Economic Conditions and Welfare Reform edited by Sheldon H. Danziger, Kalamazoo, Michigan: W.E. Upjohn Institute for Employment Research, 1999, 321 pp. + vii, ISBN 0-88099-199-2.

This volume of papers by some of the best labour economists in the US could hardly be more timely. Just when the 1996 Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) is coming up for its five-year Congressional reauthorization and when the five-year lifetime limits on Temporary Assistance to Needy Families (TANF) are almost due, the American economy has begun to slow down. Whether the unexpectedly large declines in the U.S. welfare rolls during the 1990s can be sustained during a recession is anyone’s guess, but these authors use the evidence available as of November 1998 to provide an educated one. The outlook is not sanguine.

Three guiding questions structure the book. Chapters in the first part consider why caseloads are falling, a trend that preceded the actual 1996 welfare reform in the United States. The second section has two papers asking what happens to those who do leave welfare. Most go to work, but do not escape poverty. The final set of papers asks how the states are responding to their newly devolved responsibilities. Economic conditions affect all three of these issues, and the contributions assess the likely impact of rising unemployment and other economic factors on caseloads, recipients, and state policy.

Several observations about the volume as a whole are in order. First, the authors are meticulous in noting the limitations of their methodologies, never mistaking speculation for fact. They forthrightly posit the assumptions underlying their projected impacts of a changing business cycle, including the inadequacy of the unemployment rate as a measure of labour demand, and indicate the direction of bias in their estimates. Magnitudes of predicted effects are conservatively understated, and many chapters address the same question with several specifications, lag structures, measurement periods, or data sets. Even if the predictions are tentative, comparisons of models and methods make the conclusions more reliable, trustworthy, and useful to policy makers aiming to “reform welfare reform.”

Second, the admitted limitations of the available evidence are serious. This collection represents one of the earliest attempts to assess welfare reform empirically. Hardly two years had passed since PRWORA, although some states received earlier waivers similar to welfare reform. In some studies, pre-1996 data were the only ones available. In others, only sub-sets of welfare recipients or unrepresentative AFDC exits could be examined. As Wallace and Blank’s chapter notes, “estimates that use historical evidence on the AFDC program are probably unreliable. This is particularly true with regard to macroeconomic effects.” Recently enacted time limits, hard sanctions, diversion, and state budget constraints will all make it harder to return to TANF during the next recession than in earlier downturns. Finally, in the absence of a national effort to evaluate the course of welfare reform systematically across all fifty states, the relative success of various program designs will be difficult to estimate.

Nevertheless, these economists expertly answer the three questions posed to them about the likely impacts of an economic downturn on caseloads, recipients, and policies. First, few dispute the prediction that caseloads will rise in the next recession. They only differ over how much. Much of Part One is devoted to reconciling different estimates of the relative impacts of macroeconomic and welfare reform effects on past caseloads and projecting them into the future. Irrespective of control variables considered, both the Council of Economic Advisors and Rebecca Blank have found stronger program effects than Ziliak and colleagues who, by modeling the dynamics differently, attribute most of the caseload trends to unemployment and scarcely none to program changes. Bartik and Eberts also show that caseloads are sensitive to job loss and a changing industry mix as well as to unemployment. Shifts toward less stable jobs increase the welfare rolls even if joblessness does not rise.

In this volume, Blank also contrasts welfare to Food Stamps caseloads. This is important because PRWORA also changed the eligibility rules for legal immigrants and non-parents in the latter program. Food Stamps rolls are found to be even more sensitive to economic cycles than welfare, but they also declined more than one would have expected from program changes. This suggests that welfare reform has induced many people to leave other social programs for which they remain eligible. The interactions between income support and food, health, child care, training, job search, transportation and other support services will become increasingly important when recipients begin to exhaust their TANF time limits. Indeed, Chernick and McGuire show that there are fiscal incentives for states to substitute federal support—Food Stamps, SSI, EITC, etc.—for state-funded programs like TANF.

To hear the politicians claiming that welfare reform has been a success, one might believe that its primary goal was to reduce caseloads. However, the goal may rather be to make work pay enough to lift families out of poverty. By this yardstick, TANF is a disappointment. A sizable minority of recipients who leave the rolls do not find jobs, and many of those who do go to work do not entirely leave welfare because they have low wages or part-time jobs. Part Two of the volume considers the fortunes of welfare recipients when they leave welfare and/or go to work.

This issue can be considered from the supply side or the demand side. On the supply side, all of the national and state administrative data confirm that most welfare recipients who work remain poor, whether or not they leave the rolls entirely and whether or not their wages rise over time. The persistent poverty of most working welfare recipients or former recipients reflects their low human capital, family structure, and insufficient work hours. Moffitt uses CPS data to find that state program waivers induced less educated women to increase their work effort, but had no impact on wages. Only better educated women on welfare enjoyed rising earnings. Cancian and colleagues examine this issue with administrative data from Wisconsin and other survey sources, drawing upon unemployment insurance records for accurate work and earnings information of beneficiaries. They find that, while two-thirds of women leaving welfare do go to work, only a tiny percentage of them work full-time for all five years after exiting, while a majority never work full-time full-year. Two-thirds of the leavers also receive some other type of assistance after exit. Leavers are better educated, have fewer children, and are more likely to have recently worked than the stayers. Leavers with lower education, who had been sanctioned, or were on SSI, or lived in high-unemployment counties work less and earn less than others who left welfare. Given their low wages and insufficient work hours, a recession cannot bode well for their future economic well-being.

On the demand side, Holzer considers the impact a recession may have on employers’ demand for the sort of unskilled female workers still on welfare. He takes a more microeconomic perspective than the authors in Part One, based on a study of employers in Michigan at a time when the labour market was very tight. These market conditions of high job vacancies had little impact on wages of welfare recipients, but did increase employers’ willingness to hire, train and support recipients they employ, and participate in wage subsidy and related programs. In fact, over forty percent of employers claimed they hired someone in the last two years whom they believed was a welfare recipient. However, the demand for these workers, especially in small retail establishments, is likely to fall between 25 and 40 percent in the next recession and even more so in the long term.

Chapters in Part Three of this volume consider how state programs are likely to respond to changing economic conditions. In a recession, states will no longer be flush with block grant funds used in the current period of falling caseloads to promote employment, raise hours worked, and hike wages. The “high road” to welfare reform offering supportive services to an increasingly hard-to-employ caseload may then be replaced with a stronger emphasis on “Work First,” shortened time limits, diversion of applicants, tightened sanctions and eligibility rules, or even reduced benefit levels. The definition of “work” may also change. Pavetti compares two counties in Virginia with very different rates of both unemployment and participation in Work Experience Programs to warn states to develop alternative work activities and subsidized jobs now before private, non-subsidized jobs become scarce. Indeed, as the American economy has slowed in the new millennium, one increasingly hears that community service, wage subsidies, or public employment programs like New York’s WEP workfare program may become necessary.

Rising caseloads in a recession, even if buffered by unemployment insurance, work mandates, or sanctions, will put fiscal pressure on states. Chernick and McGuire consider the possibility that states will then cut benefit levels. Yet the authors do not expect reductions of over 10 percent for several reasons. There is a federal maintenance of effort floor under TANF payments. States do not “race to the bottom” by following their neighbours’ benefit levels. Nor have states reduced expenditures on non-welfare programs in previous recessions. But other federal programs can substitute somewhat for cash assistance. Levine also ponders whether there will be enough money for TANF in a recession or whether state expenditures will have to be reallocated to meet balanced budget requirements without raising taxes or cutting welfare or other benefits. He finds that some states without large caseload declines during the recent boom will not have adequate reserves nor receive sufficient transfers from the federal contingency fund to cover welfare obligations. By analogy, he examined how state unemployment insurance programs reacted to recent recessions. Many of them exhausted their trust funds, which were replenished but slowly, even though they now cover fewer workers. If states anticipate having insufficient funds for welfare as they have with UI, Levine conjectures they could be “playing a sophisticated game of chicken” with the federal government to win a bailout.

The prophesies of doom are already coming true. In 2001, for example, Connecticut cut into its welfare budget to reduce state spending. Even with the windfall block grant savings from more than halving the caseload in six years, fiscal pressures led to new legislation hardening time limits, tightening eligibility for extended benefits, eliminating a work-study pilot program, and toughening sanctions for noncompliance with work requirements. The changes are expected to reduce the rolls further, just as the faltering U.S. economy increases the need for support. Other states with insufficient “rainy day” reserve funds may follow Connecticut’s path as their tax revenues plunge in a recession. Only the federal government can effectively address the need for counter-cyclical spending. Although the “contingency fund” may be too little, too late, the federal government can always intervene anew to revise time limits or exemptions from work. This is a clear policy implication that emerges, either explicitly or implicitly, from reading the informed predictions in Danziger’s useful collection.