Number A-18 in Series, "New Federalism: Issues and Options for States"

The nonpartisan Urban Institute publishes studies, reports, and books on timely topics worthy of public consideration. The views expressed are those of the authors and should not be attributed to the Urban Institute, its trustees, or its funders.

The Personal Responsibility and Work Opportunity Reconciliation Act of 1996 consolidated three federal-state match-grant programs, Aid to Families with Dependent Children (AFDC), Emergency Assistance (EA), and the Job Opportunities and Basic Skills (JOBS) training program, into one block grant program. The new program, Temporary Assistance to Needy Families (TANF), gives states considerable spending flexibility, but also imposes new work requirements and time limits for welfare recipients. Estimating the fiscal impact on states of this switch to TANF is fraught with uncertainty, in no small part because it is dependent upon the economy and on state responses to the new grant structure.

This uncertainty does not mean, however, that TANF's fiscal impact is totally unpredictable. Since TANF is fairly precise about how much federal assistance will be made available between now and 2002, one can calculate how much fiscal burden would be placed on states under simplifying but revealing assumptions. Such calculations do not provide a prediction of states' policy response to the new legislation, but they do reveal major fiscal incentives that will shape that response.

For the first few years, most states will receive more federal funds for welfare than they received in 1996, which they can spend in one of three ways: reducing their own welfare spending, increasing total welfare spending, and/or saving funds for future welfare needs. Our analysis demonstrates that there are sufficient incentives for all three to occur, with states almost certainly varying in their response. We find that future caseload changes, either down or up, can force state spending to fall or rise by a greater percentage than the percentage caseload change. Further, this fiscal impact will be aggravated by large multiplier effects of the work requirements—with an increase in the caseload, under a variety of circumstances, requiring states to increase the number of recipients in an approved work activity by more than the caseload increase.

Initial Change in Federal Funding

Federal funding for TANF is about 7 percent higher in real terms in 1997 than funding in 1996 for the programs it replaced (table 1). Indeed, state officials were given a strong incentive to accept the new TANF program once Congress based grants on historical spending in years when the economy was less buoyant and the AFDC caseload (by far the largest of the programs replaced by TANF) was significantly higher. ¹ The TANF impact on federal spending varies significantly by state, however, ranging from a 66 percent increase to more than a 9 percent decrease—with 8 states getting at least a 20 percent increase and 9 getting decreases in 1997. Even among the 9 states with less total funding, many will receive more funding per recipient in 1997. The largest percentage increases in 1997 (relative to 1996) are going, by and large, to states with the largest recent caseload declines. ²

The correlation between caseload decline and change in federal funding is not perfect, however, because TANF replaced programs other than AFDC. For example, the District of Columbia gets one of the higher increases in 1997 (relative to 1996) despite relatively low caseload decline, because its EA and its program administrative costs were much lower in 1996 than in the year used to determine its TANF grant (i.e., the comparison between 1997 and the base year would not show such a large increase). Colorado, at the other end, gets one of the largest reductions in funding in 1997 (relative to 1996) despite medium caseload decline. Many of the nine states that receive less funding actually experienced declines in AFDC caseload from their TANF grant base years to 1996 but still receive less total welfare funding in 1997 (relative to 1996) because they spent more on EA and/or administrative costs in 1996 than in their TANF grant base years.

Since the TANF grants are not matching grants, increased federal funding allows states to reduce their own welfare spending in 1997 and still provide the same level of support. To provide a simple benchmark of the surpluses potentially available to states, we estimate the total (state and federal) welfare spending in 1997 that would be necessary to maintain the same level of support as in 1996 for each state—under the assumptions that caseload and nominal benefits per case are the same as in 1996, while work and training costs, ³ emergency support costs, and administrative costs grow with inflation at an annual rate of 2.7 percent. State spending necessary to maintain the same level of support in 1997 is simply this total less the federal TANF grant to the state (table 2). This benchmark is not a prediction of actual state spending in 1997, because it ignores other features, such as TANF's maintenance-of-effort requirement. Under this requirement, states must maintain at least 75 percent of their own FY 1994 spending on EA, JOBS, and AFDC (plus transitional and at-risk child care) or risk the federal government reducing their TANF grants by the amount of the shortfall. ⁴ (Possible uses of the surplus funds, such as spending more on welfare because of the maintenance-of-effort requirements or other reasons, are discussed below.)

The 7 percent increase in real federal spending means that states, as a whole, could reduce their own real spending by 12 percent from 1996 to 1997 and still provide the same level of support to the same number of beneficiaries (table 2). In the absence of any maintenance-of-effort requirement, eight states could reduce their 1997 spending by more than 40 percent (one by over 100 percent) and still provide the same benefits to the same-size caseload as in 1996. Absent future caseload or benefit changes, four states would have to spend more in 1997 than in 1996.

As one would expect, the states that could reduce their spending the most in percentage terms under the benchmark are those states that experienced high AFDC caseload declines ⁵ from the early 1990s to 1996 (table 2). Ignoring the maintenance-of-effort requirement, these states could provide the same level of support and still have a surplus of 39 percent of 1996 spending. In terms of dollar impact on state taxpayers, the story is a bit different. Here we use real state welfare spending per resident—that is, state welfare spending divided by resident population—as an approximation of state taxpayer burden. States with the largest percentage caseload declines should not be expected to always have the largest declines in cost per resident. A simple reason is that each caseload decline in a high-benefit state saves much more money than it does in a low- or medium-benefit state. Another reason is that independent of average benefit levels, a given percentage decline in caseload reduces more cases per resident in a state with a larger baseline of welfare recipients per resident. Thus, it turns out, at least initially, that medium caseload decline states save less per resident under the benchmark than do the low caseload decline states (table 2). What drives this result is that many of the low caseload decline states provide higher levels of average benefits and support a larger number of welfare recipients per resident.

Responses to Temporary Increases in Federal Funding

States are likely to spend their excess TANF funds in 1997 on some combination of the following: (1)diverting previous state welfare spending to nonwelfare uses; (2)increasing total welfare-related spending, especially for child care and other noncash efforts; and (3)saving TANF funds for future shortfalls. We deal briefly with each possibility in turn.

Diverting State Welfare Spending

The temptation to spend some of this money—as well as any further saving due to caseload decline during the continuing economic expansion—for nonwelfare needs is strong. Recent evidence indicates, for example, that states will be under increasing fiscal pressure due to declines in other federal grants as well as to commitments to other spending, including prisons and health care, whose costs continue to grow faster than the economy and state revenues. ⁶ These pressures will encourage states to shift any TANF saving toward other functions. In addition, of course, TANF itself provides incentives to reduce welfare spending relative to former welfare law because of the switch from matching to block grants. Previously, states only paid a fraction of each additional dollar of welfare spending; now they must bear the full cost of additional spending. As mentioned above, however, states with significant surpluses will be constrained in their ability to divert money to other uses by the maintenance-of-effort requirement.

Increasing Total (Federal and State) Welfare Spending

Alternatively, states could respond to federal funding increases by maintaining their own budgetary outlays for welfare closer to 1996 levels. Note that this would increase total welfare spending and, if the number of recipients falls, the amount of spending per recipient would rise at an even faster rate than the total. Some states may not be able to avoid increasing the total (federal plus state) spending, to the extent that the maintenance-of-effort provision is effectively enforced or state legislators are simply hesitant to reduce the state's own spending by much. For instance, even if Indiana spends only a small amount of its own money in 1997, total welfare spending in the state will increase.

Where in the welfare system would this money go? Given that cash benefits per family have been falling in real terms for many years now, it seems unlikely that such money would be used for more generous cash benefits. The only alternative is spending on related services—including services intended to move welfare recipients toward jobs. Child care is an obvious possibility. Despite increased reliance on block grants for child care funding, the federal government continued to provide some matching grant money for child care as part of the same reform legislation, thereby adding to state incentives to spend their resources here. Increased spending on welfare-to-work programs such as workfare, job placement, training, child care, and transportation is another likely outcome, since such spending should make it easier for states to meet increased work requirements in later years, as long as the added benefits do not themselves stimulate increased welfare participation.

A small boost toward increasing total welfare spending came in the 1997 Balanced Budget Act, which granted an additional $1.5 billion for FY 1998 and $1.5 billion for FY 1999—much of it directly to the states—in work-to-welfare grants if states match one-third of the newly available funds on top of meeting their normal maintenance-of-effort requirements. Governor Tommy Thompson (R) of Wisconsin, an architect of state-led welfare reform initiatives, has asserted that "I have told them (conservatives) that changing a system from dependence to independence is going to cost more, because you have to put money into child care and into job training and medical care and transportation." ⁷

Rainy Day Funds

One of the more intriguing possibilities is that states will take advantage of temporary surpluses by setting up "rainy day" funds. Each year's federal TANF grant, but not its new small Balanced Budget Act supplement, appears to continue to be available until expended. ⁸ If so, it could be "banked" by a state until needs increase. A state could also save for future years by reducing its own spending, within the constraints of the maintenance-of-effort requirement, ⁹ and placing those savings in a rainy day fund as well. Saving for the future could be used to help pay for work requirements that, in the absence of significant caseload decline, become more stringent over time. "Rainy day" funds could be particularly important if caseload increases are triggered by a state, regional, or national recession.

Fiscal Burden in the Out-Years

Although more speculative, a benchmark estimate of state fiscal burden for years after 1997 also provides some insight into what states should be planning for. Over time, TANF grants fall by the rate of inflation, but so will cash benefits if states follow past precedent. ¹⁰ On the assumption of constant nominal benefits, caseload, real emergency support costs, and real work and training costs, total state spending necessary to maintain the 1996 levels of support in 2002 would be 4 percent lower than in 1997 (table 3), with 28 states spending less under the 2002 benchmark than the 1997 benchmark. Essentially, this calculation does little more than show the relative effect of inflation on real cash benefits (excluding other costs and emergency assistance) and real federal grants. When the former is higher than the latter, then the decline in real benefits due to inflation is larger in absolute terms than the decline in federal support, and a state's cash support fiscal burden declines over time. Again, these estimates are not meant to be predictions of total state spending but simply benchmarks before taking account of items such as training costs, which could rise or fall with changes in work requirements and caseload.

Changes in caseload after 1997 will be a very important factor in determining how much fiscal burden states face in the out-years, particularly because they often interact with work requirements. We consider caseload increases and decreases in turn.

Increases in Caseload

Suppose, first, that an economic downturn or demographic change causes caseload to be 18 percent higher in 2002 than in 1996 and the 1997 benchmark case. An 18 percent increase over five years is equivalent to five years of 3.3 percent annual growth, the average annual growth rate in national caseload since 1970. Under this scenario, in 2002 federal grants per recipient would be 26 percent lower in real terms than in 1997 and 20 percent lower than in 1996, with all but three states also receiving less per recipient in 2002 than in 1996. In the years before 2002, states experiencing caseload growth can potentially qualify for contingency matching funds if their unemployment rates or food stamp caseloads grow enough. This will not be a major mitigating circumstance, however, because only $2 billion was set aside for the entire 1997–2001 period, less than 2.5 percent of total TANF funding for those years, and no grants are available in 2002.

We estimate state spending necessary to maintain 1996 support levels in 2002 with an 18 percent increase in caseload, assuming (as before) constant nominal benefits, constant real work and training costs, and constant real emergency support costs and ignoring the maintenance of effort. These estimates establish a probable lower bound on state costs, since they assume either that states are able to meet the new work requirements without spending more in real terms or that the federal government chooses not to penalize states for failing to meet them.

According to these estimates, real state spending necessary to maintain 1996 support in 2002 would be 28 percent higher on average than in 1997 (table 3), with 12 states having to increase spending by more than 50 percent (not shown). The increase is less in comparison to 1996. But as a political matter, if states spend less of their own money on welfare in 1997 than in 1996, state legislators are likely to make their subsequent comparisons with 1997.

An 18 percent caseload increase in certain states would necessitate large expenditure increases. This finding reinforces the importance of saving excess TANF funds initially, especially for the states that saw large AFDC caseload declines from the early 1990s to 1996 and as a group will save the most in 1997 (see above and table 2). In these states, own-source spending necessary to maintain 1996 levels of support in 2002 would be more than 53 percent higher than in 1997.

Caseload growth interacts with the work requirements to potentially increase state fiscal burden in a way not reflected in the estimates above. Take the year 2002, when 50 percent of recipients must participate in an approved work activity. Under the Caseload Reduction Credit provision of TANF, this percentage is reduced point for point by the percentage to which caseload has declined between 1995 and the previous fiscal year, which in this case is 2001. If the caseload declines 3 percent between 1995 and 2001, for example, the TANF work requirement in 2002 among the remaining recipients is 47 percent, not 50 percent.

The incremental impact of caseload changes on work requirements, however, can be much higher than 50 percent. Figure 1 shows the number of additional persons in the caseload that must participate in an approved work activity in 2002 for every additional 100 cases added to the rolls in the previous year. As can be seen, the result depends crucially on caseload size in 2001 as a percent of caseload size in 1995, the base year selected in the legislation. When the caseload is 50 percent or more below its 1995 level, no work activity is required. When the caseload is at its 1995 level or above, 100 additional cases requires 50 additional cases in an approved work activity. Between these two extremes, a caseload increase has a multiplier effect that increases as the 2001 caseload approaches its 1995 size, because the increase progressively erodes the Caseload Reduction Credit at the same time as it adds people to the rolls.

Let us take a specific example (table 4). Suppose that caseload in both states A and B was 100,000 in 1995 but that in 2001 A has 90,000 cases while B still has 100,000 cases. In 2002, on the simplifying assumption that the caseloads are the same in 2001 as in 2002, A would have to place 36,000 recipients [(50 percent requirement minus 10 percent reduction in caseload) times 90,000] while B would have to place 50,000 recipients [(50 percent minus 0 percent) times 100,000]. The 10,000 additional cases in B lead to an additional 14,000 recipients who must be placed in work. The same type of phenomenon, although less severe, can occur in all years after 1996 when caseload is first below 1995 levels and then grows. Even if these 10,000 people move to welfare because an economic downturn shrank the number of jobs, the welfare offices must find 14,000 additional jobs for their welfare recipients in order to avoid a cut in federal funding! Caseworkers would need to expand their work programs by 39 percent (from 36,000 to 50,000) because of an 11 percent increase (from 90,000 to 100,000) in the number of cases. Thus, the interaction between caseload growth and TANF work requirements when caseloads are increasing from below 1995 levels magnifies the consequences of caseload changes and economic cycles within states.

Declines in Caseload

The AFDC caseload fell 11 percent from 1994 to 1996 and is likely to fall further as long as the economic expansion continues, TANF's focus on work placement is successful, or states respond to TANF incentives to reduce caseload. According to recent data from the U.S. Department of Health and Human Services, the AFDC caseload fell 14 percent from May 1996 to May 1997. ¹¹ If caseload in every state falls between 1998 and 2002 at half the annual rate the national caseload fell between 1994 and 1996, the 2002 caseload would be 13 percent lower than our 1997 benchmark and real federal funding per recipient would be 1 percent higher than in 1997 (recall that inflation still erodes the grant). Ignoring the maintenance of effort, we estimate state spending necessary to maintain support in 2002 given a 13 percent decrease in caseload and constant nominal benefits, real work and training costs, and real emergency support costs. Nationally, state spending necessary to maintain support could fall 28 percent from 1997 to 2002 (table 3). If not for the potential maintenance-of-effort problems, 2 of the 50 states would be able to provide the same level of support in 2002 as in 1997 without spending any money of their own.

When caseload falls, the multiplier effect of the work requirement operates in the opposite direction. If a state's caseload declines by 50 percent or more from 1995 to 2001, its overall work target in 2002 shrinks from 50 percent to zero.

Conclusion

None of the estimates presented in this brief should be interpreted as predictions of how new work requirements and five-year time limits will play out, in good part because we cannot know (1) how much states will spend to place a recipient in a work activity, (2) what the effect of each placement will be on caseload and recipients' countable earnings, or (3) what impacts there will be on child care and other job-related spending. Even so, our simulations do provide insight into some important fiscal implications for states of the TANF funding rules.

Most states have more than enough funds to implement TANF's requirements in 1997, and on average states will be able to maintain 1997 support levels for several years without spending additional funds. At the same time, total federal plus state spending per recipient is likely to increase as noncash spending rises to help meet work requirements and to comply with maintenance-of-effort goals.

Whether TANF later becomes a fiscal burden on states depends heavily on changes in caseload. If caseload remains stable, the states still might have sufficient or excess federal funds unless work requirements become much more costly. But substantial changes in caseloads will impact state budgets both because the TANF grant is fixed and because of the interaction between the work requirements and caseload.

If caseloads continue to decline as rapidly as over the last couple of years, the probability of having excess funds to implement TANF's requirements is high. But an economic downturn or adverse demographic change in any state could spell fiscal trouble for it. States with large increases in caseload will have difficulty meeting the large year-to-year fluctuations upward in demand, while the rising caseload itself is likely to have a multiplier effect on the proportion of recipients that must be placed in work. This suggests strongly that states should consider setting aside some of their TANF monies in "rainy day" funds as one mechanism to counter the procyclical aspects of the TANF legislation.

Finally, all our calculations assume federal policy remains unchanged until at least 2002. By that time, states are unlikely to assess their relative needs, or right to a "fair" share of total grants, by the relative size of their AFDC, EA, and JOBS expenditures 7 to 10 years earlier (as TANF specifies). If only a minority of states incur an economic downturn, they may not have sufficient voting strength to generate a change in the block grant formula. Even if a majority of states join the voting block for a change in the formula, it is not clear what that change should, or will, be. Adjusting federal grants upward for increases in caseload, for example, would send a very strange message because it would reward the very states that failed to achieve Congress's goal of reducing caseload. What is pretty clear is that the fiscal formula for sharing TANF funds—like most grant-sharing formulas—is unlikely to remain stable for very many years.

Notes

1. A state's grant is based on the larger of (a) its average federal grants for AFDC, Emergency Assistance, and JOBS for FY 1992–1994, (b) its FY 1994 grants, or (c) its FY 1995 grants.

2. We rank states by the change in AFDC caseload in 1996 relative to the same base years used to determine the TANF grant. High, medium, and low caseload decline categories are of equal size.

3. The work requirement states that 25 percent of all TANF families must "engage in work" in 1997. We assume states will not have to spend more in real terms to meet the work requirements in 1997 because (a) the participation rate is reduced by one percentage point for every percentage point caseload is below 1995 levels, and nationally caseload had already declined 8 percent from FY 1995 to FY 1996; (b) over 10 percent of AFDC recipients participated in work activities such as JOBS in recent years; (c) about 10 percent of AFDC families had earnings in recent years, and paid work in many instances counts toward the participation requirements; (d) about 20 percent of AFDC cases did not include an eligible adult and therefore will not be subject to the work requirements; and (e) states have the option of excluding single adults with children under age one from the work requirements.

4. If states do not meet all the work requirements, the maintenance-of-effort condition rises to 80 percent.

5. High, medium, and low categories are again of equal size.

6. See Andrew Reschovsky, "A Balanced Federal Budget: The Effect on States," The LaFollette Policy Report, Winter 1997, vol. 8, no. 1, pp. 8–21.

7. Jason Deparle, "Getting Opal Caples to Work," New York Times Magazine, August 24, 1997.

8. See H.R. 3474, §404(e).

9. Saving for future welfare spending will not count toward the maintenance of effort.

10. The erosion by inflation is offset somewhat from 1998–2001 by the supplemental grants for states that provided the least assistance per poor person in 1994 and/or experienced high population growth from 1990–1994. Nationally, the supplemental grants only increase real funding modestly, by 2 percent, but, according to Congressional Research Service estimates, could be quite significant for many of the states that receive them, increasing real funding by as much as 10 percent. However, there are no supplemental grants after 2001 or in 1997.

11.Department of Health and Human Services, Administration on Children and Families. Report no. ACF3637. 1997.

Tables & Figures