, the end of entitlements meant the beginning of personal responsibility.4

Second, TANF terminated open-ended funding and replaced it with block grants. Under AFDC, states were guaranteed more federal money for every adult they added to the welfare rolls. For every person they removed from the rolls, the federal government "rewarded" them by reducing federal funding (by an average of about 55 cents on the dollar).5 As conservatives often point out, this system created harmful incentives that punished states for successfully helping people leave welfare for work. In contrast, TANF provided states with block grants that gave them a fixed amount of money annually. If states were able to help people find jobs and leave the welfare rolls, they could use the freed-up federal funds to invest in education, training, childcare, transportation, or other programs that assisted low-income families.

A third major feature of the legislation focused on strong work standards. Ironically, with the help of Senator Daniel Patrick Moynihan, Republicans managed to place participation requirements in the 1988 Family Support Act. These standards required states to place a certain (but very low) percentage of their caseload into some type of constructive activity for a minimal period of time each week. Although these standards were weak-especially because over half the families on welfare were exempt and because the standard could be met by education and training without any actual employment-they nonetheless set the stage for achieving stronger requirements in the future. The future arrived in 1994, when Republicans assumed control of the House and Senate and immediately introduced legislation that contained strong work requirements for welfare recipients. The standards required that states have 25 percent of the entire welfare caseload in a work activity in 1997, 30 percent in 1998, and so on, until every state, by 2002, had 50 percent of its caseload in a work activity for at least thirty hours per week. In addition, there was a separate provision that, when fully implemented, required states to place 90 percent of their two-parent families into work programs. The legislation carefully defined work activities so that very few education and training programs could count toward the required hours. States that failed to meet these standards were subject to stiff and cumulative financial penalties.

The fourth major characteristic of the legislation required states to impose sanctions on adults who did not meet work requirements. Many conservatives believed that previous reforms had failed not only because they enforced minimal work requirements but also because they contained no serious consequences for recipients who violated those requirements. The 1996 reforms, however, not only required states to use sanctions when recipients failed to meet work requirements but also allowed them to use stronger sanctions than in the past, including complete termination of cash benefits.

The final key provision in the 1996 reforms was the five-year time limit. Next to the discontinuation of the cash entitlement, this was the provision most deplored by liberals. Even some conservatives, such as Larry Mead, did not support the time-limit provision because they thought work requirements backed by sanctions would be sufficient to convince welfare recipients that working toward independence was now a serious requirement. House Republicans, who first proposed the measure and who strongly defended it throughout the nearly two-year congressional battle, believed that if cash welfare were to become a temporary program, it would be necessary for recipients to know from the moment they applied that the benefits would end at a certain time.

Given the strong possibility that some families would not be capable of supporting themselves, Congress granted states the power to exempt up to 20 percent of their caseload from the five-year limit. In addition, states could use some of the dollars they were required to continue spending on welfare to help families who had passed the time limit. Although liberals remained critical of the time-limit provision, Republicans believed that it struck a balance between sending a strong message that welfare is temporary and recognizing that some adults would be incapable of supporting their families through work within five years.

Enacting even one of these five major provisions would have been unthinkable until Republicans assumed control of both Houses of Congress in 1994. The enactment of all five provisions constitutes the most complete change in American social policy since the War on Poverty of the 1960s, and perhaps even since the New Deal of 1935. The thrust of the provisions might be described in two contrasting ways. Clearly, they signaled that it was no longer possible for able-bodied adults to join welfare and stay dependent on it for many years. But, as seen in the House floor debate on the various versions of the legislation, these provisions were based on the affirmation that adults on welfare were capable of doing more to support themselves and their children. Permissive governmental welfare programs had induced a kind of learned helplessness in millions of young adults. As Clay Shaw, the welfare reform leader of House Republicans, stated on many occasions, Republicans thought of themselves as being on a "rescue mission." Their goal was to help people help themselves to gain independence from the welfare trap by encouraging or even forcing them to accept their responsibility to find jobs and leave welfare. It is impossible, even for liberals who bitterly opposed the legislation, to listen to or read the words of Republicans such as Clay Shaw, Nancy Johnson, Jim Talent, Bill Archer, Jim McCrery, Tom DeLay, Mike Castle, and many others, without admitting that these Republicans were convinced that personal responsibility was better for adults and children than welfare dependency-and that most welfare recipients were capable of achieving it.

Once President Clinton signed the congressional bill on August 22, 1996, the days of non-contingent cash welfare payments were over. For better or worse, the nation had adopted a radically new approach to providing for its destitute citizens.

Illegitimacy Provisions

Until 1995, illegitimacy had been of only sporadic concern to policymakers. In 1965, Daniel Patrick Moynihan, then an assistant secretary in the Department of Labor, electrified the world of policy with his infamous paper on the role of illegitimacy in destroying black families.6 Moynihan's perspective, equally telling in 1995 as it was in 1965, was that family breakdown was the scourge of domestic tranquility.7 To the horror of less insightful and less honest observers of American culture, he argued that any society that allowed its young men to be reared without significant male influence asked for and got what it deserved: chaos. For three decades after the publication of Moynihan's paper, the topics of minority families, in general, and illegitimacy, in particular, were treated with malignant neglect by scholars, policymakers, and members of the media. Meanwhile, the relentless growth of illegitimacy among both blacks and whites made the illegitimacy rates that prompted Moynihan's paper in 1965 seem quaint. Black illegitimacy rates more than doubled between 1965 and 1995, by which time almost 70 percent of black babies were born outside of marriage. Also, by 1995, the illegitimacy rate among whites was at the same level as it was among blacks in 1965, when Moynihan accurately predicted disaster.8

Charles Murray stirred this pot again in 1993 with a full-page op-ed piece in the Wall Street Journal.9 Murray argued that whites were now reaching the same level of illegitimacy that had existed among blacks when Moynihan wrote his provocative 1965 paper. More disturbingly, evidence that entire neighborhoods were tipping toward high-density fatherlessness was beginning to accumulate. Echoing Moynihan's argument of thirty years earlier, Murray maintained that neighborhoods with few fathers created an environment in which young males could not be properly socialized. Without the firm hand of fathers in their homes or even in their neighborhoods, adolescent males were likely to terrorize entire blocks with property destruction, intimidation, and violence. Indeed, throughout the 1970s, 1980s, and 1990s, hundreds of black neighborhoods reached this state of desperation, as Moynihan had predicted. By 1993, according to Murray, white neighborhoods throughout the nation were moving in the same direction.

If Murray was the intellectual leader of the Republicans who were arguing for the pre-eminent place of illegitimacy on the welfare reform agenda, Robert Rector of the Heritage Foundation rallied ground forces for the battle. Rector, widely respected among conservatives, was the leader of a coalition of influential family groups that included the Christian Coalition, Concerned Women for America, Traditional Values Coalition, Eagle Forum, and others. As the leader of this coalition, Rector had substantial influence among Republican leadership in the House and with a number of conservative members of both the House and Senate. Although a large majority of Republicans believed that the issue of work was the most politically attractive, important, and achievable part of the Republican reform agenda, they also agreed that illegitimacy was an exceptionally important issue. Rector and his allies, however, constantly threatened to oppose the Republican bill unless additional provisions were included to address the problem of illegitimacy.

Perhaps the two most important provisions to address the problem of illegitimacy were the family cap and benefit termination for unmarried teen mothers. The family cap would have prohibited states from providing additional cash benefits to single welfare mothers who have additional babies. There were several versions of the teen benefit termination provision, but all would have ended cash benefits for teen mothers for some period of time. The version of teen benefit termination that prohibited cash benefits until the mother reached the age of eighteen was included in the first welfare reform bill enacted by the House in March of 1995. So was the family cap, which would have denied additional benefits to mothers who conceived children while already receiving governmental assistance. But both measures were defeated on the Senate floor; even a majority of Republicans opposed the teen benefit termination provision.

These two votes on the Senate floor demonstrated that tough provisions on illegitimacy were controversial even among Republicans. Nonetheless, Rector and other conservatives repeatedly threatened to oppose the final bill, at least until the family cap provision was included. In July of 1996, as the House-Senate conference committee was preparing a third welfare reform bill (two had already been vetoed), President Clinton announced publicly that he would sign the bill. As a result, difficult negotiations between conservatives outside Congress and Republican conferees took place the night before the bill was brought to the House and Senate floors for final votes. Senate conferees refused to include the family cap or the benefit termination provision, claiming that either would doom the bill on the Senate floor. As a result, House negotiators worked with Senate Republicans and outside conservatives to include not only additional provisions on illegitimacy in the final bill but also a stipulation that money for abstinence education would be changed from funding that required an annual appropriation to funding that guaranteed five years of money at $50 million per year.

The final bill contained at least fifteen provisions that addressed illegitimacy (Table 1), but neither the family cap nor the teen benefit termination was among them. However, because federal legislation not only allowed states great flexibility in designing their welfare programs but also eliminated the entitlement to cash welfare, states were free to include both the family cap and the teen benefit termination provision in their new programs. More than twenty states have since adopted the family cap, but no state has passed legislation to reduce or eliminate payments to unmarried teen mothers.

The policies listed in Table 1 represent the first major federal attack on the problem of illegitimacy. Each of these policies is responsive to Moynihan's prescient identification of this national problem in 1965. Why it took three decades to launch a major attack on this significant problem is a subject that deserves book-length treatment. But the reason this problem was brought so forcefully to public attention in 1995-and why such sweeping and previously unheard of legislation could be enacted in 1996-was that the Republicans, especially the Murray-Rector axis, forced the issue onto the public agenda. Indeed, prior to 1995, House Republicans had been fighting this battle among themselves-and with conservatives outside Congress-for several years. In all likelihood, they would have continued fighting among themselves, and largely outside the view of the American public, had the 1994 election not intervened. Before 1994, not many people cared about Republican views on illegitimacy. And with Democrats firmly in control on Capitol Hill, no serious legislation on reducing illegitimacy rates had ever been brought to the floor of either House, and none was likely. However, in 1994, Republicans earned not only the majority in both Houses of Congress but also the right to set the reform agenda.

An unfortunate aspect of the legislation aimed at fighting illegitimacy was that no one knew how to reduce out-of-wedlock births. Murray and others had introduced the idea that welfare benefits made it possible for young women to have illegitimate children whom they could not support, and by 1995 there was extensive literature on this issue.10 Although liberals denied that the research on welfare benefits and illegitimacy showed any connection, a fair reading of the studies reveals that a majority did find a significant correlation and that more recent studies showed an even stronger correlation.11 Even so, in 1995 and 1996, when these provisions were being debated, there were few programs that had been shown conclusively to reduce the rate of illegitimate births (there are now reliable studies showing that programs for teens can reduce sexual activity, increase the use of birth control, and reduce pregnancy).12 Most of the programs involved helping adolescent girls, especially poor and minority girls, to understand the future implications of having children before they were ready. Some programs involved advice and training in abstinence, some involved instruction in birth control, some involved mentoring, and many offered a range of social services that were designed to help adolescents do better in school, prepare for jobs after graduation, and generally address issues faced by most adolescents. Several projects also attempted to help single young mothers who had recently given birth to children outside of wedlock. As Rebecca Maynard, a researcher with a long history of studying family issues, shows in an excellent review of these studies, large and expensive programs that lavished attention on adolescent girls failed to have significant impacts, even in reducing second births.13

The issue for Republicans in 1995 was not whether effective interventions were available; the issue was how to bring to national attention a devastating social pathology that was a major cause of many of the nation's domestic problems. Moynihan's original attempt to focus the nation's attention on illegitimacy and the breakdown of the family had failed; now Republicans were determined to force the issue onto the public agenda. Table 1 shows that they were successful in achieving this first stage in the battle against illegitimate births.

Other Provisions

A brief overview of the remaining provisions in the 1996 law is necessary in order to appreciate its sweeping nature. Either the work requirements or the illegitimacy requirements by themselves would have marked the legislation as historic, but there were many additional provisions that gave the 1996 law a magnitude that is still not widely appreciated.

Supplemental Security Income (SSI) for Children. The provisions on children's SSI that were included in the 1996 law were intended to tighten eligibility requirements. When Congress created SSI in the early 1970s, the program was intended to provide a guaranteed income and Medicaid benefits to individuals who were seriously disabled. For adults, the concept of disability had a practical meaning because the criterion was generally that benefits were appropriate for anyone whose disability was severe enough to prevent work. But the legislation was unclear with regard to extending benefits to children with disabilities. Hence, it was not surprising that admitting children to SSI was an inexact process. Inexact or not, Republicans, especially Representative Jim McCrery of the House Ways and Means Committee, concluded that the criteria were far too loose and that millions-or even billions-of taxpayer dollars were being wasted on providing disability benefits to children who could not, under any reasonable definition, be considered disabled.

The McCrery provision changed the definition of disability for children and also strengthened the criteria used to assess disabilities. The single biggest change was prohibiting the use of the Individualized Functional Assessment, a test that often yielded the conclusion that children qualified for SSI, even when they had only minor problems. At the time, the Congressional Budget Office estimated that the new definition of disability and the new procedures would result in about 180,000 fewer children on the rolls.14 In developing the exact procedures to implement congressional intent, the Social Security Administration (SSA) met several times with congressional members and staff. The criteria they developed to implement the changed definition and procedures would, they estimated, result in lost benefits for about 120,000 children.15

Child Support Enforcement. The child support enforcement amendments of 1996 were the most sweeping in the history of the child support program. The reforms were multifaceted, requiring states to adopt new enforcement methods, such as canceling the fishing, hunting, and driving licenses of fathers who are delinquent in payment of child support; imposing strong data processing requirements on states; strengthening paternity establishment requirements; mandating that states share more child support collections with mothers who leave welfare; requiring states to adopt several procedures that would strengthen interstate enforcement (about one-third of child support cases are interstate); and creating new databases that would promote the rapid location of, and access to, vital information on non-custodial parents.16 As surprising as it might seem, given the very partisan level of debate on TANF, the child support provisions enjoyed significant bipartisan support.

Benefits for Non-Citizens. The sweep of the provisions that severely limited welfare for non-citizens was historic. According to the new law, non-citizens must sign an agreement stating that they understand that financially they must rely on their own efforts upon entering the country, and that if they need assistance, they will have to turn to their private sponsors rather than go on public welfare. Second, non-citizens must have legally enforceable affidavits of support from individuals or organizations that have agreed to provide them with financial and other support, and those parties must be responsible for reimbursing the government if the non-citizen receives welfare benefits. Third, with a few exceptions (including medical emergencies and cases of child abuse), non-citizens are not eligible for welfare for five years after entering the country. After five years, ineligibility for SSI and food stamps continues, but states may choose to provide non-citizens with Medicaid and TANF benefits. Fourth, non-citizens are eligible for a substantial set of means-tested employment and training programs that they may use to improve their job qualifications. Once immigrants become citizens, they are eligible for welfare programs on the same basis as other citizens.

Unfortunately, in order to eliminate the budget deficit, Republicans included another provision on welfare for non-citizens that was politically unwise as well as questionable in terms of policy. Non-citizens already receiving welfare benefits were to have those benefits terminated one year after the passage of the new law. The problem was that this provision would have caused hardship among non-citizens, especially the incapacitated and elderly, who had become dependent on benefits. In signing the final legislation, President Clinton called attention to this provision and promised the nation that he would work to repeal it in the next session of Congress. Clay Shaw, joined by the Republican congressional leaders and Republican governors, agreed with the president. After extensive negotiations, Congress and the president agreed in 1997 to repeal the ban on SSI for non-citizens who were already receiving this benefit when the legislation was enacted. In addition, non-citizens who were in the country when the bill passed and subsequently became disabled were eligible for SSI.

This change in the 1996 law has been widely misinterpreted as signaling an end to the ban on welfare for non-citizens. Nothing could be further from the truth. Only those already receiving benefits or who were in the country on August 22, 1996, were removed from the ban. All new entrants to the United States, with the exception of refugees, are banned from receiving welfare benefits except under emergency conditions. In the long term, this provision will mean that virtually no non-citizens-with the exception of refugees-will be receiving welfare benefits.

Childcare. The major childcare reforms can be captured in three generalizations. First, as is the case with almost every area of federal social policy, federal childcare policy suffers from the affliction of multiple programs. In 1995, more than sixty different federal childcare programs existed. Amazingly, the two committees in the House of Representatives that had jurisdiction over most of the major childcare programs-Ways and Means as well as Education and the Workforce-agreed to terminate several of their programs and combine the funds into a single existing block grant program. This action cut red tape for both federal and state governments, and it also facilitated state efforts to provide coordinated services to poor and low-income families. Second, the amount of funding in the block grant was increased by $4.5 billion over six years. In addition, to further promote state flexibility in the use of federal dollars, states were allowed to transfer up to 30 percent of their TANF funds into the childcare block grant. In this way, states were given substantial flexibility and new federal resources to provide childcare to mothers on welfare, mothers leaving welfare, and other low-income families. Third, Republicans resisted all attempts to impose federal regulations on the nation. As a result, regulations and quality control remained under the purview of states, localities, and parents.

Results at Five Years

Based on extensive information gathered from state studies, national data sets, and formal evaluations, welfare reform has produced results that, unlike most social policies, are very close to those expected by its authors. In fact, the effects have exceeded expectations. Although a great deal more will be learned in the years ahead about the effects of welfare reform on poor and low-income families, the basic picture that has emerged is likely to stand the test of time. Reviewing the results will provide a solid basis for discussing steps toward further reform that should be pursued during the reauthorization debate, and as part of the next phase of welfare reform.

Temporary Assistance for Needy Families (TANF)

Decline in Welfare Rolls. As shown in Figure 1, the welfare rolls have declined dramatically since 1994. Although there is now a great deal of talk about how the economy has contributed to this unprecedented decline in the rolls, a review of the historical data in Figure 1 shows that historically, strong economies have not influenced the welfare rolls. In fact, between 1959 and the current decline, which began in the spring of 1994, there was not a single period of sustained decline in the rolls. The biggest decline was a mere two percentage points in 1982.

It is especially instructive to compare welfare enrollment during the economic recovery of the 1990s with the recovery of the 1980s. Although the economic expansion of the 1990s was impressive, the expansion of the 1980s was notable as well. Indeed, during the 1980s, the economy generated a net increase of nearly nineteen million jobs-almost identical to the job generation of the 1990s expansion.17 If the economy does, in fact, play a part in reducing the rolls, then they should have fallen sharply during the job-prolific economy of the 1980s. Instead, the rolls were more or less flat during the early years of the 1980s, after which they increased dramatically. In fact, over the entire recovery, the rolls increased by almost 12 percent. I have yet to see any analysis explaining how the welfare rolls could increase substantially during one of the strongest and most job-rich economic expansions in history if a growing economy does indeed draw people away from the welfare rolls and into employment.

Another misleading claim about caseload reduction-a claim fostered by many politicians-is that welfare reform has been successful merely because the rolls have declined. While it would be foolish to denigrate the movement of able-bodied adults off of welfare, saying that the reduction in the rolls is the only criterion of success is simply wrong. Indeed, Congress has provided lavishly for gathering information about the effects of the new law. First, they have required states to completely revamp and strengthen the reporting of data in their cash welfare programs. In addition, they have required states, for the first time, to report data on childcare. Second, they have allocated $15 million per year to the Department of Health and Human Services to fund scientific studies on welfare reform (such as its impact on children), evaluate state programs, conduct surveys, or sponsor any other type of scientific study to learn about the effects of welfare reform. Third, the law has given the Census Bureau $70 million over seven years to conduct a national longitudinal study on the condition of low-income children and families. I do not hesitate to claim that the 1996 welfare reform legislation included provisions for tracking and evaluation that were better and more elaborate than those in any other social legislation of recent decades. In addition, because American foundations have donated generously to welfare evaluation and research studies, there is now an array of data available for measuring the effects of welfare reform.

Before we leave the topic of caseload data, it should be emphasized that the major goal of the welfare reform law of 1996 was to end dependency. As long as able-bodied adults remain on welfare, dependency will continue to be an issue. However, the dramatic reduction in welfare rolls constitutes a prima facie case that welfare dependency is declining.

Employment. A major goal of welfare reform is to increase employment among mothers on welfare. There are at least two types of data available to measure employment. First, there are now more than forty state studies on individuals who have left the welfare rolls. These studies were recently reviewed in great detail by the Congressional Research Service,18 which found that at the time that states obtained data from former welfare recipients, an average of 57 percent of them were employed. Nearly 75 percent of people leaving welfare had been employed at some point since leaving welfare.

A second type of employment data is supplied by the annual Current Population Survey (CPS), a national survey conducted on a sample of fifty thousand households that are representative of the entire nation.19 This type of survey complements the state studies of welfare leavers discussed above because these state studies miss adults who intended to join the welfare rolls but were diverted from doing so by new state policies, such as job search assistance programs, that are designed to keep adults from going on welfare in the first place. Because the CPS represents the entire population, mothers on welfare, mothers who have left welfare, and mothers diverted from welfare are all represented in proportion to their true numbers in the population.

Consistent with data from the state leaver studies, Figure 2 shows very substantial increases in employment by single and never-married mothers since the mid-1990s. With regard to all single mothers, employment had been nearly flat for a decade preceding the welfare reform era. But in 1994, employment among single mothers began an increase that lasted until 1999-the most recent year for which data is available-at which time the employment of single mothers actually exceeded employment by married mothers. Also, during this period, employment by never-married mothers, a subgroup of all single mothers, increased even more than that of married mothers or all single mothers. In fact, although never-married mothers had been, for many years, the mothers most likely to go on welfare and stay there for long periods of time, their rate of employment increased almost 40 percent in just three years. Employment by single mothers and those formerly on welfare has increased dramatically and is now at an all-time high.

Poverty. Some conservatives argue that because breaking the culture of dependency is the major goal of welfare reform, reducing poverty is not an important outcome measure for welfare reform. However, the American public has long held that reducing child poverty is an important public policy goal.20 And, in a democracy, following public opinion is the preferred course of action-unless, of course, there are strong countervailing reasons. Moreover, there is a long tradition, observed by both Democrats and Republicans, of the federal government accepting responsibility for helping the poor and disadvantaged. This responsibility is evidenced in the substantial increase in spending on poor and low-income families since the early 1960s.21 Finally, the goals of reducing dependency and reducing poverty are not incompatible. If accepting poverty reduction as a goal of welfare reform means reducing the focus on dependency reduction, then the decision to emphasize the issue of poverty will involve trade-offs. But under most circumstances, reducing dependency and reducing poverty are entirely compatible goals.

The news on child poverty, for instance, is encouraging. During the welfare debate, many liberals argued that the harsh Republican welfare reform bill would greatly increase child poverty. Indeed, the Urban Institute, working with Wendell Primus, a top HHS official who eventually resigned from the Clinton administration in protest of the president's decision to sign the bill, predicted that the Republican bill would cast more than one million children into poverty.22 This estimate, of course, was used frequently during the welfare debate by the news media, especially the Washington Post and the New York Times. Now, after five years of welfare reform, there is sufficient data to reach a conclusion about the effects of the 1996 law.

Figure 3 shows the course of poverty since 1960. Child poverty rates declined quite substantially during the 1960s, but the general trend was upward for the following twenty-five years. Between 1969 and 1993, child poverty rates grew from 14 percent to 22.7 percent, an increase of over 60 percent. Even during the substantial economic recovery of the 1980s, poverty rates decreased by only 2.7 percentage points. Given this background, it is impressive that poverty rates have fallen every year since 1993.

Figure 4 provides a more detailed look at the decline in child poverty and the decline in the welfare rolls. In the set of three bar graphs for each year, the first depicts the percentage of decline in the welfare rolls that year; the second, the decline in overall child poverty; and the third, the decline in black child poverty. All three measures declined every year. Moreover, not only did child poverty fail to increase, as predicted by liberals, but the declines have been greater than at any time since the 1960s. The declines in black child poverty are especially notable. In both 1997 and 1999, poverty among black children declined more than in any other year since the Census Bureau began computing a separate rate for blacks in 1974. By 1999, black child poverty had reached its lowest level ever.

As impressive as these figures appear to be, Figure 5 shows that they actually underestimate the true extent of poverty reduction. The official standard of poverty is flawed-especially during an era in which work is increasing rapidly-because it ignores non-cash governmental benefits such as food stamps and taxes, including the very generous Earned Income Tax Credit (EITC). A mother with two children earning $10,000 would be eligible for a total of $6,000 in benefits from just these two programs, but the entire $6,000 is ignored by the official poverty measure.

Fortunately, the Census Bureau now calculates several experimental poverty measures that include income from food stamps, the EITC, and a few other governmental benefits, such as housing assistance to poor and low-income families. Figure 5 compares the progress against child poverty during the 1980s and the 1990s, using one of the broadest measures of poverty calculated by the Census Bureau.23 During both decades, the initial year was the first in which poverty declined during the respective economic recoveries. A comparison of the six years in which poverty declined during the 1980s with the six years in which poverty declined in the 1990s, reveals that the decline of the 1990s, as shown in the last panel, was more than twice as great. Thus, the broader poverty measure shows clearly that progress against child poverty under welfare reform is even deeper and more substantial than the already hopeful picture offered by the official poverty measure. Under either measure, progress is enormous.

Illegitimate Births

Figure 6 traces the course of three measures of illegitimacy rates between 1940 and 1999. The pattern shown by all three measures is quite similar, with increases nearly every year (although the percentage of births to unmarried women did not expand rapidly until about the mid-1960s) and a leveling off beginning in the early to mid-1990s. Given the major emphasis on illegitimacy provisions in the 1996 law (see Table 1), it is tempting to claim that there is some relationship between welfare reform and the leveling off of illegitimacy. However, most of the measures seen in Figure 6 started to level off before the federal legislation of 1996, which makes positing a causal relationship somewhat difficult.

Even so, it is certainly a hopeful sign that all three measures have shown moderation for the first time in several generations. Moreover, it is, at the very least, interesting that this moderation occurred during the era of welfare reform, especially if we date the era from the early to mid-1990s, as states began aggressively to implement their own work-based welfare reforms under federal waivers. A possible mechanism by which work could reduce illegitimacy is this: Young women might come to realize that they cannot rely on welfare and are instead going to be forced to work to support their children. This consideration might cause them to take action to avoid pregnancy, and possibly to focus more attention on their future. Similarly, aggressive enforcement of child support could influence males to avoid causing a pregnancy because they know that having a child means that they will be hounded by child support officials for eighteen years or more. And in many jurisdictions, men who do not pay their child support can face jail time. There is only modest evidence to support these mechanisms, but both are plausible and will likely be the focus of empirical studies over the next several years.24

One additional point should be made about welfare, illegitimacy, and marriage. In one of the more remarkable social science experiments of recent years, the Minnesota Family Investment Program (MFIP) required welfare families in an experimental group to face mandatory work requirements. In addition, the families were provided with substantial work incentives, not just in the form of the EITC but also through a generous state income disregard policy, whereby welfare benefits failed to decline with increased income. Under this income disregard policy, adult welfare recipients who went to work were allowed to retain a substantial part of their cash benefit. Their childcare was also guaranteed. With all of its provisions, the MFIP benefits package meant that many of the families who took advantage of the experiment and went to work for around $10,000 or $11,000 per year received cash, food stamp benefits, and income supplements that increased their total income to about $18,000 or $19,000. Families in the control group, however, were given the regular AFDC program with no work requirements and no generous income disregards.

The results of the MFIP experiment are arresting: Among those in the experimental group, work and earnings increased, divorce decreased, marriage increased, domestic violence declined, and children's school performance improved. This set of results makes MFIP one of the most impressive social experiments ever conducted.25

These results also introduce a slight dilemma for conservatives. What if the impacts on marriage and children are produced by the combination of mandatory work and generous subsidies? Would conservatives be willing to spend the several billion dollars per year required to help families who are working full-time get to an income of $18,000 or $19,000 per year through an expanded EITC, a more generous and refundable child tax credit, or some other mechanism? My own view is that the MFIP data are not entirely convincing, mostly because the results need to be replicated in other locations. But so far, the MFIP results are by far the most encouraging for those who want to increase marriage rates while simultaneously increasing work effort.

Results for Other Programs

SSI for Children. The intent of the children's Social Security Insurance (SSI) reforms was simple but controversial: to reduce the growth in the number of children receiving SSI on the grounds that too many children who were admitted to the program were not disabled under any reasonable definition of disability. As seen in Figure 7, the reforms have had precisely their intended effect. The rolls, which were growing rapidly prior to 1996, immediately stabilized after the enactment of the welfare reform legislation. The rolls declined by about 100,000 in 1997, and they have declined by about another 30,000 since then. As suggested by Figure 7, and as individuals from the General Accounting Office testified at the House Ways and Means Committee hearing on child support enforcement, the rate of growth in the children's SSI program in 1996 would have produced at least 1.2 million SSI child recipients by 2000.26 Instead, in large part because of the reforms, there were about 850,000 child recipients. If the rolls are down by 350,000, as these figures suggest, taxpayers are saving about $2 billion per year in SSI cash benefits. There are, undoubtedly, additional savings in Medicaid.

Some would argue that the loss of SSI benefits has caused hardship in some of these families. As far as I can tell, there is no evidence of increased hardship, except in the sense that some families lost a cash benefit that they were accustomed to receiving. The real issue is whether families who have children with serious disabilities have been dropped from the program. Of course, as in the case of the 1996 debate over SSI reforms, everything hinges on the definition of disability. In this regard, the following facts are important: The Clinton administration worked closely with the Committee on Ways and Means to develop the new definition and the procedures for establishing whether the definition had been met by individual children; no administration official claimed that the reforms were leaving deserving children behind; both the Social Security Administration's inspector general and individuals from the General Accounting Office testified that the old definition was unclear, and they worked with the committee to develop the new definition; and advocates have not launched a detailed criticism of the new definition and procedures in the five years since the reform legislation of 1996. Moreover, there does not seem to be any serious debate in Congress about children who unfairly lost their SSI benefits or about changing the 1996 SSI reforms.

Child Support Enforcement. Almost all observers agree that the child support amendments that are being aggressively implemented by states seem to be producing results. Figure 8 compares two statistics: the number of paternities established and total child support collections. The improvement in paternity establishment from 1994 to 1999 is spectacular.27 Two provisions were especially important in improving paternity establishment. The first required mothers to provide extensive information on the father of their child or suffer a reduction in their benefit. The second required states to set up programs to establish paternity in the hospital at the time of birth. In many respects, these two policies stand as a kind of metaphor for the entire welfare reform legislation. One policy requires the use of sanctions to compel mothers, if necessary, to do the right thing by identifying fathers. The second policy establishes a positive and voluntary program in hospitals that approaches new fathers, asking them to acknowledge paternity voluntarily because of the many benefits it could confer on their child. By any measure, paternity establishment is a great success story and testifies to the creativity and implementation skill of the child support community.

The dramatic 60 percent increase in child support collections between 1994 and 1999 is also a success story. Here, the underlying causes will likely never be fully known. As mentioned previously, child support enforcement was reformed more extensively than any other program, with the exception of AFDC. In fact, with regard to child support, there were at least five major reforms and scores of more modest reforms, making it virtually impossible to separate the effects the particular reforms. Moreover, the child support program was also extensively reformed in 1988, and many of those reforms were still building toward their full impact when the 1996 reforms were implemented. But perhaps it is unnecessary to examine a good thing too closely. All indications are that the child support reforms of 1996 are having substantial positive impacts.

Benefits for Non-Citizens. As with SSI benefits for children, the reforms of welfare programs for non-citizens appear to have been effective. Figure 9 shows that enrollment by non-citizens in the four major welfare programs has dropped considerably.28 The declines range from about 20 to 40 percent over the three years between 1995 and 1998. It can be expected that the drop in non-citizen enrollment in these programs will continue to fall as people begin to age out of them. Eventually, unless the policy is changed, virtually no non-citizens (except refugees) will be drawing welfare benefits.

Like the reforms of SSI for children, the goal of Republicans in enacting the reforms of welfare benefits for non-citizens was simply to reduce the number of people eligible for welfare. In these broad terms, the reforms are meeting their major objective. However, many policymakers would argue that an important issue is whether these reforms are causing undue hardship. But so far, there is limited evidence of hardship. The legislation did not simply leave non-citizens to their fate; as outlined above, it provided for emergency benefits for non-citizens and for assistance to be provided by sponsors rather than taxpayers. It would be worthwhile to find out whether these new mechanisms are working.

Value judgments are often key in deciding, from a policy standpoint, whether non-citizens should receive welfare benefits. As early as 1993, as a result of the efforts of a task force on welfare reform appointed by the Republican leadership in the House of Representatives and headed by then-representative Rick Santorum of Pennsylvania and Tom Delay of Texas, House Republicans concluded that American taxpayers should not be held responsible for providing welfare benefits to non-citizens. Emergency benefits, and even education and training programs by which non-citizens could improve their prospects, were considered reasonable, but not welfare benefits. Financial assistance should be the responsibility not of taxpayers but of individual or institutional sponsors. In every major House Republican bill after 1993, the restriction or elimination of most welfare benefits for most non-citizens was a primary feature. This provision is now the law of the land, although many liberals and Democrats disagree with the policy.

Reauthorization and the Next Phase of Welfare Reform

Instead of addressing only issues surrounding work and illegitimacy, I have emphasized the entire range of reforms contained in the 1996 legislation to make two points. First, the Republicans who drafted the reforms agreed on the foremost importance of personal responsibility, which they then applied to a broad range of welfare programs. At least partly as a result of these reforms, there are now fewer families dependent on welfare, more single mothers working, fewer illegitimate births, fewer non-citizens living on welfare, more fathers paying more child support, and fewer healthy children and adults depending on the SSI disability program. The number of people in the United States pulling their own weight is almost certainly higher now than at any time since the War on Poverty began in the 1960s. And, although I have not emphasized this side of the reforms, the federal government is spending less money on these programs than it was in 1995.29 I think it would be fair to conclude that the Republican philosophy on welfare for the able-bodied now dominates domestic policy, and that the breadth and depth of success of the 1996 reforms have exceeded even what conservatives had predicted.

In the midst of these successes, Congress is poised to consider welfare reform reauthorization, which must be completed by October 1, 2002. There is much that is worth conserving in the 1996 legislation. To be sure, there are problems with the welfare reform law, but TANF is the best social policy for the poor that the United States has ever had because it is based on an accurate understanding of human nature and is grounded in an ethic of personal responsibility. It follows, then, that the basic structure of the law is solid.

Congress should recognize the success of the personal responsibility agenda and vigorously pursue a three-part strategy to continue promoting this successful campaign. First, the work agenda should be maintained and even strengthened. Second, Congress should expand the family formation agenda. The 1996 reforms intended to reduce illegitimacy have had modest success, at best. The lack of dramatic success is not due to any flaw in the goal of the reforms; rather, it is, in all likelihood, the result of deep and pervasive cultural practices. As both Moynihan and Harvard scholar Orlando Paterson have persuasively argued, black Americans have had a shortage of two-parent, married families for many generations, in part because the black family-and especially the position of black males-was severely compromised by slavery and Jim Crow.30 In the case of whites and, increasingly, Hispanics, the American culture of divorce and sexual liberty has, over the past three or four decades, led to a similar weakening of the two-parent family and marriage. But as we have seen, in recent years, these trends have at least leveled off. While maintaining a focus on reducing illegitimacy, Congress should now emphasize the importance of marriage and explore ways that government and private organizations can increase marriage rates.

The third element of the reauthorization strategy should be to expand the importance of faith-based and other community organizations in conducting the nation's welfare programs. National policy should reduce the barriers to faith-based organizations' ability to provide services to the poor, actively encourage their involvement in doing so, and provide for the careful evaluation of the effectiveness of faith-based services.

Stenghtening the Work Agenda

Funding. It is likely that the most contentious issues in the 2002 reauthorization debate will be over money. The TANF block grant is funded with $16.5 billion per year, plus about $1 billion in extras. The controversy will be initiated by proposals to cut TANF funds by between $3 and $5 billion per year. The justification for the cuts will be based on the fact that the TANF caseload is down by well over 50 percent in most states.

This reasoning is faulty on several counts, however. First, even though the welfare rolls have declined, state expenditures on work activities are necessary to keep the rolls down. New recipients are continuously coming onto the rolls, and former recipients often return. If states abandon their work programs, the rolls will begin to grow again.

Indeed, states should actually expand their work programs. Here, an irony of the 1996 legislation becomes apparent. Republicans who drafted the strong work requirements envisioned welfare reform as a program in which the majority of recipients would be in work programs. Almost as an afterthought, Congress allowed caseload reductions to count toward fulfilling the work requirements. For example, if, in 2002 a given state had experienced a 45 percent caseload reduction since the baseline year of 1995, the work requirement for 2002 would be only 5 percent (50 percent minus 45 percent). Because so many recipients have left the rolls since 1996, the caseload reduction credit means that the original Republican goal of engaging most welfare recipients in work has not been realized in most states. Indeed, many states, in effect, have a "zero" work requirement because their rolls have declined by 50 percent or more since 1995. Even so, in part because of the more generous state income disregard rules, over 30 percent of the people on welfare are now working, up from under 10 percent in 1995.

The logic behind requiring people on welfare to work is still compelling, and Congress should consider requiring states to have a certain percentage of their actual caseload in work programs, regardless of the extent to which their rolls have declined. This approach, although violating the criterion of state flexibility that infuses the 1996 law, would also ensure that states actually establish work programs for welfare recipients-a task that many states have not yet tackled. Experience with these programs clearly shows that they are expensive, complex, and difficult to administer properly.31

More importantly, budget cuts are based on a misunderstanding of the goals of the TANF program. Whereas the major goal of AFDC was to provide cash to destitute families so they could raise their children at home, providing cash is only one of the four goals of TANF. The other goals are to fight welfare dependency by promoting work and marriage, reduce illegitimacy rates, and increase the proportion of children living in two-parent families. To date, states have implemented aggressive programs to require work, but implementation of the family formation goals of TANF is spotty at best.

States need TANF money because they are just beginning to develop programs addressed to the family formation goals. Every state has activities designed to reduce illegitimate births and births to teenagers, but many states have minimal programs in place. Even more to the point, very few states have programs specifically aimed at promoting marriage. When states begin to develop such programs, they will need money to finance their efforts.

States also need money to prepare for the inevitability of a recession, which could increase their caseloads again. There is certainly no reason to panic over the impact of a recession on state funds, but prudence is the best course. Because the United States has so little experience with entering a recession with such a small percentage of the population drawing cash welfare benefits, it is impossible to predict how many people might return to the rolls. But consider, for the sake of argument, that the typical state caseload is now at about 50 percent of its highest level in recent years-the level on which its TANF funding is based. If the number of adults who return to the rolls equals half the current caseload-a high level that seems unlikely in most states-states would then be at 75 percent of their historic high level. In this unlikely scenario, states would still have a 25 percent surplus, when compared with the cost of their historic high caseload level. This example highlights what I believe to be the crucial consideration for the way that states handle recessions-namely, their ability to take money out of work support programs at a time when there are fewer workers and to make these funds available for cash benefits. If, however, Congress begins cutting TANF funds, states will not have the money needed to run a cash welfare program and a work program simultaneously during a recession.

The most important reason for not cutting TANF funds is that federal block grants are the wave of the future. Federalism has long been a tenet of Republican philosophy. According to this view, it is better to locate responsibility for developing and conducting programs primarily at the state and local levels than to try to manage them from afar. In my view, the nation would be better off if food stamps, Medicaid, and child protection programs were converted to block grants, with some mechanism in place that would allow for changes in funding levels to meet changing conditions. There is already growing bipartisan support for a child protection block grant; in addition, there is some support for a food stamp block grant-or at least more support than there was when Republicans proposed such a grant in 1995 and 1996. But if Republicans preside over cuts in the TANF block grant, then support for future block grants will be greatly reduced. Liberals will argue, with considerable justice, that Republicans are creating block grants only to cut them later. States will agree with this argument and withdraw their support for block grants. Also, if TANF is cut, even moderate congressional Republicans will begin to question block grants. For the sake of creating future block grants-thereby extending the federalism agenda-funding for the TANF block grant should not be cut.

Food Stamps and Medicaid. The food stamp program is simply incompatible with TANF. Food stamp benefits are entirely federally funded, but the program is administered by states. To reduce errors, the federal government conducts a quality control program that fines states with high error rates. Unfortunately, because it is difficult to keep benefit calculations in line with current income, the food stamp cases that cause the most errors are those in which someone in the household works. It would be hard to imagine two programs that are more incompatible: The TANF program imposes fines on states that do not reach mandatory work requirements, while the food stamp program, in effect, fines states that have high proportions of workers in their caseloads.

A second problem with the food stamp program is that many eligible families who are not on cash welfare do not receive the benefit. As the Urban Institute's Sheila Zedlewski has shown, over half of the families leaving welfare do not receive the food stamp benefits for which they are qualified. Working mothers earning $10,000 per year or less need the $2,000 in food stamps that their income would allow.32 Studies have shown that lack of knowledge about eligibility for food stamps and the administrative hassle involved in applying for and maintaining eligibility are major reasons for the low level of receipt.33 And in some states it is especially difficult for single working mothers to make the repeated office visits necessary to maintain food stamp eligibility.

Congress should consider a host of food stamp simplification procedures, including presumptive eligibility for families leaving welfare as well as continuous eligibility periods of six months. Even more importantly, Congress should allow three or four states to accept their food stamp money as a block grant and try to develop more efficient and effective mechanisms for providing food stamp benefits-or something equivalent-to working families.

Floundering Families. No policy produces all winners. In the case of female-headed families, Census Bureau data show that in 1999, the income of as many as 700,000 families in the bottom fifth of the income bracket had fallen, compared to their counterparts in 1995.34 Although poverty has dropped substantially, there are some families at the bottom of the income distribution who are worse off now, and the reason seems clear. The TANF program was designed to attack the problem of long-term dependency directly by requiring work, imposing sanctions on those who did not work, and creating a five-year time limit on dependency. Under these rules, families who could not work or refused to work suffered a reduction in benefits. In fact, in more than thirty-five states, welfare benefits for such families could be terminated. In these states, not everyone who left the welfare rolls did so in order to work. Several studies show that at any given moment, about 60 percent of those who have left the rolls are employed; over a period of several months, around 80 percent are employed. This leaves approximately 20 percent who are dependent for income on others. Some of these individuals who remain dependent on others for their income, and some who cannot hold jobs, are worse off than they were on welfare.35

In addition, there are still families on the caseload who have characteristics (addictions, depression, domestic violence, low education level, little job experience) associated with poor success in the job market. When these families reach the five-year time limit, some of them will have difficulty finding and holding jobs. States can exempt up to 20 percent of their caseload from the five-year time limit, so most states are now developing procedures to identify the adults who are unable to support their families and are therefore eligible for the exemption.

A major problem with the exemption is that many researchers, welfare administrators, and policymakers believed in 1995 that most of the mothers on welfare were incapable of holding full-time jobs. The data reviewed above, especially the employment data on never-married mothers, show that this position is simply wrong. A major strength of the 1996 reforms was that they communicated the clear message that everyone is expected to work. However, as the five-year time limit approaches (the limit will have been reached in every state by the summer of 2002), the prospect of substantial increases in destitution underscores the importance of addressing the problem of floundering families. In devising policies to help these floundering families, states need not only to find ways to maintain the pressure on every family to work and achieve as much independence as they can, but also to provide additional help-including cash welfare, if necessary-to families who fail. Thus, the five-year time limit should remain in place. In addition, the 20 percent exemption should not be changed unless it can be shown that in some states more than 20 percent of families cannot support themselves. At this point in time, no one has presented data that suggest that 20 percent is too low. Indeed, informal communication with state welfare directors suggests that very few believe that 20 percent of their caseload will reach the five-year limit.36 Even so, there are families who still have great difficulty supporting themselves. The task of helping them is rendered even more difficult by the fact that so little is known about them. The correct approach under these circumstances is for the federal government to take action that encourages states to design and implement innovative programs, and then to evaluate them in the search for successful models.

Job Advancement and Retention. The history of federal, state, and local employment and training programs is not very encouraging.37 Of the many reasons for the failure of training programs, one of the most important is that, historically, the jobs for which training was being offered were not necessarily available in the local economy. Moreover, many participants were placed in education and literacy programs on the grounds that a high school degree or its equivalent would make them more employable. But it was probably unwise to expect that young adults who had largely failed in school as youngsters would suddenly decide that school was important. Whatever the cause, evaluations have consistently shown that education and training programs produce either small impacts or no impacts at all.38

Despite this somewhat discouraging history, the need for quality education and training programs is greater than ever. There are now at least two million employed or recently employed mothers who, in previous years, would have been on welfare. The problem is that these mothers have jobs that pay only around $7.00 per hour. Moreover, previous research has shown that women on welfare who begin their careers with low-wage work make very little progress over the years. If the nation wants to help welfare mothers improve their income, it must find better ways to help them acquire more education and training.39 The congressional debate on this issue will, in all likelihood, be misleading because members will simply assert that if education and training are made available, all will be well. But both experience and research show that simply providing education will have only a modest impact.

Two education and training strategies make sense at this point. First, states and localities should mount programs in which they cooperate with low-wage workers and educational institutions to arrange courses that lead to specialized short-term training or even two-year degrees. It is essential that the training be relevant to available employment in the local area. Educational institutions, especially junior and community colleges, must be flexible in setting the length of courses and the times at which they are offered in order to meet the needs of single working parents.

The second strategy is to work with employers to identify the types of training necessary for low-wage workers to advance. Some programs of this type already exist, but many more are needed. Employers in need of skilled workers are often amenable to arrangements of this sort and may even pay part of the training costs or grant their workers release time for training. In some cases, it may be possible to work both with employers, who can identify the types of training and the skills necessary for advancement, and with local post-secondary institutions that can provide the training.

These arrangements will not be easy. Educational institutions are often rigid; employers always have their eye on the bottom line; and recipients may lack motivation or have a difficult time juggling work, child rearing, and educational activities. Nonetheless, programs have the potential to pay for themselves and to help many former welfare recipients escape poverty and join the middle class.

The reauthorization legislation should establish a pot of money, perhaps $50 million per year for five years, to support education and training programs similar to those outlined above. States should compete for the money and should be required to use their own TANF or state funds to pay for at least half of the cost of these programs. Congress should also fund a high-quality evaluation to determine whether the programs produce worthwhile impacts on skills, job advancement, and income.

Family Formation

As with the work agenda outlined above, the 1996 reforms took the appropriate direction with regard to family formation. Although the reforms aimed at reducing non-marital births have not had the same sweeping impacts as the work reforms, there is no question that the dangerous trends of illegitimacy and family dissolution of the 1960s, 1970s, and 1980s have been partially neutralized. It is reasonable to conclude that the 1996 reforms have played at least some role in this progress. Even more importantly, it makes sense to argue that, because the measures promoting marriage have not been aggressively implemented by states, more widespread and forceful implementation by states and localities-combined with active participation by the private and faith-based sectors-may further progress on this issue.

Consider the 1996 reform establishing the abstinence education program. Having worked in Washington for many years, I can testify that the reaction to the abstinence message has changed dramatically since 1996. Previously, the idea of teaching the hopelessly old-fashioned idea that fifteen-year-olds should not be having sex was viewed as a joke.

The reasons for this attitude change may not be clear, but the fact that key public officials have spoken out against single parenthood has certainly played a role. So has the national movement to have junior and senior high students sign abstinence pledges. So have the improved methods and growing use of birth control. So have the TANF work reforms themselves, as today's young women have watched their mothers work rather than simply draw welfare benefits. And welfare reform's abstinence education program could now be one of the most important fronts in the war to convince adolescents to wait not only to have babies but to have sex.

The goal of the abstinence education provision was clear and simple: to give states money to establish local programs that would send a clear and consistent message that sex outside of marriage is wrong and that society expects abstinence from its young people. Now, five years later, every state except California has accepted the funds and has established programs, largely in the public schools, to send the unambiguous message that abstinence is the standard for adolescent behavior. An excellent random assignment evaluation of the program is underway, although the chances that the evaluation will show impacts on sexual attitudes or behavior has been reduced somewhat by the refusal of several successful abstinence programs to participate. Nonetheless, a total of eleven programs are being evaluated, five by experimental design. Data from the experimental part of the evaluation-which will likely be published in the spring or summer of 2002-will be the first published data from a random-assignment experiment on the impacts of abstinence programs.

Even so, many of those who previously argued that the genie could not be returned to its bottle now claim that abstinence is the best message to send to children. Even many safe-sex advocates now realize the moral imperative of the abstinence message, especially because the only completely effective protection against sexually transmitted diseases is abstinence. One suspects that abstinence is the message that most parents favored all along, but only now do they find social support for their convictions.

Similarly, because of the 1996 reforms, every state now has effective programs to promote paternity establishment and collect child support. The paternity establishment program has been exceptionally successful, primarily because states have approached fathers in the hospital when their babies are born and have asked them to voluntarily acknowledge paternity. Every state also has declared goals for illegitimacy reduction and has presented plans for how those goals are to be achieved.

According to this data, states are at least thinking about ways to reduce non-marital births, and a few have been aggressive in implementing programs. Now, the key is to increase the level of activity. The 1996 legislation included a provision that offered cash bonuses of up to $25 million per year to states that reduce both their illegitimacy and abortion rates. The bonus has been successful in getting some states to determine how they might be able to reduce non-marital birth rates and institute new programs. However, after two rounds of awarding bonuses, it is clear that there is no obvious relationship between developing new state programs and winning the bonus. In fact, the most important factor in determining which states win the bonus appears to be the racial composition of the state population. For reasons that are not well understood, blacks have enjoyed greater reductions in non-marital birth rates than any other group in recent years. Thus, states that have high percentages of blacks are at an advantage.40

Because the illegitimacy reduction bonus does not appear to be producing the intended effect of stimulating state and local programs, the criteria for bonus payments should now be directed toward rewarding states that have the highest percentage of their youth or at-risk youth in prevention programs. The greatest need now is for program expansion; Congress can determine which programs are effective, either during the emphasis on expansion or after most states have substantially increased their focus on pregnancy prevention programs.

While maintaining the focus on reducing illegitimacy, Republicans should also expand the family formation agenda to include a much greater emphasis on marriage than they did in 1996. The fact that so many poor children are being reared apart from their fathers simply must become a greater focus of national attention. As Assistant Secretary of Health and Human Services (HHS) Wade Horn and Brookings Institution senior fellow Isabel Sawhill have shown, in 1960 about ten million children lived apart from their fathers.41 Today, the number is over twenty-four million, nearly four of every ten children. About one million children arrive at this status each year through divorce; the rest (about 1.3 million), through birth outside of marriage. A large and growing body of empirical evidence on divorce, illegitimacy, and marriage shows unequivocally that both adults and children are better off living in married, two-parent families.42 For example, children being reared by single parents, compared to those with two parents, are four times more likely to live in poverty.43 Indeed, Sawhill has shown that the increase in poverty since 1970 can be attributed almost entirely to the increase in the percentage of children living in female-headed families.44 Children who grow up without fathers are also more likely to fail in school or drop out, to have emotional or psychological problems, to engage in early sexual activity, to become pregnant, and to have drug and alcohol problems.45

Given the facts showing that marriage is vital for adults and children in families, the promotion of marriage should be a central goal of the next phase of welfare reform. However, as with the issue of illegitimacy in 1996, there are virtually no programs that have conclusively been show to promote marriage effectively. On the other hand, few programs that explicitly attempt to encourage marriage have been implemented. There are many ideas that should be put to the test, but there are at least three policies that should be supported by Congress during reauthorization.

First, the Earned Income Tax Credit (EITC) contains a marriage penalty that causes many couples to lose EITC income if they get married. This problem is aggravated because, for those in the income range of about $15,000 to $20,000, other benefits are also phasing out, and taxes are phasing in. This burden on marriage could be alleviated, however, by starting and ending the EITC phase-out range at a higher income level for married couples.

Second, states should end all forms of discrimination against married couples in their welfare programs. In fact, it would be especially interesting to grant preferences to married couples in some types of programs. For example, if housing units and housing voucher programs could reserve some portion of their benefits-say 30 percent-to provide a preference for married couples, then public policy would begin to send a clear message that marriage is a preferred status. In addition, public housing programs need more married couples as residents in order to provide role models for the children who live there.

Third, Congress should use some of the money currently being used for state performance bonuses to establish a fund for states that want to try new programs to increase marriage rates among the poor. Recent research shows that half of the couples having a first child outside marriage are cohabiting at the time the child is born. In 30 percent of these cases, the parents stated that they were involved in a committed relationship.46 If 80 percent of couples having babies without being married are committed to each other and to the baby at the time of birth, it makes sense to help them move their relationship toward marriage. States should be encouraged to help these couples by providing job assistance to both parents, counseling (if necessary), and perhaps even financial inducements to marry. Poor and low-income couples say they want to marry, and research shows that both they and their children will be better off if they do. Wise public policy efforts would seize this moment-when the parents are in a loving relationship and are deeply committed to their baby-to help the couple move toward creating a permanent bond. Research also shows that without such assistance, within a year or two, most of the couples will separate, one or both will move on to new relationships, and the father will have greatly diminished contact with his child.47 Everyone loses, and the trajectory of the child's development falls. In this case, Congress should offer matching funds to states wishing to promote marriage, employment, and better parenting among these couples. Congress should also appropriate funds to conduct careful evaluations of the state programs.

Faith-based Initiatives

As with both the work and family formation agendas, the 1996 reforms blazed yet another trail for American social policy. The Charitable Choice provision placed religious organizations on a level playing field with governmental and other private-sector organizations in conducting social programs that are financed by public funds. Under the terms of Charitable Choice, religious organizations can retain their religious character, including the practice of hiring people only of their own religion, and still receive governmental funds. The two major restrictions on the use of public funds by religious organizations are, first, that clients must have other choices if they prefer to avoid religious providers and, second, that governmental funds cannot be used to proselytize or support specifically religious activities.

The reason that Republicans, especially moderate Republicans, supported Charitable Choice was that they did not believe that the "wall of separation" between church and state was an absolute provision of the Constitution. In addition, they believed that a level playing field would increase competition between providers to deliver quality services. Charitable Choice represented a powerful combination: increasing competition and attracting a new, wealthy, committed, and potentially effective set of actors to the provision of social services. In addition, many Republicans and members of the American public believed then-and continue to believe now-that churches and other faith-based organizations can provide especially effective treatments because many poor and troubled Americans need precisely the hope and direction that faith can provide.

Federal policy should now expand the Charitable Choice provision to other federal programs, encourage more churches and other faith-based organizations to participate, and conduct evaluations to determine whether faith-based organizations can provide effective services. By establishing an office in the White House and five cabinet-level agencies, the Bush administration appears to be pursuing the first two goals. The hiring of John DiIulio, a distinguished social scientist and research advocate, to direct the White House Office of Faith-Based and Community Initiatives suggests the administration's commitment to research as well. As the State of Indiana has shown through its own office of faith-based initiatives, a governmental office that focuses specifically on reaching out to the community and offering technical assistance to faith-based organizations can increase the number of such organizations that are actually providing services to the poor. Although the Charitable Choice provision in the 1996 legislation has not yet produced a significant increase in the number of churches involved in helping the poor, the Bush initiative, if pursued vigorously, may substantially increase the number of faith-based services.48

To encourage states to expand their faith-based initiatives, Congress should require every state to list, in the state plan required under TANF, how it plans to involve faith-based organizations in its poverty programs. The reporting requirement should include four elements: 1) an overview of steps taken to ensure that faith-based organizations are aware of the opportunity to compete for governmental funds; 2) a description of the types of services and other responsibilities that faith-based organizations will provide; 3) a list of changes in these state policies since the last reporting period; and 4) a detailed account of the types of evaluations in place. In keeping with the devolved nature of the TANF block grant, states should not be required to conduct any of these activities, only to report on those activities they choose to conduct. Promoting the involvement of faith-based and other community organizations in fighting poverty must be pursued by administrative agencies and by the president's use of the bully pulpit. The major task now is to encourage faith-based organizations to become involved.

The Bush administration and others who are trying to promote the involvement of faith-based organizations should take action to promote the use of vouchers. In 1990, Congress enacted childcare legislation requiring every state to offer parents vouchers as part of their overall strategy to provide childcare. Recent studies have shown that, across the nation, parents are using vouchers to make childcare arrangements, including the use of informal facilities in neighborhoods and care by relatives.49 Perhaps one-third of this care is provided by churches. If parents use vouchers to select faith-based services, the constitutional issues of church and state are minimized. Although it may not be possible to use vouchers in all cases, an emphasis on parental choice wherever possible is a wise course, especially since, in addition to avoiding constitutional concerns, this practice reinforces a major purpose of all social programs-namely, to emphasize self-reliance and personal responsibility.

A word of caution is in order here. One of the most frustrating aspects of the War on Poverty and subsequent social programs has been the willingness of policymakers, scholars, and the media to believe, on the basis of flimsy evidence, that social programs are effective. Since the 1960s, the science of program evaluation has emerged, showing clearly that most social programs do not work. Indeed, the noted scholar and evaluation guru Peter Rossi has asserted, in his "Iron Law of Evaluation," that the expected impact of any given social program is zero.50 The repeated conclusions of these studies have shown that there are few differences between experimental and control groups; this has been a bitter pill to swallow for liberals who defended the programs even when there was no evidence of their effectiveness or ineffectiveness. Now it seems that conservatives may commit the same mistake.51 Those who make decisions about the investment of public funds should not accept conservatives' claims about the effectiveness of faith-based programs until evidence from reliable evaluations exists to support their claims. Fortunately, further research in this often-neglected area of study is on the way.

Conclusion

The 1996 welfare reforms have created a solid foundation for helping poor and low-income families who are willing to help themselves. The work agenda has played a major part in leading at least two million single mothers into the labor force and producing the most substantial reductions in child poverty since the 1960s. The family formation agenda, although only partially implemented across the nation, has nonetheless helped to neutralize growth in the rate of non-marital births. The incipient faith-based agenda is bringing new actors to the business of reducing dependency and increasing the practice of virtue. If the nation holds fast to these three agendas and promotes their expansion as part of the next phase of welfare reform, compassionate conservatism holds the promise of achieving the social policy goals that have been so passionately pursued by both liberals and conservatives.

Notes

1. Lawrence M. Mead, Beyond Entitlement: The Social Obligations of Citizenship (New York: Free Press, 1986) and The New Politics of Poverty: The Non-Working Poor in America (New York: Basic Books, 1992).
2. Charles Murray, Losing Ground: American Social Policy, 1950-1980 (New York: Basic Books, 1984).
3. Mickey Kaus, The End of Equality (New York: Basic Books, 1992).
4. Lawrence M. Mead, Beyond Entitlement: The Social Obligations of Citizenship (New York: Free Press, 1986).
5. Committee on Ways and Means, House of Representatives, Green Book: Background Material and Data on Programs within the Jurisdiction of the Committee on Ways and Means (Washington, D.C.: GPO, 1998), section 7.
6. Lee Rainwater and William L. Yancey, The Moynihan Report and the Politics of Controversy (Cambridge, Mass.: M.I.T. Press, 1967).
7. Daniel P. Moynihan, The Negro Family: The Case for National Action (Washington, DC: Department of Labor, 1965).
8. Stephanie J. Ventura and Christine A. Backrach, "Non-Marital Childbearing in the United States, 1940-1999," Vital Statistics Report 48, no.16 (2000).
9. Charles Murray, "The Coming White Underclass," Wall Street Journal, 29 October 1993, p. A14.
10. Ron Haskins, "Does Welfare Encourage Illegitimacy?" American Enterprise 7, no. 4 (1996): 48-49; Charles Murray, "Family Formation," in The New World of Welfare, ed. Rebecca M. Blank and Ron Haskins (Washington, DC: Brookings Institution, 2001).
11. Ron Haskins, "Welfare Reform and Illegitimacy: Republican Policies Designed to Reduce Illegitimacy" (unpublished manuscript, 2000); Charles Murray, "Family Formation," in The New World of Welfare, ed. Rebecca M. Blank and Ron Haskins (Washington, DC: Brookings Institution, 2001).
12. Douglas Kirby, Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy (Washington, DC: National Campaign to Prevent Teen Pregnancy, 2001).
13. Rebecca Maynard et al., "Changing Family Formation Behavior through Welfare Reform," in Welfare, the Family, and Reproductive Behavior: Research Perspectives, ed. Robert A. Moffitt (Washington, DC: National Academy Press, 1998).
14. Congressional Budget Office, CBO Memorandum: Federal Budgetary Implications of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (Washington, DC: GPO, December 1996).
15. This estimate was made by analysts at the Social Security Administration based on the definitions and procedures for SSI application and qualification arrived at through negotiations between, on the one hand, the House Ways and Means Committee and Congressman Jim McCrery, and, on the other, officials representing the Social Security Administration. The negotiations took place in the fall of 1996 after the enactment of the welfare reform law.
16. Committee on Ways and Means, House of Representatives, Green Book: Background Material and Data on Programs within the Jurisdiction of the Committee on Ways and Means (Washington, DC: GPO, 2000), section 8.
17. Office of the President, Economic Report of the President (Washington, DC: GPO, 2001).
18. Christine Devere et al., Welfare Reform Research: What Have We Learned since the Family Support Act of 1988? (Washington, DC: Congressional Research Service, 2000).
19. Census Bureau, Poverty in the United States, Series P60-210 (Washington, DC: GPO, 2000), C1-C9.
20. Martin Gilens, Why Americans Hate Welfare: Race, Media, and the Politics of Antipoverty Policy (Chicago: University of Chicago Press, 1999), chapter 3.
21. Robert Rector and William F. Lauber, American's Failed $5.4 Trillion War on Poverty (Washington, DC: Heritage Foundation, 1995).
22. Sheila Zedlewski et al., Potential Effects of Congressional Welfare Reform Legislation on Family Incomes (Washington, DC: Urban Institute, 1996).
23. Census Bureau, Poverty in the United States, Series P60-201 (Washington, DC: GPO, 2000), 28-33, definition 14.
24. Irwin Garfinkel, "Assuring Child Support in the New World of Welfare," in The New World of Welfare, ed. Rebecca M. Blank and Ron Haskins (Washington, DC: Brookings Institution, 2001).
25. Lisa Gennetian and Cynthia Miller, Reforming Welfare and Rewarding Work: Final Report on the Minnesota Family Investment Program, vol. 2, Effects on Children (New York: Manpower Demonstration Research Corporation, 2000).
26. Subcommittee on Human Resources of the Committee on Ways and Means, Child Support Enforcement and Supplemental Security Income 1995, 104th Congress, 1st session, 13 June 1995.
27. According to analysts at the Department of Health and Human Services, states have been highly successful in establishing paternity for current non-marital births through the use of voluntary acknowledgments by fathers in the hospital at the time of birth. This success, combined with far more aggressive efforts by states to establish paternity for children on welfare