Thursday, June 30, 2005

Statistics: Wefare Reauthorization

I'm reposting, verbatim, the excellent testimony of Dr. Lawrence Mead on the subject of welfare reform.

Here's the low-down: When cash welfare for single moms was changed in 1996 from an entitlement to a work based, temporary benefit (the program is now called Temporary Assistance to Needy Families, or TANF), it was hugely successful. The economy was good, more single mothers were moved into work, poverty and the number of TANF cases went down.

But now that the economy is weak, poverty is up, and the remaining TANF cases are those that face much more serious barriers. So possible solutions are much more difficult. One of the biggest barriers to serious reform is the lack of reliable welfare data, driven by the fact that the enforcement mechanism for moving cases into work, the work participation rate (like most government statistics) is a Rube Goldberg machine of exceptions, credits, exclusions, etc. What you see is as a simple percentage is really a bizzare smoke and mirros misrepresentation.

work participation rate

A diagram of how the work participation rate is calculated.

This testimony outlines the major problems of TANF reform and the work participation rate. If you're a welfare geek (like I am) it is indespensible. To everyone else, I imagine it will be impregnable reading. Sorry, but I don't have the time right now to write an annotated version...

The Reauthorization of TANF:
Work and Child Care Provisions

Testimony of Lawrence M. Mead
before the
Subcommittee on 21st Century Competitiveness
Committee on Education and the Workforce
U.S. House of Representatives
109th Congress, 1st session

Department of Politics
New York University
726 Broadway
New York, NY 10003

March 15, 2005

I am a Professor of Politics at New York University and a longtime student of welfare reform. I’ve written several books on the subject, most recently a study of welfare reform in Wisconsin. I appreciate this chance to testify on the work and child care provisions of H.R. 240, which would reauthorize Temporary Assistance for Needy Families( TANF).

The Success of Reform

Welfare reform is unquestionably a success. Since their height in 1994, the rolls in AFDC/TANF have plummeted by over 60 percent. The overall poverty rate fell from 14.5 percent in 1994 to 11.3 percent in 2000, before rising to 12.5 percent in 2003 due to the recent recession. For children, the equivalent figures are 21.8, 16.2, and 17.6 percent. These gains are less dramatic than the caseload fall but still notable. Other research establishes that the noneconomic effects of reform on families and children have also largely been positive.

Most analysts think that the main force behind these gains was that work levels among poor heads of family rose. In 1993, only 44 percent of poor female heads with children were employed, only 9 percent full-year and full-time. These figures rose by 1999 to 64 and 17 percent, before ebbing to 55 and 16 percent in 2003. Like the caseload fall, the work gains reflected TANF’s stiffer work requirements as well as good economic conditions and new subsidies for wages and child care. Yet most studies conclude that welfare reform was the most important of these forces.

The new work requirements diverted many families into jobs who might previously have gone on aid. Under JOBS, the predecessor of TANF, the share of AFDC cases meeting work participation norms rose from 22 percent in 1994 to 33 percent in 1996. Under TANF, much more of the caseload was made mandatory for work and the hours demanded increased, yet the work participation rate still rose from 31 percent in 1997 to 34 percent in 2000, before falling to 33 percent by 2002. By a broader measure, 19 percent of cases were active in 1994, rising to 43 percent in 2001. While the majority of cases were not yet meeting the work test, the pressure to work was sufficient to transform welfare in much of the country, at least in the conditions of the last decade.

The ideal in welfare reform is to link benefits as tightly as possible to work. That requires a clear work test that employable recipients must meet as soon as they apply for aid, not sometime later. Equally important, there must be ample benefits to support working, particularly child and health care. That combination was realized most fully in Wisconsin, the subject of my recent book. For that reason, the Badger State achieved almost the greatest caseload fall in the country as well as almost the highest work participation rate—69 percent in 2002.

Reauthorization should maintain pressure on states to move the remaining recipients toward work. That in my view mainly requires fixing problems in TANF that have shielded many recipients from a need to work at all. Raising formal work standards should be secondary. I close with some shorter comments about child care.

Fixing Problems in TANF

Recently, due to shortcomings in the original law, some states have found ways evade TANF’s work demands. Some of these problems are addressed by H.R. 240, but some are not.

Caseload fall credit

TANF demanded that states raise the share of their cases where adults were in work activities by increments, until 50 percent were so engaged by 2002. But the law also allowed states to count against those targets any percent by which their caseloads fell after 1995. Because the fall was unexpectedly great, the credit cut the standards states had to meet to trivial levels. In 2002, the threshold was zero for twenty states. In that year, all states met these reduced standards, but only twelve states would have met the original 50 percent norm, only five of them without benefit of a waiver (see further below). The national participation rate reached only 34 percent in 2000, in 2002 33 percent.

The credit adds complexity, making monitoring the states more difficult. Most important, it is duplicitous, reducing the actual work standard states have to meet far below what TANF claims to the world. The case for the credit is also weak. When PRWORA was drafted in 1995-6, some states feared that rapid caseload fall might drive the most employable cases off the rolls first, making it impossible to meet the new work participation levels on the rolls. The credit allowed states, in effect, to get work credit for the decline itself. This was plausible in TANF’s early years, when massive diversion occurred and work levels soared off the rolls. It is less plausible today, when the caseload has changed little for several years and work levels off welfare have drifted down. The main task now is no longer to divert people from welfare but to make cases already on the rolls more active. To do that, TANF’s original activity norms must finally be enforced.

To that end, H.R. 240 would replace the current credit benchmarked on 1995 with one based on the four previous years. The Senate bill has an employment credit. While both versions improve on current law, they are still complicated and misleading. I would rather omit the credits and offset this by keeping, rather than raising, the 50 percent work participation norm (see further below).


Another major limitation of TANF is that it allowed states to sanction cases only partially if they failed to fulfill work requirements. A dozen states fail to end grants even in the face of open-ended noncompliance. Among these are California and New York, which have the largest caseloads, comprising 31 percent of the national caseload in 2002. In these states, reform cannot be fully implemented because much of the caseload is allowed to defy the work test. In New York City, 31 percent of the employable cases cannot be engaged because they are tied up in sanction status or in adjudication that may lead to sanctions.

H.R. 240 would mandate a full-family sanction for cases that defied activity requirements for two months or more. The Senate bill, I am told, has no such clause. It is essential, in the eventual conference, that the House insist on its provision.

Child-only cases

An emerging crisis in welfare is that more and more of the caseload is made up of “child-only” cases. These are cases where the children receive assistance but not the caretaker. The share of AFDC/TANF cases including no adult was under 10 percent in 1988, but by 2001 it had soared to 37 percent. While there is little applicable research, many of the caretakers in these cases are thought to be aliens. Their children are American-born and thus eligible for aid, but they are not, either because they are legal aliens disqualified by PRWORA or because they are illegal. By this route TANF helps to finance illegal immigration. That is reason enough to address the problem. But what is relevant here is that these cases are not subject to TANF’s time limits or work requirements.

Under AFDC, the caretaker in a child-only case could not be the biological parent; commonly, it was another relative who took charge of children when the parent was incarcerated or incapacitated. But this restriction ended with TANF. Recently, some states have begun to classify some cases as child-only even when the biological parent is still present. This allows them to exempt these cases from the work test or time limits and still draw TANF funding for them.

The child-only “out” must be ended. One option is to restore the AFDC ruling that excluded biological parents as caretakers in such cases. This would force these parents back into regular TANF, where they would face the usual work test and time limits. Another option would be to expand eligibility to cover some alien caretakers, who in turn would face normal work tests and time limits. A third option is to bring these cases under the work tests indirectly, by including them in the denominator for the work participation rate calculation. This would put force states either to limit child-only cases or to enforce work more strongly on the rest of TANF.

The idea that only the children receive support in these cases is a fiction. Now that family welfare is a work-based program, it is inappropriate for TANF to fund cases where no adult shares responsibility for the family through employment. In Wisconsin, such reasoning led the state to exclude from TANF (the state’s W-2 system) cases where the adult was unemployable or not legally responsible for the child. These families were diverted to separate programs based on SSI or kinship care. Those programs still draw TANF funding but are closely controlled and have not undercut W-2. Through reauthorization, TANF must work toward the same outcome nationwide.

Separate state programs

Similar abuses have arisen in connection with separate state programs (SSP). These are programs that states run for cases that they cannot support on TANF. Of these cases, 64 percent are in California. That state and some others use SSP mainly to support two-parent cases. The reason is to escape the very high work participation standard—currently 90 percent—that TANF demands for these families. Other states use SSP to support aliens ineligible for TANF. New York uses SSP to support the many cases that go beyond TANF’s five-year limit due to the state’s weak sanctions. In New York City, these cases comprise 40 percent of family aid.

SSP is another “out” from the work test. The problem is smaller than with child-only. Just 84,697 families were on SSP in 2001, or 4 percent of the TANF caseload in that year, although the programs have grown recently by some accounts. SSP is also less abusive than child-only, because the programs do not draw TANF funding directly. However, states’ SSP spending counts toward their maintenance of effort (MOE) requirements, so the programs are indirectly part of TANF.

One solution is to end the special work participation target for two-parent cases, as H.R. 240 proposes. This would remove the largest impetus behind SSP. Another choice, as with child-only, would be to include these programs in the denominator for the work participation rate calculation. This again would force states to limit the programs or else enforce work more seriously in TANF.


A final out is the waiver programs run by some states. These were experimental approaches to aid that many states initiated prior to PRWORA, and then were allowed to continue afterward. In the AFDC era, these programs usually toughened work requirements beyond what was then permitted by normal federal rules. Since PRWORS, however, they have done the opposite. Typically, the programs exempt more of the caseload and expect less effort to fulfill the work test than would be allowed under TANF. Massachusetts, for instance, exempts parents with children under age 6, allows indefinite job search to count as a work activity, and demands only twenty hours of activity weekly. In contrast, TANF exempts only parents with children under 1 at state option, limits job search to six weeks a year, and demands thirty hours of effort a week.

In 2002, fifteen states ran waiver programs, and in every case the program recorded higher work participation rates than they would have under regular TANF rules. Only one of these states would have met TANF’s original 50 percent norm for 2002 (in advance of the caseload fall credit) without its waiver. Seven others met that standard only with the waiver. The remaining seven fell below 50 percent even with the waiver.

The solution is to phase out waiver programs. H.R. 240 would forbid their renewal. I understand the Senate bill is unclear. Again, the House should insist on its provision.

Raising Work Standards

I would be more cautious about raising TANF’s formal work standards than in fixing the above problems. The Bush Administration’s proposals and H.R. 240 embody some good ideas, but in some case they overreach.

Full engagement

Both the Administration and H.R. 240 require“ universal engagement,” and I support this, but the meaning has to be clear. The basic idea is that recipients cannot ignore the work test. They must enroll in the work program and enter its activities when they first go on aid. What that requires has to be defined clearly in the law or, perhaps, in regulations. H.R. 240 would require that each case have a “self-sufficiency” plan, but this might easily become mere paperwork. More meaningful might be to require actual participation in some activity such as orientation or job search.

Work participation standards

The Administration has recommended raising the all-family work participation target from 50 to 70 percent of the caseload. H.R. 240 and the Senate bill would both do so. On its face, this is too ambitious. Seventy percent is more than double the national participation rate actually achieved in 2002, only 33 percent. A real activity rate of half the caseload is probably as much as most states can achieve, given the practical difficulties of getting welfare mothers out of their homes and into programs or jobs. Wisconsin’s W-2 program achieves rates above 60 percent only through intense case management and lavish support services. Most other states are not yet capable of this.

As if realizing the difficulties, the current bills would offset the 70 percent target with many credits and exemptions, including the modified caseload fall or employment credits. These would reduce the effective rate that states had to achieve to something like the current 50 percent. I would rather keep the 50 percent, phase it in over several years, and omit the credits and exemptions. That would be more honest and also more effective, because it would make clearer what was expected.

Required hours

The Administration and H.R. 240 would also raise the weekly hours of activity required to qualify a case as active from the current 30 (35 for two-parent cases) to 40. Hours required of actual work within this total wold rise from 20 to 24. As above, however, the rise would be more apparent than real because the activities that count as work would also be broadened. The hours between 16 and 40 would now be more loosely regulated, with previous curbs on vocational education eliminated. And for three months out of every 24, clients could go into full-time substance abuse treatment or other remediation. States would also get pro rata credit for hours worked short of 40.

Again, it would be better to expect fewer hours but have the demands be real. It is unrealistic to expect an actual work week of 40 hours from poor single mothers. Even Wisconsin, with its intense administration, could not achieve this. In W-2., in practice, for most of the caseload the demand fell to 30 hours of actual work, usually in a community service job, with perhaps some education or training on the side. New York City has constructed an effective program combining 20 hours of public service employment with 15 hours of job search or training for most recipients. While most localities will prefer unsubsidized employment to government jobs, this general approach is sound.

I would keep TANF’s current 30- or 35-hour standard for overall activity, its 20 hours for actual work, and its current rules for “creditable” work activities. Omit the pro rata credit. To raise expected hours simply generates unjustified demands for increased child care funding (see below).

H.R. 240 would calculate a state’s work participation rate using the total number of countable hours worked per month, rather than the number of families meeting the participation standard. This would simplify the calculation of the pro rata credit, but it would probably concentrate hours worked on fewer cases. The number of families actually participating could be reduced. Since the goal of reauthorization should be to broaden the reach of the work test, this would be a step backwards.

Permissible work activities

Under existing law, recipients can go to school and receive work participation credit for no more than one year, and the share of recipients meeting the work test this way is capped at 30 percent. H.R. 240 would restrict educational programs to four months but remove the 30 percent cap. The Senate bill would allow longer educational programs than before, in some cases even four-year college. Both of these changes would probably lead to a higher share of the caseload meeting the work test through education than before.

This would be a mistake. It would take welfare work policy back toward the era of the Family Support Act and JOBS, when most recipients were allowed to substitute school or training for actual employment. Evaluations demonstrated that “work first” was a better strategy. The fact that many recipients today are more disadvantaged than those who left the rolls earlier does not change this verdict; they, too, are likely to profit most from actual work. To allow recipient to turn welfare into a college scholarship also offends equity, since many of the taxpayers who pay for welfare lack the same opportunity. On both grounds, TANF should continue to stress work first. I would keep current rules on permissible work activities.

Performance standards

H.R. 240 would have states define their own performance measures for TANF. I find this unrealistic. Not all states can do this well. The resulting measures would also not be comparable across the country, making holding states accountable more difficult. The dangers are illustrated by school reform, where No Child Left Behind has allowed states to define their own tests for student performance. Coupled with tough federal standards, the result has been chaos.

Welfare reform should do the opposite: Let states choose goals, but control measures centrally. The objectives could include employment outcomes, such as job entries, wages, or job retention, but also reduction in poverty or nonmarital births. Up a point, states could state their own mix of objectives. But the definitions and indicators themselves should be developed nationally. It would then be clearer what states were doing and how they compared to one another. To draft indicators may require a regulatory process, but the new TANF legislation should authorize it.

Child Care

Whether child care funding is adequate for welfare reform has become a major issue in TANF reauthorization. Advocates contend that funding is insufficient to achieve the higher work participation rates contemplated in both the House and Senate bills. As now written, neither bill would raise those levels as much as appears. If my recommendations were followed, work levels would rise somewhat more, but I still think planned funding would be sufficient.

Federal funding for child care across all programs rose from $8.9 billion to $14.1 billion from 1994 to 1999, or by 60 percent. And this increase occurred in the face of sharply declining welfare caseloads. I have seen no systematic evidence that lack of child care has impeded states’ ability to move recipients off welfare and into jobs. Arguments to the contrary are unpersuasive.

Critics charge that only a minority of families leaving welfare have claimed the subsidized child care that is offered to them. But this is probably because they do not need or want it, not because they cannot get it. Critics also note that there are long waiting lists for subsidized care, and only 15 percent of eligibles received subsidized care under the Child Care and Development Block Grant (CCDBG) in 1999. But child care is a normal market good. Most of it is bought and sold privately, not provided through government. To provide a subsidy lowers the cost to consumers and raises demand; hence the waiting lines. But the fact that people seek a subsidy does not establish that they cannot afford child care without it, let alone that they cannot find care at all.

It is true that states have found CCDBG funding insufficient to meet demand. In 2002, $3.7 billion in federal TANF money was spent either directly on child care or transferred to CCDBG for that purpose. On the other hand, over 1997-2001, states spent only $62 billion of $81 billion in total federal TANF grants. It is thus implausible to say that they have done all they can to fund child care and that large funding increases are needed.

While certainty is elusive, the $1 billion increase in funding contemplated by H.R. 240 is probably enough to cover the child care needs of single mothers leaving welfare. One can argue for more money only if one posits other goals, such as providing more subsidized care to families already off welfare or improving child care quality. Those aims might be valuable, but they go well beyond the needs of welfare reform. Reauthorization should not be held hostage to them.


Welfare reform has succeeded largely by enforcing work requirement on more of the caseload than under previous law. Reauthorization should expand the reach of the work test until, in every state, aid to needy families is closely tied to employment by the parents.

The main challenge now is not to raise formal work demands but to overcome the weaknesses in TANF that have allowed much of the caseload in some states to escape the work test entirely. If we do that, there will be little need to raise work standards. The logic is the same as in tax reform: Broaden the base to which requirements apply, and what is demanded can be quite modest.

To make work standards more transparent is also important. The caseload began to fall in 19994, well before TANF was even enacted, let alone implemented. It was driven as much by politics as by formal requirements. Due to the debates over welfare, recipients got a message that work would now be expected of them. Many then went to work and left the rolls before welfare told them to. But to maintain that pressure, recipients and the public alike must understand what welfare demands. The rules under TANF are already complicated. H.R. 240 as now written would make them more so. Let us instead seek simplicity and clarity. Let us seek a more definite and more certain work test rather than a tougher one.


<-Back to the Main Page