Ending Welfare May 1995



Jeffrey Lehman and Sheldon Danziger

May 1995

During the 1992 Presidential Campaign, Candidate Clinton promised, in Putting People First, “to make work pay” and to “end welfare as we know it”:

It's time to honor and reward people who work hard and play by the rules.  That means ending welfare as we know it — not by punishing the poor or preaching to them, but by empowering Americans to take care of their children and improve their lives.  No one who works full-time and has children at home should be poor anymore.  No one who can work should be able to stay on welfare forever.

Shortly after taking office, President Clinton created a Welfare Reform Task Force to translate the campaign rhetoric into draft legislation.  The Task Force interpreted its mandate to be to craft a reform of the program that most people know as “welfare” — Aid to Families with Dependent Children (“AFDC”).  The reform was expected to resonate with “the basic American values of work, family, responsibility, and opportunity.”

Welfare reform debates have always been, at least implicitly, about the four values invoked by the Task Force.  Since AFDC was first created by the Social Security Act of 1935, each generation has changed the program to reestablish its understanding of what is required to respect those values while providing cash assistance for the “truly needy.”  Each round of statutory amendments has recalibrated the balance among (i) the interests of needy single parents, (ii) the interests of needy children, and (iii) the interests of the larger society in expressing its commitment to all four values.  To be sure, it is not easy to forge a legislative consensus (much less a societal consensus) on how the balance should be recalibrated.

President Clinton made deficit reduction his top domestic priority in 1993 and universal health coverage his top priority in 1994. Thus, despite a good deal of discussion concerning welfare reform, the first two years of the Clinton Administration ended without either House of Congress giving sustained committee consideration to any welfare reform bill. Several bills were introduced, however, and their titles emphasized at least three of the four key values -- work, family and responsibility. The most significant bills included the Administration’s “Work and Responsibility Act” (hereafter the Clinton Plan), a more liberal bill, the “Family Self-Sufficiency Act of 1994” introduced by Representative Matsui (D-CA) and nineteen cosponsors, and a more conservative bill, a “Welfare Reform Act of 1994” introduced by Senator Faircloth (R-NC) and Representative Talent (R-MO) (“Faircloth-Talent”). In September 1994, House Republicans unveiled “The Personal Responsibility Act,” as a major component of their proposed “Contract with America.”

The fourth value -- opportunity -- which was the featured value during the War on Poverty/Great Society era (the 1964 Economic Opportunity Act) does not appear in the title of any of these bills. As we discuss below, the bills differ fundamentally in their approach to opportunity. The more liberal bills assume that significant barriers to opportunity remain for welfare recipients, whereas the more conservative bills assume that opportunities are available but that the current welfare system encourages recipients to forgo them.

The Republican Congressional majority in the 104th Congress placed welfare reform at the center of its Contract with America and made it an important item on the 1995 legislative agenda.  And, in March 1995, the House passed The Personal Responsibility Act and sent it on to the Senate.  The debate has been cast in terms of all four of the key values of work, family, responsibility, and opportunity. Is it likely that these values could be translated into a welfare reform plan that would produce a better world than the one which would exist if the status quo continued unchanged? This simple question raises a number of difficult normative issues about how individuals and governments ought to behave. It also raises difficult empirical questions about the consequences that individual actions and government programs bring about. While we cannot resolve all the normative and empirical issues, we can offer an analytic framework within which to consider them. In this article, we highlight some of the relevant normative commitments that provide the principal motivations for the 1995 welfare reform debate and we set forth some of the background empirical and analytic considerations that are relevant to this round of difficult collective self-definition.

This article is organized as follows. In the next section, we briefly describe how welfare operates today. Then we describe the current economic context and its implications for welfare reform. The two subsequent sections focus on the four values, first work and opportunity, then family and responsibility. We consider how those values are expressed in existing law, and how they would be reconceptualized under proposed reforms. The final section concludes that welfare reform has the potential, if properly implemented, to make things better than they are now. However, given the current political environment in Washington, we doubt that the welfare reform proposal that is likely to emerge from Congress will do so.  Indeed, if the House-passed version of the Personal Responsibility were adopted by the Senate as well, we would like the President to veto it, as worse than no reform at all.


Aid to Families with Dependent Children is an income support program that responds to immediate financial hardship.  It embodies a commitment to support a subgroup of the poor that was, at one time, thought blameless: low-income families with young children and a missing or financially incapacitated breadwinner.  To qualify for benefits, a family must show that it has virtually no assets, that it has very low income (each state sets its own eligibility ceiling), and that a child in the family is deprived of at least one parent’s support because the parent is (a) not living with the child, (b) incapacitated, or (c) a recently unemployed primary breadwinner.

AFDC is primarily a program for single mothers and their children.  A few single fathers participate, and a somewhat larger number of two-parent families satisfy the more stringent requirements for two-parent eligibility.  But among the roughly 5 million families receiving AFDC benefits in a typical month in fiscal year 1993, about 90 percent were fatherless.

AFDC has two aspects: a safety net aspect and a transitional aspect.  Although the issue of time-limits has figured prominently in welfare discourse since the 1992 election, it is already the case that half of all families that begin a welfare spell leave the rolls within one or two years.  For those families, AFDC ensures a meager, but potentially vital, safety net.  In 1994, a nonworking welfare mother with two children and no earnings received $366 in cash and $295 in Food Stamps in the median state, or about 69 percent of the poverty line.  Importantly, AFDC recipients qualify for family health insurance in the form of Medicaid.

The transitional aspect of current law is embodied in JOBS, the Job Opportunities and Basic Skills Training program created by the 1988 Family Support Act. At present (ignoring special state options, and ignoring the special requirements imposed on teenage high school dropouts), AFDC is a program of three-years-per-child-and-then-participate-if-you-can. Once the mother’s youngest child reaches age 3, she must participate for up to 20 hours per week in JOBS. Once that child reaches age 6, she may be required to participate for up to 40 hours per week.

Participating in JOBS means agreeing to a reasonable “employability plan” the state devises, as long as the state provides for child care, transportation, and other work-related expenses. However, if the state has not appropriated sufficient funds to provide a JOBS slot (and many states have not), the recipient is not punished for the state’s failure. Any recipient who complies with legitimately-imposed JOBS requirements continues to receive a welfare check. Any recipient who fails to comply, without good cause, may be sanctioned by having her monthly grant reduced to reflect a family with one fewer person.             As of 1992, states were (on average) providing JOBS slots for only about 16 percent of their non-exempt participants or about 7 percent of all adult welfare recipients. Under current law, each state will have to place at least 20 percent of non-exempt participants during fiscal year 1995 or face the prospect of losing some federal funds. Many welfare reform bills would expand this participation requirement.

This transitional aspect of AFDC currently imposes no time limits on its safety-net aspect. Recipients may enter AFDC, enroll in JOBS, find a job, lose that job, return to the rolls, and re-enroll in JOBS. It does, however, incorporate the notion of mutual responsibility by which recipients are expected to take advantage of training and work opportunities provided by the government.


One of the most significant changes in America’s welfare programs over the past two decades is the decline in the level of cash benefits they provide.  Throughout that period, inflation  eroded the effective purchasing power of a welfare grant. Moreover, during the 1990’s, a number of states cut benefits in nominal terms.  Thus, in the early-1990s in the median state, the combined AFDC and Food Stamp benefit was about 70 percent of the poverty line for a nonworking mother with two children, down from about 85 percent in the mid-1970s. In the absence of reform, welfare benefits would most likely continue to erode and leave recipients increasingly distant from the poverty line.

The declining economic position of AFDC recipients is, to be sure, not unique.  The past two decades have been characterized by economic distress for the middle class, the working poor, and the unemployed, as well as for welfare recipients. There has been relatively little economic growth over the past generation, and the gains from growth have been very uneven.  In the two decades following World War II, “a rising tide lifted all boats.”  During economic recoveries most families gained -— the poor as well as the rich, less-skilled workers as well as the most-skilled.  During the 1980’s recovery, however, a rising tide became an “uneven tide,” as the gaps in living standards widened between the rich and the poor, between the rich and the middle class, and between the most-skilled workers and the least-skilled workers.

In America today, economic hardship is remarkably widespread.  Popular portrayals of economic hardship tend to focus on inner-city poverty or single-mother families or displaced factory workers, and associate poverty with their lack of work effort or lack of skills.  But during the 1980’s, inequalities increased within most broader groups across the population as well.  While white-collar workers fared better on average than blue-collar workers, and married-couple families fared better on average than mother-only families, many white-collar workers and many workers in married-couple families were also laid off or experienced lower real earnings.

Not even the most educated groups were spared.  To be sure, the average college graduate continues to earn substantially more than less-educated workers, and the earnings of the average college graduate grew much faster than the earnings of other workers in the 1980’s.  Nonetheless, a college degree no longer guarantees a high salary.  In 1991, among 25-to-34 year old college graduates (without post-college degrees), 16 percent of men and 26 percent of women worked at some time during the year but earned less than the poverty line for a family of four persons.

Because economic hardship is this extensive, one should be suspicious of claims that there are any simple solutions to the “welfare problem.” The “welfare problem” is one aspect of a broader “poverty problem,” which, in turn, is part of a broader problem of economic hardship -- slow economic growth, high poverty rates, and rising income inequality. In effect, the economy has generated new welfare cases faster than an unevenly growing economy and government policies can reduce the rolls.

The primary source of this increased economic hardship has been a set of structural changes in the way the labor market operates. Less-educated workers have found it harder to secure employment, and those who are hired tend to receive low wages. Many factors moved the economy in the same direction. The decline in the percentage of the work force that was unionized, reductions in the percentage that works in manufacturing, increased global competition and the consequent expansion of the import and export sectors all lowered the wages of less-skilled workers. The automation which accompanied the introduction and widespread use of computers and other technologial innovations also increased demand for skilled personnel who could run the more sophisticated equipment. Simultaneously, there was a decline in the demand for less-skilled workers, who were either displaced by the automated systems or had to compete with new imports.

These changes in the general structure of the labor market have important implications for the current welfare reform debate. Because most welfare recipients have limited education and labor market experience, the contemporary economy offers them fewer opportunities even when unemployment rates are low. Moreover, in many communities, the unemployment rate has exceeded 6 percent for most of the past 15 years; in many inner cities, the unemployment rate has remained well-above 10 percent for this period. And the shift in the skill mix required in today’s economy means that, even if an employer extends a job offer to a welfare recipient with low skills and experience, that employer is not likely to pay very much.

It is simply not the case that most of today’s welfare recipients could obtain stable employment that would lift them and their children out of poverty, if only they would try harder. Fear of destitution is a powerful incentive to survive; it will not, however, guarantee that an unskilled worker who actively seeks work will be able to earn enough to support her family. The harsh realities of today’s labor market mean that changes in welfare mothers’ economic incentives are unlikely to make much of a difference unless they are accompanied by changes in their economic opportunities.

As long as America remains committed to the view that a child should not have to live in poverty merely because his or her single parent is unemployable, debates about welfare reform should continue to be primarily debates about what kind of government intervention we would like to support.  Do we want to continue to support families outside the paid workforce? Or do we want to try to improve the labor market prospects for welfare recipients? How does placing a time limit on the receipt of welfare change society’s “work” expectations of recipients? How does it change government’s responsibility for protecting needy children?

These are not technical questions of policy science. They are questions about the kind of society in which we want to live. To answer them systematically, it is helpful first to consider what existing programs say about the kind of society we aspire to be. We can then consider what changes in those programs might make it easier for us to realize those aspirations.


The most widely discussed aspect of the current welfare reform debates is a two-year time limit -- the proposal that, after two years, an AFDC parent’s obligations would change so that she could no longer receive cash assistance in return for caring for her own child. There are, however, many possible definitions of what it means to impose a time limit on the receipt of cash assistance. These range from a strict “two years and out” rule in which no cash assistance or any other public support is available, to a “two years and work” rule in which some safety net is still provided to recipients, but where the form of assistance shifts from cash during the first two years to a minimum-wage job offer in subsequent years.

When it is suggested that a time-limit in the Clinton proposal or any other proposal would “end welfare as we know it,” the implicit claim is that such a change in the structure of AFDC would signal a radical shift in society’s expectations of single mothers.  It is useful to situate discussions of the relationship between AFDC and social attitudes towards paid work by single mothers in a broader historical context of legislative reform, because the evolution of AFDC since 1935 has reflected a steady change in the implicit understanding of what it means for a single mother to “work.”

In AFDC's early years, the implicit concept of work was linked to other markers of social status.  A stylized interpretation of conditions during the 1930’s and 1940’s might run as follows:  White widows “worked” vicariously through their late husbands and directly by maintaining a “suitable home” for their children.  Over time, more white divorcees and unwed mothers claimed welfare benefits; they “worked” by satisfying the “suitable home” standard and, if the caseworker thought they were capable, by accepting “appropriate” work for wages.  During that same time period, and especially in the south, black single mothers were expected to do whatever house or field work was demanded by local employers.  In all cases, the mother, through her “appropriate behavior” justified public support for the fatherless child.

During the late 1960's, the federal AFDC statute began to embody a different notion of what kind of work was required from single mothers in return for welfare.  In response to growing public dissatisfaction over the rising welfare caseload, one which coincided with a rapid increase in married white women's participation in the paid labor force, Congress amended the statute to provide greater economic incentives for maternal labor force participation and to provide that some women (although, admittedly, few at first) would be required to participate in work training programs.

Since 1967, the statutory expectation for workforce participation by single mothers has steadily expanded.  Traditionally, mothers of very young children were exempted.  But over time, the definition of a “very young” child has fallen from “under six” to “under three” (and at state option to “under one”).  At the same time, Congress has appropriated progressively larger amounts of money to fund state programs that attempt to move mothers off of welfare and into a job.

As we mentioned, the Family Support Act of 1988 emphasized training, education and work for AFDC recipients through the JOBS program. JOBS became a central feature of the FSA in part because of the favorable evaluations of many state “workfare” demonstration programs that were undertaken in response to the Reagan Administration’s emphasis on work. In the early-1980s, many liberals opposed workfare programs and considered them to be punitive. When these state programs were evaluated by the Manpower Demonstration Research Corporation (MDRC) during the 1980s, however, many were judged modestly successful in reducing welfare receipt and increasing earnings. More important for liberals was the finding that many participants found the programs to be fair and helpful in connecting them to the work force. The evaluation results were promising enough so that by the late-1980s, moving welfare recipients into employment had become a bipartisan “new consensus.”

Liberals and conservatives still disagreed on other goals for welare-to-work programs. Liberals thought welfare reform should offer opportunities for a welfare mother to receive training and work experience which would raise her family’s living standard through increased work and higher wages. Conservatives emphasized work requirements, obligations owed by a welfare mother in exchange for the government’s support, even if her family’s income did not increase.

The MDRC evaluations produced some results that satisfied both groups. In their comprehensive review, Gueron and Pauly (1991) found that successful programs made modest reductions in welfare dependence and encouraged some recipients to leave the welfare rolls. The early state programs primarily consisted of job search assistance and employment counseling. By the end of the 1980s, programs also provided recipients such services as subsidized child care and transportation, follow-up counseling after employment began, and education and training services.

As liberals came to accept work as the key goal for welfare reform, David Ellwood, in his influential book, Poor Support: Poverty and the American Family (1988), incorporated the new consensus, but extended it. He proposed converting welfare into a transitional system that would provide cash support for a limited period of time. At the end of the transitional period, a recipient would be expected to earn wages in a regular job or a work opportunity provided by the government. Low wages would be supplemented by expanded tax credits, access to subsidized child care and health insurance, and guaranteed child support. Ellwood’s proposal captured the attention of Candidate Clinton, who proposed to “end welfare as we know it” by imposing a two-year time limit. President Clinton appointed Ellwood as one of the co-chairs of the Welfare Reform Task Force that developed the Work and Responsibility Act of 1994.

Thus, contemporary discussions of a two-year time-limit might be viewed as an (untested) extension of the demonstration and policy lessons of the recent past.  On the other hand, current proposals might be seen as an attempt to accelerate the evolution of welfare by putting a very strict limit on the time during which single mothers may fulfill their societal responsibility “merely” by rearing their own children.  

One can capture some of the cultural stakes behind time limits with an analogy to the world of insurance.  The proposition that welfare should not be “a way of life” implies that the “premium” a family pays to society by rearing its own children is a limited one, one that will only allow it to collect a limited “insurance benefit” should it be struck by the calamity of poverty.  In other words, proposals to time-limit AFDC are  proposals to make AFDC more like time-limited Unemployment Insurance, and less like Social Security, whose benefits continue indefinitely. 

Whether or not one concludes that this program change is desirable will depend on one’s views regarding the opportunities available to a welfare recipient who comes up against the time limit. If one believes that opportunities are readily available to anyone who wants to work and that unlimited welfare benefits undermine the incentive to seek work, then a time limit should generate increased work and even an increase in a recipient’s income. Our view, however, is that, given the labor market changes of the past two decades, many welfare recipients and other less-skilled and inexperienced workers face limited labor market opportunities. As a result, a change in the understanding of a mother’s responsibility to “work” can only be made tolerable by a comparable change in the understanding of the government’s responsibility to provide some “opportunity” for those who actively seek to work but cannot find an employer who will hire them. Without such balanced change, a “two years and out” time-limit for single mothers would be both unrealistic and unfair.

Given the complexity of welfare, even a “balanced change” in expectations of and opportunities for welfare recipients would have complicated effects. The most important feature of the welfare caseload is that it is heterogeneous, so that opportunities may be available to some recipients, but unavailable to others. We find it useful to place welfare recipients into five stylized cases.

First, there are undoubtedly some recipients who are physically able, but unwilling to take available low-paying jobs. When a time limit is imposed, their benefits will be terminated. This will yield some savings on welfare expenditures, but cause hardship for their children unless the time limit causes them to accept the jobs they had previously rejected. There is no way to know how large a group this is, but our intuition is that it is very small. If we are correct, then both the budgetary savings and the heightened hardship that flow from a time limit would be small.

A second group of welfare recipients currently supplement their benefits (in violation of welfare’s rules) by working off the books in low-wage jobs. When a time limit is imposed, these families would probably leave the rolls as well. Their hardship could be cushioned to the extent that being off welfare would permit them to move to more visible, but higher paying employment.

A third group includes recipients who are ready to work and have the skills to be productively employed, but who have not taken jobs because of the work disincentives currently implicit in AFDC. This group is likely to consist primarily of mothers who have strong concerns about health insurance, or who live in high-benefit states. If their welfare benefits are terminated, they will presumably be able to get jobs and earn wages that would be supplemented by the Earned Income Tax Credit. They will incur child care expenses and health insurance expenses, so their disposable income may not increase.

A fourth group of recipients fall between the work expectations of the AFDC program and the Supplemental Security Income Program (SSI). SSI provides a safety net for those whose disabilities prevent them from working, but has a very strict eligibility criterion. What will happen to AFDC recipients who have health problems that make it hard for them to succeed in the marketplace, but who are not ill enough or disabled enough to qualify for SSI? Will they be offered work opportunities in sheltered workshops? Or will they be exempted from the time limit? Or will they simply lose all cash assistance? 

A final group includes parents who want a job but cannot find an employer to hire them. Our presumption, based on the recent labor market experiences of low-skilled workers and the MDRC evaluations of welfare-to-work demonstrations, is that this group would be quite large. Gueron and Pauly, for example, show that many program participants have difficulty maintaining stable employment after the programs end. What kind of work opportunities will be available for those who take the responsibility to actively seek work? Will jobs be provided or will they be expected to rely on family support or private charity? Will many of those who “play by the rules” and actively seek work find themselves without jobs and without a safety net?

We suspect that most Americans would find the changes in choices open to these different groups of welfare recipients more or less acceptable, depending on whether they believe that every family that comes up against the time limit has some mechanism by which it could pull its weight. Whether such mechanisms are adequate is the central question along the work-opportunity axis of welfare reform debates. But it is not the only such question. Because even if adequate work opportunities for welfare mothers exist, one must still consider the effects on children of requiring their mothers to accept those opportunities.

The empirical literature on this issue is inconclusive. Some children will benefit, and some will suffer, but there is no reason to be confident that overall the effects on children will be distinctly positive or negative. We are aware of no studies that consider the effects of different forms of child care (maternal or paid) on the children of welfare recipients. One can imagine that the two-year-old child of a disadvantaged welfare recipient might benefit from the stimulation of a day care center; one could as easily imagine that she might suffer from disruption in her intimate relationships. Ultimately, the effects on children will reflect both (a) the quality of the AFDC recipient child’s new day care environment and (b) the extent to which increased experience in the paid workforce provides the mother with a transition to a higher standard of living and with a set of life opportunities that make her a more successful parent.


The other axis of values that has long been central to welfare reform debates links the value of parental responsibility to the value of two-parent families.  Can welfare protect children from some of the economic costs of divorce without encouraging divorce?  Can welfare protect children from some of the economic costs of being born out of wedlock without encouraging nonmarital births?

Such questions have always been an important part of welfare policy discussions.  During the 1980’s and early 1990’s, however, a broad political consensus emerged in which the dominant reform goal became the challenge of maintaining a social safety net while fighting the alienation of welfare recipients from the paid workforce. Family structure was a real, but secondary issue.  The past twelve months, however, have seen a breakdown in the consensus, as conservative politicians rejected the work consensus and placed a concern with out-of-wedlock childbirth at the center of the welfare reform debate. Indeed, a major difference between the Clinton Plan and the Republican Plan is that the former emphasizes moving welfare recipients into work, whereas the latter emphasizes eliminating eligibility and benefits for children born out of wedlock.

Although there is disagreement concerning how welfare reform can support families, there is no disagreement that rapid changes in family structure have occurred. The number of young children who live with only one parent has skyrocketed since the early-1950s.  In 1960, only 9 percent of children under 18 lived with one parent, and less than 0.5 percent lived with a single parent who had never married.  In 1992, 27 percent of children under 18 lived with one parent, and 9 percent lived with a single parent who had never married.

Because AFDC assists low-income children in one-parent families, the demographics of recipient families have changed in tandem with the changes in society as a whole. In 1935, the typical AFDC family was headed by a widow; in the 1950s, by a divorced or separated mother. Since the mid-1980s, however, most AFDC-recipient children have lived with a never-married parent.

In the 103rd Congress, several legislators proposed denying AFDC benefits to children born out of wedlock. And such a proposal is a key feature of the Personal Responsibility Act as passed by the House. Just as Ellwood’s book provided the intellectual rationale for time-limiting cash welfare benefits, Charles Murray’s writings have provided the rationale for denying benefits to unwed mothers. In 1993, Murray published a Wall Street Journal editorial page column under the headline, “The Coming White Underclass.” The column has proven to have sufficient political importance to warrant a thorough discussion. As Mickey Kaus has pointed out, after Murray’s column appeared many Republicans abandoned the view that “work” was the primary welfare problem and adopted the view that “illegitimacy” was.

In “The Coming White Underclass,” Murray revived the polemical style that he had deployed in Losing Ground a decade earlier and more recently in The Bell Curve. He constructed an argument with eight structural characteristics:

(1) Murray presented a troublesome social fact.  In Losing Ground, the troublesome fact was the increasing rate of pre-transfer poverty.  In “White Underclass,” it was the increasing rate of out-of-wedlock childbearing.

(2) Murray presented the troublesome social fact in a variety of ways, using quantitative measures from several different data sets.

(3) Murray speculated in apocalyptic terms about the future implications of the troublesome social fact.

(4) Murray hinted darkly that the troublesome social fact had been concealed from the average American.  While “headlines” reported one thing, Murray suggested that the “real news” had been suppressed.

(5) Murray expressed his vision of society in quotable aphorisms.  “In the calculus of illegitimacy, the constants are that boys like to sleep with girls and that girls think babies are endearing. . . . Bringing a child into the world when one is not emotionally or financially prepared to be a parent is wrong.  The child deserves society's support.  The parent does not.”

(6) Murray offered a simple account of how the troublesome social fact could (in theory) have resulted from the rational responses of self-interested individuals to government social welfare programs.

(7) Murray insisted that the troublesome social fact would disappear if government disappeared (in this case, by eliminating many social welfare programs and denying an unwed mother any right to collect child support from the child's father). His proposal “does not require social engineering. Rather, it requires that the state stop interfering with the natural forces that have done the job quite effectively for millennia.”

(8) Finally, Murray offered assurances that the costs of his recommendation would be minimal because the 

world of private, voluntary exchange (families and charities) would be an effective substitute for the public safety net.  “How does a poor young mother survive without government support? The same way she has since time immemorial.”

An important part of what makes Murray's polemic effective is the clever way it baits academics.  For the structural characteristics that we numbered (3), (4), and (5) in the list above seem calculated to goad professorial critics into making analytically sound, but politically unpersuasive, criticisms. 

Consider an example.  In Murray’s argument, a key premise is that having a child out of wedlock is detrimental to both the mother and the child —- a premise that would meet little resistance with the general public and that would seem to be supported by data showing a correlation between nonmarital births and unfavorable measured outcomes.  To an academic reader of Murray, however, the claim evokes two responses.  First, observed correlations between out-of-wedlock childbearing and, say, poverty might be “spurious.”  Nonmarital births might not cause poverty. Rather, nonmarital births could be the consequence when young people grow up in impoverished surroundings and see little potential for escaping their conditions.  Alternatively, both nonmarital births and poverty might be caused by some other pernicious social force.   Second, even a supposedly-causal connection could be “contingent.”  In other words, even if out-of-wedlock childbearing is harmful to children under current conditions, it might not be so harmful if social programs or educational or economic opportunities were changed.

As a theoretical matter, these responses to Murray are completely sound.  Social science methods are too limited to provide uncontrovertible proof of social causation.  And social phenomena are virtually all contingent. Our point, however, is that, while such responses might expose theoretical weaknesses in Murray's argument, they do not present counter-evidence to demonstrate that the relationship between out-of-wedlock births and poverty is in fact spurious.  Nor do they demonstrate that American society could realistically be transformed to make the phenomenon benign.  For policymakers, the knowledge that a social fact might not be inevitably troublesome is worth very little, especially if Murray's “troublesome” thesis (if not the “apocalypse” thesis) resonates with most people’s intuitions about how the world works and is likely to continue to work.

It would be unfortunate if academic criticism of Murray’s argument got bogged down in the logical failings of the way he used characteristics (3), (4), and (5).   The danger is that the serious flaws reflected in characteristics (6), (7), and (8) of Murray’s argument would remain unexposed.  Accordingly, for purposes of discussion, let us stipulate that out-of-wedlock childbearing is a troublesome social phenomenon and that its recent rise is a troublesome social fact.  Let us even stipulate that government might consider replacing the programs of the War on Poverty era with Murray’s War on Illegitimacy.  The problem is that Murray has not even remotely begun to make the case for the idea that the first step in his War should be to deny unwed mothers access to the social safety net.

Murray suggests, first, that the rise in nonmarital childbearing was caused by the growth of the welfare state; second, that eliminating the welfare state would reverse the trend; and third, that the side effects of eliminating the welfare state would be tolerable. These three propositions are independent. Even if the first were true, it would not necessarily imply the second; and the second would not necessarily imply the third.

Unlike his quasi-empirical discussion of the fact and consequences of nonmarital childbearing, Murray’s discussion of the causes of the rise in nonmarital birth is purely theoretical. He attributes it to a change in the “calculus” of young boys and girls. He believes that the rise in nonmarital births has followed from a drop in their “costs.” And he believes that eliminating welfare would, directly and indirectly, raise those costs enough to lead young girls (and maybe even young boys) to act differently.

Why does Murray believe that welfare encourages births out of wedlock? Because, under existing law, two-parent families can receive AFDC benefits only if (a) one parent is incapacitated or (b) the primary earner was recently employed and is now unemployed. Thus, it is not surprising that in 1991 only about 11 percent of AFDC children qualified for the program while living with both parents. Indeed, the fact that AFDC treats single-parent families better than two-parent families has been a concern of policymakers since at least the early 1960s.

Here it is Murray who indulges in theoretically interesting but P.R.A.ctically irrelevant speculation.  As a matter of pure theory, Murray could well have been right that the structure of AFDC eligibility brought about the rise in out-of-wedlock births.  But it is just as easy to construct a story on the theoretical plane about why Murray’s account of the rise in nonmarital childbearing is completely wrong.  

The key point, ignored by Murray in “White Underclass” just as he ignored it in Losing Ground, is that merely knowing the direction of an economic incentive does not tell us anything about how big an effect the incentive actually has.  When it comes to the decisions to have sex, to bear a child, and to raise a child, a host of other factors can easily “dominate” or dwarf the effects of AFDC's benefit structure.

We do not need to resign ourselves to theoretical speculation, however, because social scientists have attempted to measure the effects of welfare's incentives on family structure. In a comprehensive review of the literature, Robert Moffitt considered the time-series data. He concluded, “the evidence does not support the hypothesis that the welfare system has been responsible for the time-series growth in female headship and illegitimacy.”  He then considered the econometric analyses of the effects of variations in the level of welfare benefits on the likelihood that a child lives with two parents. Moffitt concluded that, while some of the most recent studies had begun to show some evidence of a detectable effect on rates of female headship, the magnitude of the effect was small.  “The failure to find strong benefit effects is the most notable characteristic of this literature [on the relationship between welfare and female headship].”  

Summarizing the studies that looked specifically at the relationship between welfare benefits and nonmarital childbearing, Moffitt concluded that there was “mixed evidence” of any effect at all.

In addition to eliminating welfare, Murray also proposes to reduce young men’s and women’s interest in bearing children out-of-wedlock by denying unwed fathers any legal right to have a relationship with their children, and by eliminating any obligation of unwed fathers to support their children. We have already challenged Murray’s implicit assumptions about how such changes would affect young men’s and women’s perceptions of their own self-interest. Even more importantly, children who were born out-of-wedlock if his proposal were adopted, would find themselves with no claims for financial support from either the father who was one cause of their predicament or the state that insulated him from responsibility.

In contrast, for the past 20 years Congress has steadily expanded the federal role in child support enforcement. A domain that was almost the exclusive province of state law as recently as 1974 is today subject to a complex and pervasive umbrella of federal regulation. The system is designed to enforce financial responsibility, not only on the part of fathers who never marry, but also on the part of fathers who divorce. Under current federal law, each state is required to entrust its child support operations to a single agency. It must demand the Social Security numbers of both parents as a condition of issuing a birth certificate (unless it finds good cause for not doing so). It is given economic incentives to improve the technology used to establish paternity, and to increase the percentage of cases in which paternity is established (except in cases where doing so would be contrary to the best interests of the child).

Having established paternity, states are also required by federal law to establish uniform presumptive guidelines for child support awards. And in the case of AFDC recipient children, the state has a financial incentive to obtain support orders. Finally, the enforcement techniques for collecting child support, once it has been awarded, have been strengthened -- states must maintain parent-locator services; they must cooperate with one another; they must utilize Federal parent-locator services and even the federal courts, if necessary. Where necessary, states must use other enforcement mechanisms, including measures to withhold child support from fathers’ tax refunds and unemployment compensation checks, and to impose liens on fathers’ property. Since the beginning of 1994, states have been required to implement immediate withholding from fathers’ paychecks (regardless of whether the father is behind) for all new child support orders. Nonetheless, despite these efforts, billions of dollars of potential child support payments remain uncollected.

In his January 1994 State of the Union Address, President Clinton suggested that he would press ahead even further. He said that we should “say to absent parents who aren’t paying their child support, if you’re not providing for your children, we’ll garnish your wages, suspend your license, track you across state lines, and if necessary, make some of you work off what you owe. People who bring children into this world cannot and must not walk away from them.” We have no way of knowing to what extent such measures would reduce the frequency of out-of-wedlock births, but they would have symbolic value. In addition, a prominent researcher has concluded that such reforms can reduce welfare dependency and poverty by increasing the amount of child support collected.

We are not challenging Murray’s concern about the increasing nonmarital birth rate. Rather, we reject his suggestion that eliminating AFDC and absolving unwed fathers of any child support responsibility are the best ways to resolve this problem. Indeed, the research suggests that any beneficial effects from the small decline in nonmarital births that would be likely to occur would be greatly outweighed by the increased hardships that would have to be endured by the children who would still be born out of wedlock.

This conclusion, based on our reading of the empirical evidence, does not mean that we reject Murray’s assumption that young women, sometimes teenagers, are making rational decisions about whether to become single parents. Assume for the sake of simplicity that a young woman chooses between two options: (i) having a baby alone, and (ii) waiting until she has married. Assume also that there is a social concern that too many teens are now choosing option (i). Murray believes that if AFDC and access to child support were eliminated, option (i) would suddenly become substantially less attractive than option (ii). He is almost certainly wrong to think that the effects would be so dramatic that the “War on Illegitimacy” would suddenly be won. But that should not prevent us from considering other policy interventions that could make option (ii) more attractive than option (i). Indeed, there is some evidence, for example, that, holding other characteristics constant, women are less likely to have a child before they graduate from high school if they attach a higher economic value to the diploma.

Anyone seriously concerned about out-of-wedlock births should carefully consider two possibilities: (1) that narrow, focused changes in specific provisions of the AFDC statute might reduce the relative benefits (or increase the relative costs) of deferring childbearing, without significant attendant social harms, and (2) that other government interventions might raise the perceived “benefits” of deferring childbearing in ways that would influence behavior. For example, the government might subsidize higher education for any high school graduate who had not borne or fathered a child out of wedlock, or provide a guaranteed job, or do more to ensure that any “opportunity” provided for single mothers trying to get off welfare will be equally available to young women who avoided welfare by not having a child.

In sum, the statistical evidence fails to support Murray’s strong historical claims that the current “crisis of illegitimacy” resulted from the structure of AFDC.  It offers even less reason to believe Murray’s suggestion that we could dramatically reduce out-of-wedlock births by denying unwed mothers access to public support and by freeing unwed fathers of all child support obligations.


One way to frame the ultimate policy question is as follows: Should an informed citizen endorse the kind of welfare reform embodied in the Clinton Plan, the Personal Responsibility Act, or neither? That question, however, entails some difficult tactical judgments. Before embarking on such a tactical calculation, it is prudent to begin by coming to terms with a simpler question: Given the general framework of welfare reform issues that is on the table at the present time, how might one recognize a package of changes that could plausibly constitute an improvement over the status quo? Let us turn to major features of the Clinton and Republican welfare reform plans. We give parallel consideration to each plan even though the President has not re-introduced his plan in the 104th Congress and the Personal Responsibility Act has already passed the House.

The Clinton Plan

The Work and Responsibility Act of 1994 included these major provisions: 

* A two-year time limit on unrestricted cash assistance that would restrict most AFDC recipients to a lifetime maximum of 24 months of cash assistance. There would, however, be exemptions and deferrals for mothers with disabilities, those caring for disabled children, and the mothers of infants.

* A WORK program that would offer subsidized employment opportunities to recipients who exhaust their eligibility for cash assistance.

* Child support reform to require paternity establishment and raise additional financial support from absent fathers.

* A series of demonstration programs designed to test new approaches to teen pregnancy prevention.   

The Clinton Plan reflected the consensus embodied in the Family Support Act of 1988 that welfare recipients should be moved from cash assistance into employment and extends the JOBS participation mandate to a larger portion of the caseload. The plan would have provided that the parent of a young child would no longer be exempt from particpating in JOBS until the child’s third birthday. Rather, the parent would have been required to participate full-time as soon as the child turns one year old. Moreover, if a teen parent lacked a high school diploma, she would have had to immediately enroll in an education program. For recipients who conceived a child while already on AFDC, the exemption from JOBS would have lasted for only twelve weeks. Other exemptions would have existed for special cases, including people who were found to have an illness or incapacity that prevented them from accepting employment or participating in training.

Once the period of exemption ended, the parent would have had to enroll in JOBS, and the JOBS program would itself have been changed. Federal funding for JOBS programs would have been expanded. A parent participating in JOBS could have been thrown off of AFDC for six months for declining an offer of employment of 20 hours per week or more.

By far the most significant change would have been the creation of a time limit -- the so-called “two years and work” provision. Under this provision, a parent could generally not have received AFDC benefits for more than 24 months during her lifetime. That time limit would have been subject to several important qualifications. First, it would have been phased in, applying only to parents born after 1971. Second, some months of AFDC participation would not have been counted towards the 24-month limit: months during which the parent was less than 18 years old, months during which the parent was exempt from participating in JOBS (for example, because of the child’s age), and months during which the parent met a “minimum work standard.” Third, if a parent left AFDC with less than 6 months of eligibility left, she could receive an additional month of eligibility for every four months away from the program, until her “bank” were restored to a full six months.

At the expiration of the 24-month time limit, the Clinton Plan would have required the parent to leave the JOBS program. At that time, the parent would have become eligible for a new subsidized job program to be known as “WORK,” assuming he or she had not refused an offer of an unsubsidized job without good cause.

WORK would have offered subsidies to public or private employers to encourage them to take on AFDC recipients in work-like positions; the employer would have issued a “paycheck” in an amount that equaled the former welfare check, in exchange for however many hours of work (between 15 and 35 hours per week) that “paycheck” could have bought at the minimum wage (or, if higher, at the wage the employer otherwise paid for comparable work). WORK participants would have been required to change assignments every twelve months, and would have received an intensive reassessment every two years. They would not have been eligible for the Earned Income Tax Credit (and thus would always have received lower earnings than someone in comparable unsubsidized work), but they would have been eligible to receive AFDC benefits if their WORK wages were low enough. They would also have been eligible to continue receiving Medicaid and subsidized child care.

The Clinton Plan was estimated to increase federal AFDC spending by $11.8 billion over five years. (Note that in a single year the total costs of AFDC are now about $25 billion.) Most of the increase would have funded additional education, training, placement, and child care costs, as well as subsidies for the WORK positions.

In the child support area, like the work area, the Clinton Plan would have extended the Family Support Act and built on the consensus that absent fathers should accept greater responsibility for the support of their children. The plan would have expanded on existing child support enforcement efforts.  It would not have denied benefits to unmarried mothers, but it would have denied benefits to mothers who are unable to identify all possible fathers or are unwilling to help locate them.  And it would have required all teenage parents to live with an adult relative unless the home circumstances were dangerous or no adult relative were willing to have the teenager in the home. The Clinton Plan would have required (a) hospitals to establish paternity at birth, (b) child support payments to be adjusted as fathers’ income changes, and (c) wage-withholding and suspension of professional, occupational and drivers’ licenses to parents who refused to pay.

Our reading of the Clinton Plan is quite favorable. However, at least three key analytic questions remain unanswered:

     (1) Would the new WORK positions have provided a more effective pathway into the workforce than currently exists for welfare parents? Over the years, the federal government has supported many different forms of job creation and job training, from CETA through the WIN Demonstration projects to JOBS. Evaluations of those programs have rarely shown huge long-term benefits, but they have often shown noticeable marginal improvements. Much would depend on the details of program design and implementation.

     (2) Would the changes in young people’s opportunity sets that might have resulted from the Clinton Plan lead them to defer childbearing until more appropriate times? This is a question both about the substance of reform and about the way in which that substance comes to be understood by ordinary citizens. To the extent the impetus for welfare reform is a desire to shape behavior, the effectiveness with which reform is explained to the larger public may be as important as its actual content.

    (3) How many people would have fallen through the cracks, and how far would they have fallen? In most states, AFDC is the last meaningful safety net for children who live in poverty. Under a reformed system, what will happen to those children whose parents are unable or unwilling to comply with the greater demands of that system?

Without endorsing the Clinton Plan as written, we can comfortably conclude that it provides an appropriate framework for discussion.  While there are profound social costs associated with any changes that risk reducing support to needy children, the status quo has proven inadequate to meet the needs and desires of AFDC parents to participate in the paid workforce.  The Clinton Plan would have invested additional resources in child care, subsidized work slots, education, training, and job placement.  There could be enormous social benefits associated with a meaningful expansion of opportunity for people who are currently unable to participate effectively in the workforce.

On the other hand, the Clinton Plan no longer sets the terms of discussion. Republican control of the House and the Senate meant that the Personal Responsiblity Act of 1995 was the starting point for the current welfare reform debate. The Personal Responsibility Act rejects the consensus about work, opportunity, and responsibility that has evolved over the last decade. In that regard its conceptual framework is of great interest even if a modified version of the P.R.A. emerges from the Senate in the Spring or Summer of 1995.

The Personal Responsibility Act of 1995

The Personal Responsibility Act as passed by the House of Representatives has two critical structural features with respect to AFDC:

While the P.R.A. imposes few constraints on states’ abilities to spend their own funds to provide welfare benefits (the most notable exception being with regard to certain aliens), the restrictions on the use of the federal block grant funds are striking.  Under existing law, the federal government creates a mixture of entitlement criteria, which presumptively require the states to award benefits, and conditions on participation, which prohibit states from providing AFDC benefits to certain individuals.  Under the P.R.A., the conditions on participation — the restrictions on who may be helped with federal funds — are toughened.  But that is just the beginning.  Because all entitlements would be abolished, states would be free to set even stricter restrictions — as strict as they like.

It might be argued that the new conditions on participation are not that significant because states are free to use their own funds (funds which would be substantial if the states maintained their current levels of financial effort) to support individuals who are excluded by the new conditions.  But the new federal conditions have symbolic importance, as they signify the federal government’s perception of who is not worthy of support.  And they are likely to have economic importance as well, because many states may be reluctant to reverse a federally created presumption concerning who deserves public assistance.  In addition, the states are not required to maintain their current levels of funding.  Thus, their ability to set their own conditions may be limited as future recessions or budget crises tempt them to fund benefits entirely from the federal block grant.  

Some of the key new conditions on participation for recipients of federal funds are as follows:

* Individuals may (with a few exceptions) not receive federal funds for more than 60 months, even if they are willing to work for their benefits.

* Federal cash aid is denied to any child born to an unmarried mother under age 18 until the mother turns 18.  The federal money saved by this provision would be returned to the states for use in pregnancy prevention programs, orphanages, or similar programs, but could not be used for direct cash support of the children or families.

* Federal benefits are reduced for children for whom paternity has not been established, including children currently receiving benefits. Even if the mother cooperates fully in providing information needed to help locate the father, benefits would be reduced until the state locates the father and establishes paternity. (Families still on assistance would receive the withheld benefits once paternity was established.)

* All other federal requirements for participation in education and training activities would be eliminated, effectively making the JOBS program optional, although states would be allowed to impose rules of their own.

On balance, it is clear that the P.R.A. would reduce the number of children receiving federal aid by making them ineligible for benefits. Because of the paternity establishment, teen parent, and unconditional 60-month cutoff provisions, millions of children would be dropped from welfare, whether or not their parents were able or willing to work. Ultimately as many as half would be cut off.  While states may be free to support such children with their own funds, there is little reason to believe that many states would do so.

The package of normative and empirical assumptions that drives the Personal Responsibility Act is remarkable.  If one were to put the proposal in its most favorable light, one would say that it reveals an astonishing degree of faith in the power of “shock therapy.”  Even though the empirical evidence does not support the view that welfare mothers could all find work if only they tried harder, supporters of the Personal Responsibility Act might believe that the prospect of starvation offers an unprecedented spur to entrepreneurial innovation.  Even though the empirical evidence does not support the view that reducing AFDC reduces nonmarital birthrates, supporters of the P.R.A. might believe that clear and unequivocal ineligibility for AFDC will lead unmarried teenagers to find abstinence, abortion, or marriage more attractive than they do now.

One could also construct a somewhat darker rationale for the Personal Responsibility Act.  One could say that it reflects a sense of deep despair and desperation about the future of America’s poor children.  Supporters of the P.R.A. might believe that children born out of wedlock pose such a grave threat to American society that we must take big chances, to make dramatic use of the symbolic power of government in an effort to change the behavior of individuals.  They might even believe we should not invest too much energy in worrying about whether such dramatic moves injure some truly blameless mothers and children along the way.

A third rationale for the Personal Responsibility Act is simpler, and even more disturbing. The P.R.A. includes a broad range of proposals for changes in the American welfare state.  In addition to the AFDC restrictions discussed above, it would deny government benefits and services to legal resident aliens, would merge many federal nutrition programs into a block grant, would greatly reduce total spending on the programs to be “blocked,” and would place substantial spending caps on a variety of public assistance programs. The Congressional Budget Office estimates that the Act would reduce federal spending by $69.4 billion over the first five years.  It is surely possible that the Personal Responsibility Act reflects nothing more complicated than the narrow self-interest of a group that does not identify with other people who collect means-tested benefits.

Whatever the actual personal motivations of the proponents of the Personal Responsibility Act, we are unable to defend it by reference to any set of normative and empirical assumptions that we would find palatable. It would significantly increase hardship and absolve government of numerous responsibilities that it now attempts to fulfill.  If the Senate follows the lead of the House of Representatives, the President should veto it. 

There is room in the current welfare system for a program of balanced reform.  Changing social and economic conditions make it plausible that America could find a way to constructively recalibrate the balance among the values of work, family, responsibility, and opportunity.  Although we have our doubts about whether the 104th Congress will ultimately produce such a recalibration, here are five benchmarks that we will be using to evaluate the evolving legislative package:

1.  Jobs.  What happens when a welfare recipient has, in good faith, struck out in the private labor market, even after the federal government has deployed all the education and training and employment subsidies and other incentives it can afford? One possibility is that the person would be left to the mercy of private charity. We find that alternative unacceptable. A second possibility is that the recipient would be provided a government job at the minimum wage. Guaranteed public sector jobs programs (often referred to as “public service employment” or “PSE”) have traditionally been criticized on four grounds:

* The jobs will be “bad jobs” -- unfulfilling, unable to provide workers with transferable skills.

* The jobs will “substitute” for jobs that would otherwise exist, perhaps even replacing high-paying jobs with low-paying ones.

* The jobs will go to the non-needy because of “creaming” or patronage abuses.

* The jobs will be more expensive than they are “worth.” PSE positions might have some value and might create some new employment opportunities, but not enough to justify the overall cost.

Notwithstanding the force of such criticisms, we are convinced that a substantial amount of funds should be invested in PSE. The reason for providing a PSE job of last resort to welfare recipients and other poor unemployed adults is not to provide “good” jobs, or to increase overall employment, or to create highly valued output, but rather to offer somewhat better opportunities to those rejected by the private market. Like all redistributive programs, PSE has efficiency costs. The question is whether the equity gains of providing employment to those whose skills are not valued by market forces outweigh the program’s costs and the alternative of letting recipients who can’t find work go without cash assistance either. Given the public distaste for welfare, a PSE program, with all its faults, is a necessary component of a fair time-limited reform.

These welfare-PSE jobs cannot be the “good jobs” that we might wish every person to have. They are the minimal opportunity offered to persons who want to work, but are left behind by the private market. Supplemented by food stamps and the Earned Income Tax Credit, such jobs would raise the standard of living of PSE job-holders quite close to the poverty line. That is more than we can say for current welfare programs.

2.  Health Care. There is some evidence for the proposition that the loss of Medicaid is one of the biggest concerns of welfare recipients considering entrance into the paid workforce.  Under current law, people who leave welfare are entitled to retain transitional Medicaid benefits for a year. If one of the aims of reform is to make paid work more attractive than welfare, further discussion of universal health insurance remains a necessity.

3.  Child Care.  Any ambitious program of welfare reform rests on the premise that children’s long-term interests are served by requiring their single mothers to participate in the paid workforce.  That assumption about the long-term is necessarily linked to some assumptions about the short-term.  In particular, it assumes that children who are currently being cared for by their mothers will receive adequate care once their mothers have jobs.  But good child care is expensive.  Reformers must be willing to come to grips with that fact and account for it in their budgets.  Otherwise, the presumed long-term benefits of reform are destined to be an unfulfilled promise.

4.  Child Support Enforcement.  Welfare reform cannot be just about women and children.  No welfare reform plan can speak meaningfully about a “new commitment to enforcing parental responsibility” if it ignores fathers.  Given the high rate of joblessness among the fathers of children on welfare, that probably means attention to fathers’ opportunity sets as well.  A comprehensive welfare reform package should include fathers paying child support within the scope of experimental PSE programs that provide them opportunities.

5.  Willingness to Spend Money.  Any reform package that aspires to make a significant change along the dimensions of work, family, responsibility, and opportunity will be expensive.  In the current economy, it will cost a lot to expand work opportunities for single parents who may lack marketable skills. But if welfare reform is to be worth pursuing, it must proceed on a principle of balanced responsibility: welfare recipients and prospective parents must take responsibility for themselves and their children; the government must take responsibility for providing meaningful employment opportunities for all.  Only when everyone, regardless of fortune, has agreed to do more, will it be appropriate to speak of a new social contract.


The Great Depression taught Americans an important lesson:  that individuals can be poor and unemployed through no fault of their own.  The New Deal and the War on Poverty expressed a solidaristic commitment to help such individuals.  Working Americans would pitch in to contribute, to help create opportunities for others, so that the country as a whole would be able to live out the values of productivity and compassion.

The welfare reform debates of 1995 will show whether the values of the New Deal and the War on Poverty can endure into the twenty-first century.  It is possible that the expansion of economic inequality and insecurity — the “uneven tides” of the 1980’s — have merged with other social forces to undermine identification with the poor.  It may be that members of the middle class are now having trouble seeing themselves, their parents, their children, or their friends standing in the shoes of the poor.  If that is indeed happening, then much more than AFDC is in jeopardy.