Block Granting Welfare: Fiscal Impact on the States

Assessing the New Federalism is a multiyear Urban Institute project designed to analyze the devolution of responsibility for social programs from the federal government to the states, focusing primarily on health care, income security, employment and training programs, and social services. Researchers monitor program changes and fiscal developments. In collaboration with Child trends, the project studies changes in family well-being. The project aims to provide timely, nonpartisan information to inform public debate and to help state and local decisionmakers carry out their new responsibilities more effectively.

Key components of the project include a household survey, studies of policies in 13 states, and a database with information on all states and the District of Columbia, available at the Urban Institute's web site. This paper is one in a series of occasional papers analyzing information from these and other sources.

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.

---

Contents

Block Granting Welfare: Fiscal Impact on the States

Welfare Funding under PRWORA
Methodology
State Characteristics, 1987-1993
Simulation Results
Conclusions
References

Notes

Appendix A

About the Author

---

Block Granting Welfare:
Fiscal Impact on the States

One of the biggest changes to the welfare system embodied in the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (PRWORA) is switching from an open-ended match to fixed block grants as the federal funding mechanism for state-run welfare programs for families with children. Some economists are particularly concerned that the quite different incentive structure of a block grant regime will lead to dramatic declines in state welfare expenditures as states cut basic benefit levels, perhaps even triggering a "race to the bottom."⁽¹⁾ Further, with the new authority over the design of their welfare programs granted to them by PRWORA, states may be tempted to change more than just benefit levels in response to their changed fiscal environments. For example, states might redesign eligibility rules or eliminate citizen entitlement to welfare in an effort to control costs. Advocates for the poor worry that fiscal concerns may become major determinants of states' welfare policy designs.

The immediate fiscal impact of the changeover in funding systems, resulting in initial windfalls to most states, has clearly been positive for the states. However, the block grants are nominally fixed over the horizon of PRWORA, so the possibility of future declines in federal funding for welfare, compared with the funding levels that would have prevailed under the previous regime, merits serious concern. Relative declines in federal funding will ultimately be determined by such factors as population growth and economic conditions, including unemployment and inflation. Although it is early in the debate, many economists believe that the withdrawal of federal funds from the welfare system will ultimately be of a serious magnitude (Chernick 1998).

Under the block grant system, the qualitative nature of states' funding commitments is also changed. Because a state must usually contribute the "marginal dollar" if its welfare program is to expand beyond the mandated funding level under PRWORA, states may face far greater potential volatility in their welfare expenditures than under the matching regime, ceteris paribus. At least in the aggregate, caseloads and expenditures expand during recessions, just when state fiscal capacity is often declining. Such cyclical funding risks are thought to be borne more efficiently by the federal government, because shocks to expenditures and revenues are pooled regionally and because of the relative ease with which the federal government issues debt.

State policymakers, who had a major role in shaping PRWORA, have tended to stress brighter block grant funding scenarios. States will be "rewarded" with more federal funds under block grants than under matching grants when caseloads decline because of effective welfare policy changes or an improved economy. Under appropriate conditions, states could end up with a larger cumulative federal contribution over the PRWORA funding cycle than if the old matching regime had prevailed. The U.S. General Accounting Office (GAO) estimates that in the first full year under welfare reform the states collectively enjoyed a $1.4 billion windfall of federal funds as a result of the initial funding formula, caseload declines since 1996, and the fixed block grants (GAO 1998). If states do not use the windfalls merely to replace their own expenditures but instead choose to invest these additional funds in policies that permanently reduce welfare dependency, this is a benefit of the new regime that should legitimately be considered. To the extent that states carry over these surpluses in rainy-day funds, states may also be able to better match federal funding with actual needs over time.⁽²⁾ Additional federal funds are available for states in recession and for states with relatively high population growth. From this perspective, it is perhaps less surprising that economists' fears of catastrophic withdrawals of federal funding were not shared by many state policymakers when PRWORA was under development.

This report undertakes a basic analysis of the change in funding regime, based on historical expenditures over a recent seven-year period. The goal is to assess the impact on state fiscal environments of the switch from matching to block grant federal funding. The broad question to be examined is, How much of the burden of welfare finance does PRWORA effectively transfer to the states? Specifically, how large a withdrawal of federal funds from the welfare system does PRWORA represent overall, and how much will the stabilizing properties of federal funding be compromised? As the introductory discussion suggests, the answer to these questions depends on each state's welfare expenditure needs over the funding cycle, the feasibility and desirability of using the Contingency Fund, and the ability to accumulate funds in Treasury and state accounts, which are intended to act as safety valves in years when current needs exceed the Temporary Assistance for Needy Families (TANF) block grant.

The methodological approach makes the strong assumption that historical state expenditures would still represent desired state expenditures in a post-PRWORA world. The analysis is primarily an accounting exercise, in the sense that the determinants to state welfare spending, including the "price effect" of the switch from matching to block grants, are ignored. However, it is still possible to learn a great deal about the post-PRWORA fiscal environment in which states will operate. For example, one needs only weak assumptions about the course of state spending to model use of the Treasury account. Where the findings might depend strongly on the assumption of no behavioral response is noted at appropriate points.

The next section discusses the new funding regime under PRWORA. The following section lays out the methodology of the simulations. States' historical experiences with Aid to Families with Dependent Children (AFDC) program expenditures over the period relevant to the simulations are discussed, and federal support for state welfare programs that would have prevailed from 1987 through 1993, if the PRWORA funding rules had been in place, are simulated. The last section discusses the findings and conclusions.

Welfare Funding under PRWORA³

Until fiscal year (FY) 1997, the federal government contributed from $1 to $4.88 for each $1 a state spent on AFDC. The rate at which state payments were matched depended on each state's characteristics, including poverty rates and fiscal capacity. Higher funding rates were generally negatively correlated with the welfare benefit offered to families. For example, states in the low-benefit southern region had matching rates that were among the nation's highest. The first column of table 1 presents average federal funding shares for each state over the period 1987-1993. Mississippi (0.80), West Virginia (0.76), Arkansas (0.75), and Utah (0.75) had the highest shares. Twelve states received the minimum share (0.50) in most years.

With PRWORA, federal welfare funding changed to a system composed of a TANF block grant based on combined past federal contributions for AFDC, the Job Opportunities and Basic Skills (JOBS) program, and Emergency Assistance (EA). For most states, the federal block grant is based on 1994 expenditures and is nominally fixed through FY 2002.⁴ For 20 states, PRWORA authorizes modest growth in the TANF block grant (2.5 percent each year through FY 2002). These states are marked with an asterisk in table 1. The 20 states selected to receive these additional funds have either experienced significant population growth (exceeding 10 percent between the 1990 census and July 1995) or paid AFDC benefits that were 35 percent or less of the national average benefit in FY 1994. PRWORA commits $16.5 billion annually through FY 2002 to provide TANF block grants to all the states.

Because of dramatic declines in AFDC caseloads since 1994, the majority of states received an FY 1997 TANF block grant payment that exceeded the federal matching payment they would have received under the AFDC system. The second column of table 1 reports these state windfalls (as a percent increase over the hypothetical funding that would have been provided in the AFDC regime), as estimated by the GAO (GAO 1998, table II.4). Forty-five states shared an FY 1997 windfall of $1.4 billion. The size of individual state windfalls ranges widely (from minus 7% in Pennsylvania to 70% in Indiana), depending primarily on caseload declines since 1994. The median windfall represents a 9 percent increase over hypothetical federal funding under the matching regime.

Most states must maintain their own welfare funding at 80 percent of their FY 1994 expenditures on AFDC, JOBS, EA, and welfare-related child care programs or face large penalties to their block grants.⁵ Minimum requirements on state spending are termed "maintenance of effort [MOE]" levels; in this report, this particular provision is called the "80 percent MOE" funding level. States that meet the job targets in PRWORA may reduce their spending by an additional 5 percent, to a 75 percent MOE level.⁶ As is the case with the federal TANF block grants, for most states, MOE spending is based on spending in a period with substantially higher caseloads. However, because the state MOE level is 20 percent below earlier expenditures, states' required spending is less than actual state spending on comparable programs in 1996, according to the GAO. Only one state, Indiana, which experienced an enormous windfall, was required to spend more under PRWORA (GAO 1998, p. 44).

In combination with the matching funding system, the entitlement status of AFDC meant that federal funding expanded automatically with caseload expansions. Of course, state spending was also required to increase. Similarly, the state shared proportionately in any welfare spending declines. Under a block grant, there is no year-to-year variation in federal expenditures; thus, with PRWORA, the marginal welfare dollar is contributed by the state. If the state is trying to meet a targeted amount of total spending, which may change from year to year because of underlying economic and social conditions, the volatility of state spending would be greater under PRWORA than under AFDC.⁷ Some elements of PRWORA are intended to partially preserve the fiscal stabilization aspect of federal funding that is lost in the change from a matching to a block grant.

Each state may carry unlimited TANF funds forward for use in later years, giving states some power to reallocate their federal funding stream over time. States can thus make spending from the federal source more variable, possibly reducing the variability of their own expenditures. Clearly, this feature is an imperfect stabilizing mechanism, as states' saving decisions are based on forecasts of future developments and thus are prone to error. Also, because of the wide disparities in windfall amounts and individual developments in states' funding needs over the life of PRWORA, there will be large disparities in states' potential to accumulate savings accounts from federal funds. However, this tool is unlikely to be used to the fullest by the states for two reasons. First, state policymakers have expressed concern that large balances held at the Treasury will signal to Congress that TANF is overfunded at the federal level. Second, unresolved government accounting issues have created uncertainty about the conditions under which states will be able to retrieve these funds, further discouraging their use (GAO 1998).

States, of course, are free to carry their own funds forward for use in future years. By saving and dissaving, states would be able to smooth total welfare spending. Savings for this purpose do not count toward MOE spending in the year they are set aside, but withdrawals constitute MOE spending. The extent to which states would wish to create rainy-day funds from their own resources probably depends on the size of federal windfalls relative to total desired spending on TANF, because states would presumably set aside funds that can be replaced by federal spending instead of immediately diverting the freed-up state funds to expenditures on other programs. Again, states that begin under PRWORA with large windfalls are in the best position to establish such accounts.

Additional federal "stabilization funds" will be made available through the Contingency and Loan Funds. PRWORA set $2 billion aside to finance limited federal matching, subject to changes in states' unemployment rates or Food Stamp caseloads.⁸ Thus, the federal government is in principle willing to share some of the "downside" risk to program funding (and to reduce the marginal cost of financing welfare program expansions under certain conditions). A state's unemployment rate must exceed 6.5 percent and represent at least a 0.10 percentage point drop from the same months in either of the two preceding years. Alternatively, the Food Stamp caseload must have increased at least 10 percent over either its 1994 or 1995 levels. Several conditions will probably discourage great state reliance on the Contingency Fund. First, states experiencing significant, ongoing economic distress may still fail to meet the exact economic criteria for accessing the fund. Second, states have to increase their own spending significantly to induce a federal match. States must meet a 100 percent MOE in years the fund is used.⁹ They are then eligible to receive matching funds on expenditures above the 100 percent MOE at Medicaid matching rates (similar to AFDC matching rates).¹⁰ The total amount of federal matching monies available to a state is limited to 20 percent of its TANF block grant in any year. Finally, despite the fact that the initial allowance of $2 billion for the fund was decried as insufficient by many observers in 1996, the Adoption and Safe Families Act of 1997 further reduced the fund by $40 million over four years. Although this reduction is relatively minor, the method of accomplishing the reduction appears to impose large penalties on states that happen to need the Contingency Fund in years when most other states do not (GAO 1998). The Contingency Fund has many unattractive features, but the most frequently voiced concern is that because the secretary of health and human services is directed to make contingency funds available on a first-come, first-served basis, the fund will be depleted before some states in need become eligible to use it.

The $1.7 billion Federal Loan Fund for State Welfare Programs allows states to borrow from the federal government for up to three years at rates comparable to Treasury instruments with similar maturity. The Loan Fund is not included in the simulations in this report. Because the funding must be repaid, the Loan Fund is essentially equivalent to the options of raising taxes or reducing expenditures on other programs in order to finance welfare spending.

---

Methodology

The effect of the new funding regime on states' fiscal environments is simulated here by using information on the states' actual funding experiences under AFDC.

Because PRWORA authorizes only a seven-year funding cycle and because it is extremely difficult to predict the levels at which the block grants might be renewed for FY 2003 with any accuracy, the analysis is restricted to a single seven-year period. Of course, the choice of a particular seven-year period is critical to the findings. FY 1987-FY 1993 is chosen for several reasons. First, state experimentation with welfare reform (with the exception of Wisconsin) is not a major factor over most of this period (the Bush administration began encouraging states to seek waivers to deviate from the standard AFDC program in the early 1990s). Therefore, although this period is fairly recent, the sequence of welfare expenditures is not unduly influenced by a significant regime change in welfare policy. In fact, a chief difficulty of evaluating the influence of the change in funding regimes in the post-PRWORA world is that states will likely institute policies that independently make the change to block grants appear highly favorable. To take one example, a policy of discouraging those who approach the welfare office from actually completing an application, if effective, will drive down caseloads and costs, making the TANF block grant appear more generous relative to the perceived need for welfare funding.

Second, the aggregate unemployment rate in 1987 is roughly comparable to the aggregate unemployment rate in 1996. This is convenient because the unemployment-rate-based rules for using the Contingency Fund are couched in terms of absolute, not relative, levels and changes.¹¹

Third, the FY 1987-FY 1993 period is qualitatively similar to the initial period of PRWORA because it begins with a subperiod of improving economic conditions and declining AFDC caseloads and expenditures. It also contains a moderate recession (1990-91), an interesting feature as the sufficiency of federal funding over the next recession (which is likely to occur during PRWORA's seven-year funding cycle) is a major concern.

Block Grants and MOE Levels

The TANF grant for each state is computed by grossing up its FY 1987 federal AFDC expenditures by the actual windfall (in percentage terms) it is estimated to have received from PRWORA, as reported in column 2 of table 1. It is important to note that these windfalls are primarily based on state economic conditions in 1994. (Although one could combine the rules for setting the TANF block grant with pre-1987 data, using the relative windfalls actually received by the states should better indicate states' post-PRWORA fiscal positions relative to each other.) For most states, the TANF grant amount constitutes the certain funding they receive from the federal government in each of the seven years. For 20 states, however, the TANF block grant is assumed to grow by 2.5 percent each year after the first (note that all states' MOE levels remain fixed). A state's usual MOE level is computed as 80 percent of its share of grossed-up (by the appropriate windfall) actual expenditures in 1987. (Because of the difficulty of predicting which states will meet work targets over seven years, the possibility of a 75 percent MOE level is ignored.) Once the TANF block grants and state MOE levels are computed, it is a simple matter to examine the minimum certain funding that states' welfare programs will receive under PRWORA from both federal and state sources, and to contrast these figures with historical funding streams.

Treasury Funds

States have some flexibility in reallocating the federal funding stream over time by accumulating TANF funds in accounts at the U.S. Treasury. To compute states' savings from federal block grants, two assumptions are made about state funding choices. The first assumption is that, given a choice, states would rather spend federal than state funds on their welfare programs. The second assumption is that states want to maintain their total welfare expenditures at historical levels, if possible, but they do not wish to spend more than this amount in total; that is, actual historical expenditures continue to represent states' demands for welfare spending. Although the latter assumption does not seem unrealistic on its face, it does ignore the incentive structure of the overturned matching regime and the influence that the varying state "prices" of welfare spending (i.e., the match rate) may have had in determining the total size of state AFDC programs.¹²

With these two assumptions, one can readily determine whether a state would accumulate a savings account at the Treasury and the amounts it would save and dissave over time. First note that states are required to spend the 80 percent MOE level each year. Therefore, they are concerned with financing the residual of total historical expenditures less MOE expenditures, if this amount is positive. In years when the TANF block grant exceeds this residual, the state is able to deposit the surplus in its Treasury account for future use, where the funds are assumed to earn interest at the three-month Treasury bill rate. In years in which there is a positive Treasury account balance, the state will dissave from the fund whenever its needs (as indicated by total historical expenditures) exceed its current resources (the 80 percent MOE plus the federal block grant). To model the use of the Treasury fund, it is not necessary to make additional assumptions about state behavior with regard to making up any shortfall from historical expenditures once all federal funds have been exhausted. If one does wish to model state spending, two possible extreme alternative assumptions are that the state will spend only its MOE level and that the state will finance 100 percent of any shortfall.

The Contingency Fund

Introducing states' possible use of the federal Contingency Fund, which provides federal matching funds to states under certain conditions, is somewhat more involved. Determining which states meet the economic requirement is straightforward. The unemployment rate condition rather than the Food Stamp condition is used here, as states will probably wish to determine their eligibility as quickly as possible. Unfortunately, only simplified unemployment conditions can be simulated; the part-year use of the Contingency Fund cannot be simulated because of a lack of expenditure data at greater-than-annual frequencies. In addition to meeting the economic criteria, a state must perceive that its own funds, together with available federal funds (from the current block grant and those accumulated from past block grants), fall short of the desired amount of welfare spending. Clearly, states that wish to spend only the minimum possible amount on welfare under TANF will never use the Contingency Fund. These simulations are based on the extreme assumption that states wish to achieve the historical level of total expenditures and that they are willing to bridge any gap between federal resources and desired expenditures entirely with state funds, if necessary. Thus, the assumptions underlying these simulations imply maximum use of the Contingency Fund by the states.

To qualify for federal matching funds, a state must first increase its spending from an 80 percent MOE level to a 100 percent MOE level. In addition, the state must pay its share of the remaining shortfall from historical expenditures after the block grant, any Treasury account balance, and the 100 percent MOE have all been spent, in order to receive federal matching funds for the remainder of this amount (up to 20 percent of the TANF block grant). Because states are required to increase their own spending in order to access the Contingency Fund, it is only rational for the state to seek matching funds if the total cost to the state is less than the cost of financing the entire shortfall itself. Rationality requires that the 100 percent MOE amount does not itself exceed historical total expenditures less federal block grant resources (which may come from Treasury account dissaving as well as the current block grant payment). Thus, two conditions must be met simultaneously for a state to rationally use the Contingency Fund. First, the state must meet the objective economic criteria in a given year. Second, the perceived shortfall in welfare funding remaining after block grant and Treasury account resources are exhausted must be sufficiently large to justify the additional state expenditures required to induce the federal match.

Volatility of State Expenditures

Historically, total AFDC expenditures have varied substantially over time within states. The AFDC system had two features with important implications for the stability of state expenses. First, AFDC was an entitlement, implying that states were committed to increase or decrease funding with caseload increases or decreases. Thus, states had little control over the variability of their expenditures under AFDC. However, the second salient feature was that the federal government shared in these funding changes, thereby substantially reducing the states' exposure to AFDC funding risk. Under PRWORA, federal block grants are invariant with respect to the states' current needs.¹³

With the possible exceptions already noted, states will need to use own-source revenues to service any additional demand for welfare spending they may face (similarly, spending declines will be absorbed completely by state dollars down to the MOE spending level). This increased variability in potential state liability for welfare spending may exacerbate financing problems at the state level. First, welfare is not a popular spending category in state budgets, and increased appropriations at any time may be hard-won. Second, states' tax bases are strongly variable (e.g., see Dye and McGuire 1991), and major sources of state revenue, such as personal and business incomes, tend to move strongly procyclically. Welfare spending, on the other hand, is thought to move countercyclically.¹⁴ Thus, increased variability of welfare spending resulting from a move to block grants may contribute to a lack of congruence between state spending liabilities and revenue sources.

Obviously, if states choose an MOE-only policy, the observed volatility of state funding would decrease to zero. Of greater interest is some measure of states' increased potential "liability" for the variation in welfare finance as a result of PRWORA. Therefore, it makes sense to judge the potential increased volatility of state spending against a "full-funding" benchmark; that is, holding total welfare expenditures constant, what happens to the funding risk facing the states when the PRWORA funding regime is instituted? This aspect is examined using an index of the increased volatility of states' potential welfare liability: dividing the hypothetical coefficient of variation of state expenditures under PRWORA, under the assumption of a state commitment to full funding of the program, by the historical coefficient of variation of actual state expenditures over the period.¹⁵ This measure indicates the relative increase in fiscal pressure on the states as a result of their greatly increased responsibility for marginal welfare funding.¹⁶

---

State Characteristics, 1987-1993

This study's assessment of the impact of PRWORA depends on the pattern of historical expenditures. States with large windfalls and low expenditure growth over the horizon of the legislation may actually gain federal funds (relative to AFDC) over the life of the block grant. Further, the use of the Contingency Fund depends on both historical expenditures (which proxy for desired total expenditures) and the unemployment rate-determined eligibility conditions. This section describes the data on state unemployment conditions and expenditures over the period 1987-1993.

Unemployment

Most states experienced a period of increased unemployment during the reference period. Nationwide, the unemployment rate was 6.2 percent in 1987. It fell to 5.3 percent in 1989 before rising to a peak of 7.5 percent in 1992. By 1993, the unemployment rate stood at 6.9 percent. Although the timing of recessions varies somewhat across states, prospective demand for the Contingency Fund is heavily concentrated in 1991 and 1992. Appendix A indicates the years in which each state would have met the necessary unemployment rate conditions to access the Contingency Fund. Before 1991, few states would have qualified (six in 1987, none in 1988 and 1989, and three in 1990). In 1991 and 1992, 50 percent or more of the states (25 and 32, respectively) would have qualified, and 15 states would have met the economic criteria in 1993 (12 of these would have met the criteria because their 1993 unemployment rate, while representing a decline from 1992, still exceeded their 1991 rate by 0.1 percentage point). Of course, some states experiencing recessions over this period still would have failed to qualify for the Contingency Fund.

Expenditures

As discussed earlier, column 1 of table 1 presents average federal AFDC funding shares for the period. Matching rates change very little across years within a state, so for most states these are very similar to the statutory rates. Nationwide, about 55 percent of welfare spending in this period is financed by the federal government. In the next section, these shares are compared with simulated shares under various assumptions to assess how much the federal government's overall responsibility for welfare funding is increased or decreased by PRWORA.

Column 3 of table 1 presents average annual growth rates of state expenditures over the period. Nationwide, growth averaged 7 percent, but states' experiences are diverse. Arizona (19.5%), Florida (17.9%), New Hampshire (17.2%), Pennsylvania (16.3%), and Nevada (16.2%) led the nation in expenditure growth. In contrast, Iowa, Michigan, Illinois, and Wisconsin experienced essentially no growth over the period. Of the states provided with a 2.5 percent growth allowance in their TANF block grants, only about half experienced above-average growth, which largely reflects the fact that there are two criteria for being a "growth fund" state. Ten states receive a growth allowance because they pay relatively low benefits; presumably, the allowance will force these states to modestly increase welfare funding over the funding cycle, relative to other states.

It is interesting to ask whether states that meet the economic criteria for accessing the growth fund are among the states that face the greatest expansions in AFDC expenditures. For each year, the average expenditure growth for the group of states meeting the economic criteria is computed and compared with average growth for the group of states not meeting the criteria. Of 1991, 1992, and 1993 (the years in which a reasonably large number of states meet the criteria), only in 1992 do the states that qualify to use the fund also experience above-average expenditure growth from the preceding year. In 1991, growth in qualifying states is only slightly greater, while in 1993, it is actually lower.¹⁷ This is consistent with some past studies and anecdotal evidence that the relationship between welfare expenditures and the business cycle is a weak one in many states.¹⁸

Column 4 reports on the variability of state expenditures as measured by the coefficient of variation. As before, states' experiences over the period are diverse, ranging from nearly constant nominal spending in Michigan, Louisiana, and Mississippi to quite variable spending in New Hampshire, Arizona, New Mexico, and Nevada. Although the high-expenditure = growth states usually experienced relatively greater expenditure variation as well, variability can differ dramatically across states with similar average expenditure growth. For example, Pennsylvania and Nevada experience high and similar expenditure growth, but the variability of state spending is much higher in Nevada.

---

Simulation Results

This section presents the findings from simple simulations of state welfare finances under PRWORA, comparing them with actual historical expenditures over 1987-1993. The discussion covers the amount of "guaranteed" funding for welfare under PRWORA, both from federal and state sources; the possible use of the Contingency Fund and Treasury accounts; and two facets of the change in the funding burden on states—the potentially increased share of state spending in total spending and the potential for increased state funding volatility.

"Guaranteed" Welfare Funding under PRWORA

States are guaranteed to receive at least the TANF block grant each year. Because the block grants are fixed, their enduring value over the seven-year cycle of PRWORA, relative to the comparable federal funding commitment under the matching regime, depends on inflation, underlying changes that influence welfare spending (e.g., caseload fluctuations, changes in welfare policy), and the size of the initial windfall to states. Committed federal funding is simulated for states, assuming that a block grant regime, with initial windfalls to states comparable to those under PRWORA, had been instituted in 1987. The first set of results is presented in table 2.

Column 1 of table 2 compares total federal funds provided through block grants to actual federal expenditures over the period, assuming no use of the federal Contingency Fund. The aggregate national difference of -4.3 percent indicates a fairly modest withdrawal of federal resources under block grants. However, a glance down the rows of column 1 reveals enormous differences in funding changes across states. For 18 states, federal funds available through block grants exceed actual federal expenditures over the period. States such as Wisconsin, Michigan, and Indiana, all of which receive extremely large windfalls, experience huge gains from the switch to block grants (55%, 37%, and 35%, respectively). However, for the majority of states, the block grant funding scheme leads to a net withdrawal of federal resources from their welfare systems. The hypothetical block grant funds of Arizona, Florida, Hawaii, Nevada, and New Hampshire amount to less than 70 percent of the value of federal funds actually received over the period. Not surprisingly, these five states are also the leaders in expenditure growth. None receives a sufficiently large initial windfall or (in the case of Arizona, Florida, and Nevada) a sufficiently large growth allowance to make up for the essentially fixed nature of the block grant.

However, TANF block grants are only a portion of the total federal financial guarantee to states embodied in PRWORA. As mentioned earlier, the Contingency Fund is an additional source of federal funds for states that are willing to pay a higher MOE level and contribute matching funds.¹⁹ In column 2, the percentage difference between hypothetical federal funds provided under PRWORA (now also considering rational use of the federal Contingency Fund by the states) and historical federal expenditures is reported. Note that the percentage change in federal funding remains the same for the 24 states that never chose to use the fund under these assumptions. Once the Contingency Fund is considered, the impact of the PRWORA funding regime on aggregate financial flows to the states is minor (minus 1.4%). Again, however, the disparities among states are enormous, although the states that suffer the greatest loss from the change to block grants are able to close the gap significantly by using the Contingency Fund. For example, Arizona, Florida, Nevada, and New Hampshire are able to reduce their funding gaps by 7.4 percentage points, 9.5 percentage points, 17.2 percentage points, and 15.4 percentage points, respectively, although the total withdrawal of federal funds remains large.²⁰

States are also required to fund TANF at the MOE level under PRWORA. Just as the block grant formula's emphasis on a year of generally high caseloads caused the federal windfalls to most states, it is also true that states' funding commitments are based on a year of relatively high expenditures. However, states need pay only 80 percent of this amount to satisfy their MOE requirement, suggesting that only states with enormous windfalls, or precipitously declining welfare expenditures over 1987-1993, are likely to be required to spend more under TANF than under AFDC. Column 3 of table 2 demonstrates that some states (Indiana, Louisiana, Michigan, West Virginia, Wisconsin, and Wyoming) would be required to spend more for welfare under the TANF funding system than they would have on AFDC over the same seven-year period. However, most states could significantly reduce their spending on welfare without incurring a federal block grant penalty. Consequently, nationwide, states have the potential to reduce their own welfare spending by 25.6 percent and still remain in the TANF system.

The figures in columns 2 and 3 can be combined to examine the total change in "mandated" funding (minimum required spending on welfare) brought about by PRWORA. The total required spending of 14 states meets or exceeds historical spending under AFDC. Iowa, New Jersey, South Carolina, and Utah could spend about the same. Alabama, Montana, Ohio, and West Virginia could run moderately larger programs. A few states would be required to run much larger programs under TANF (programs in Indiana, Louisiana, Michigan, Mississippi, Wisconsin, and Wyoming would be at least 20 percent larger). Many states could cut their programs substantially, if they so desired. For example, Alaska, Arizona, California, Connecticut, Florida, Hawaii, Nevada, New Hampshire, New Mexico, North Carolina, and Tennessee would have been free to cut spending by more than 25 percent over this era without repercussion, had the PRWORA funding structure been implemented in 1987.

The Contingency and Loan Funds

This section examines which states use the Contingency and Loan Funds, describes how their balances evolve over the period, and discusses the overall adequacy of the Contingency Fund.

Regarding Treasury accounts, column 1 of table 3 lists the number of states with a positive balance, and column 2 lists the aggregate balance in each year. Under the assumptions used throughout this study, more than half the states would have a positive Treasury account in some year. Twenty-five states would have opened Treasury accounts at the outset, with one more state doing so in the next year. After 1988, the number of states with open Treasury accounts would begin a slow, steady decline, but a large number of states (15) would end the period with a positive balance. Column 2 lists the aggregate amount of states' deposits. As state officials fear, these balances would be huge, growing to $5.7 billion by 1993, or 57 percent of all TANF block grants that would have been paid out in 1993. Three states' activities dominate the Treasury accounts: By 1993, 74 percent of the funds would be owned by Wisconsin, Michigan, and Indiana. Presumably, states with such large accounts would be prime targets for block grant reductions when the funding cycle rolls over.

The remaining columns of table 3 describe states' use of the Contingency Fund. It is important to ask which states would be motivated to use the fund. Column 3 lists the number of states in each year with a funding "gap"; that is, after spending the 80 percent MOE amount and block grant source funds (be they from the current block grant or dissaved from the Treasury account), these states would still be spending less on welfare than under the AFDC regime. From 1989 to 1993, the majority of states would have such a gap in each year. Not surprisingly, because of the fixed nature of the block grants and overall growth in historical expenditures, the number of states "in need" would grow from 25 in 1987 to 35 in 1993. However, the funding gap must be sufficiently large to make it worthwhile for the state to enhance its funding to induce a federal match; otherwise, it is cheaper for the state to finance the entire shortfall itself. The fourth column of table 3 lists the yearly number of states that would have sufficiently large funding gaps to rationalize use of the Contingency Fund.

Comparing columns 3 and 4, in the early years, it is obvious that the simulated funding gaps must be quite small. Over time, as federal funding would become increasingly inadequate, the size of the funding gaps would grow until, in the 1990s, nearly all states with a gap could rationalize spending the 100 percent MOE necessary to trigger access to the Contingency Fund. As discussed, however, "need" as measured by historical welfare expenditures does not mesh perfectly with "need" as determined by the unemployment rate criterion in PRWORA (the number of states satisfying the latter is given in column 5).

Column 6 lists "fund users"—states that would meet the economic criterion and find use of the fund rational. Until the onset of the recession in 1991, only two states would use the fund (Colorado in 1987 and Alaska in 1990). In 1991, 18 states would use the fund, drawing $862 million in federal matches, or less than 10 percent of total TANF block grants paid out in that year. Use of the Contingency Fund peaks in 1992, with 22 states drawing a little over $1 billion. In 1993, nine states would rationally use the fund, seeking an additional $769 million of federal money. However, only $39 million of this could be awarded without exceeding the $2 billion limit.21 As with the Treasury accounts, Contingency Fund activity would be dominated by a few states. California would seek $410.1 million in 1991, 1992, and 1993 (the maximum amount, or 20 percent of its TANF block grant, in each year). New York would be a distant second, seeking $139.1 million in 1991 and a bit more than half of what California seeks in 1992 and 1993. The top three fund users (California, New York, and Pennsylvania) combined would receive 72 percent of all contingency funds in 1991 and 65 percent in 1992. In 1993, California and New York alone would account for 82 percent of funds sought. (The smallest fund payment to any state is the $1.2 million paid to Colorado in 1987.)

As discussed, use of the Contingency Fund imposes a large cost on states, in that they must not only match federal spending (at rates similar to the AFDC matching rate) but must also meet a 100 percent MOE in the year of the match. Column 8 presents the amount of additional state spending (i.e., all spending above the 80 percent MOE) that states would be required to undertake to induce the desired federal match. Colorado and Alaska would spend many times their small federal matches in additional state funds. (In general, smaller fund payments—where "small" is defined relative to the overall level of state welfare spending—will require more excess state expenditures per additional federal dollar, because the state has to absorb the fixed cost of getting up to the 100 percent MOE regardless of the size of the federal match payment.) In 1990, 1991, 1992, and 1993, aggregate additional required state spending would be roughly double that from federal matching funds. However, depending on each state's unique fiscal situation, the amount of state funding needed to induce the match would vary widely. For example, in 1992 Illinois would have to put up $95.6 million to induce $9.2 million in federal contingency funds, while Georgia would spend $43.2 million to receive $39.2 million.

There is a great deal of interest in whether $2 billion is an adequate funding level. The figures presented in table 3 suggest that, over this period, the Contingency Fund would have been exhausted by early 1993: $732 million would have been requested but would not have been funded. The states affected include all nine seeking funds in 1993: California (seeking $410.1 million), New York ($220.3 million), Texas ($45.8 million), Washington ($43.2 million), Oregon ($17.1 million), Maine ($12.0 million), Illinois ($9.4 million), New Mexico ($8.8 million), and Nevada ($2.0 million).

Changes in States' Welfare Financing Burdens

Table 4 presents findings on changes in states' fiscal burdens, as measured by the share of welfare funding that they would be responsible for under PRWORA and the change in the volatility of state expenditures. Column 1 presents historical state funding shares (equal to one minus the federal shares reported in column 1 of table 1). Column 2 presents funding shares if the required minimum is spent on welfare, with no use of the Contingency Fund. Nearly all states experience a decrease in their overall responsibility for welfare funding under this assumption. At the aggregate level, the share of funding provided by states falls from 45 percent to 40 percent. The majority of states would experience declines of 5 percent to 10 percent in their share.

Column 3 recomputes states' funding shares if states are committed to fully funding their welfare programs to match total expenditures under AFDC. In most cases, the state funding share is higher than under AFDC. Nationally, the share of state funding rises by a modest 0.04 percentage point, to 0.49 percent. Fifteen states' shares would be about the same. The shares of some states (e.g., Wisconsin and Wyoming) would go up because the 80 percent MOE is so much higher than historical state spending. Most of the rest of the state shares would increase because full funding of welfare under the new regime would require significant new state contributions.

In column 4, the index of volatility of state expenditures is presented. First note that for the 12 states that are able to spend the MOE level in each period and still achieve full funding of their welfare programs, state expenditures do not vary and the index equals 0. For the remaining states, volatility increases. The states that experience the largest increases in volatility are Oklahoma, Kansas, and North Dakota. Only a few states (most significantly Colorado) are hit simultaneously with exceptionally large federal funding losses and large volatility changes.

---

Conclusions

This report has examined the hypothetical path of spending under PRWORA funding rules, had they been in place from 1987 through 1993. Because history will not repeat itself, the findings for specific states cannot be read as predictions; instead, they are illustrative descriptions of what happens when funding rules interact with specific state and federal spending patterns. Despite the limitations of this approach, however, several useful general insights emerge. This section summarizes the major findings and discusses the insights gained from this fiscal accounting exercise.

Findings

The significant findings follow.

• If economic trends in the period 1996-2002 mirror those from 1987-1993, total costs to the federal government under TANF and AFDC will be similar. Without considering Contingency Fund spending, TANF would cost only 4 percent less. Assuming an unlimited Contingency Fund, total costs under the two regimes are within 1.4 percent.
• Minimum mandated funding facing different states is extremely variable, implying huge diversity in states' fiscal constraints. Some states are required to spend far more under TANF than they would under AFDC. Other states could drop funding substantially. For example, over the period examined here, California could have cut welfare spending 25 percent without penalty, while Louisiana, Wyoming, and Indiana would all have been required to increase welfare spending by 25 percent. Although these requirements would cause some generally poorer states to bring their welfare program spending more in line with national averages, this is not always the case. For example, Kentucky and New Mexico, two very low-benefit states with high matching rates under the old regime, could cut welfare spending by 22.6 percent and 30 percent, respectively.
• States that qualify to use the Contingency Fund through the unemployment rate criterion are often not states experiencing particularly high expenditure growth.
• Even ignoring inflation, the $2 billion Contingency Fund would have been exhausted before 1993, leaving $732 million in desired claims that could not have been met.
• States could amass huge Treasury account balances, growing to a total of $5.7 billion in 1993—57 percent of all TANF block grants paid out that year.
• Not surprisingly, given the concentration of welfare recipients and expenditures in a minority of large states, Contingency Fund and Treasury account activity is dominated by a few large users. California and New York dominate the Contingency Fund, while Wisconsin and Michigan amass enormous Treasury accounts.
• States that benefit most from PRWORA will experience reduced volatility of welfare spending, as they usually need to spend only their MOE. Most other states, however, face increased volatility of their own welfare expenditures.

Interpretation

These findings have implications for the amount and distribution of federal and state welfare spending that may be expected under PRWORA. The following summarizes the major implications of the findings for current policy and suggests possible modifications that might be considered when PRWORA comes up for reauthorization in 2002.

• Although the nature of federal support has changed enormously in its qualities, it probably will change little in its quantity. The cost of welfare to the federal government under PRWORA differs little from the expense of AFDC in the time period studied.
• The distribution of federal funding among states is disparate and essentially arbitrary. The TANF block grants codify huge funding differentials across states for two reasons. First, large windfalls are based on the arbitrary criterion of how much states reduced welfare spending over the two years before the official beginning of welfare reform. Second, block grants are fixed over the horizon of the legislation, and the portion of federal funding that is responsive to states' evolving funding needs is extremely limited. Thought should be given to setting less arbitrary criteria when block grant amounts are reauthorized in 2002 and to schemes for adjusting block grants during the funding cycle in response to fiscal realities that are not easily manipulable by the states.
• Required state spending and required total spending are also disparate and arbitrarily determined. Although some poor states will be required to spend as much or more under PRWORA than in the past, presumably encouraging them to close the disparities between their programs and the rest of the country's, other poor states will be able to reduce welfare spending dramatically at their own discretion.
• Alternative criteria for Contingency Fund use should be considered. Many states experiencing extremely high expenditure growth from one year to the next are not eligible to use the Contingency Fund, while a relatively high proportion of states that are qualified to use the fund do not experience great expenditure growth. The alternative Food Stamp criterion might be a better indicator of need but it is only available with a long lag. Allowing the use of preliminary Food Stamp data, caseload growth in the TANF program, or other appropriate "flash" indicators as a supplement to the unemployment rate may be desirable.

---

References

Bartik, T.J., and R.W. Eberts. 1998. "Examining the Effect of Industry Trends and Structure on Welfare Caseloads." Mimeo (November). Kalamazoo, MI: W.E. Upjohn Institute for Employment Research.

Brueckner, J. 1998. "Welfare Reform and the Race to the Bottom: Theory and Evidence." Institute of Government and Public Affairs Working Paper #64 (October). University of Illinois at Urbana-Champaign.

Chernick, H. 1998. "Fiscal Effects of Block Grants for the Needy: An Interpretation of the Evidence." International Tax and Public Finance 5: 205-233.

Dye, R., and T. McGuire. 1991. "Growth and Variability of State Individual Income and General Sales Taxes." National Tax Journal 44(1): 55-66.

Dye R., and T. McGuire. 1998. "Fiscal Systems and Business Cycles: What Will Happen to State Welfare Spending When the Next Recession Occurs?" Mimeo (August). University of Illinois at Chicago.

Figlio, D.N., V.W. Kolpin, and W.E. Reid. 1997. "Do States Play Welfare Games?" Mimeo. Eugene, OR: Department of Economics, University of Oregon.

Powers, E.T. 1998. "Block Granting Welfare: The Outlook for State Budgets." National Tax Association Proceedings—1997, pp. 34-42. Meeting held in Chicago, Nov. 9-11, 1997.

Saavedra, L.A. 1998. "A Model of Welfare Competition with Evidence from AFDC." Institute of Government and Public Affairs Working Paper #63 (October). University of Illinois at Urbana-Champaign.

U.S. General Accounting Office (GAO). 1998. "Welfare Reform: Early Fiscal Effects of the TANF Block Grant." GAO-AIMD 98-137 (August). Washington, D.C.

---

Notes

1. The differing incentives under matching and block grants are discussed in Chernick (1998) and Brueckner (1998). Figlio et al. (1997) and Saavedra (1998) present empirical evidence that states are influenced by each other's welfare benefit choices.
2. Of 10 states recently examined by the U.S. General Accounting Office (GAO) (1998), half had chosen to "invest" all of their Temporary Assistance for Needy Families (TANF) grants immediately. Four states decided to set aside a portion of their welfare resources as a contingency against future caseload and cost increases.
3. The Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (PRWORA) was enacted in August 1996. The phase-in of state TANF programs lasted from August 1996 to July 1, 1997. Federal expenditures for each state during the phase-in were capped at the TANF block grant, regardless of when the state switched to TANF.
4. Although the formula determining the block grant amount included other factors, most states let 1994 expenditures determine the size of their block grants.
5. It is assumed throughout that states do not wish to opt out of the TANF system.
6. Although the TANF block grant supplants previous expenditures on Aid to Families with Dependent Children, Job Opportunities and Basic Skills, and Emergency Assistance, state spending on welfare-related child care also counts as maintenance-of-effort (MOE) expenditures, as do other expenditures that are reasonably calculated to accomplish the purpose of the TANF block grant. State expenditures must pass a "new spending" test to ensure that spending on unrelated programs is not credited toward MOE compliance. Failure to maintain the MOE will lead to a TANF block grant reduction equal to the shortfall from the MOE level in the succeeding year. Penalized states are obliged to make up the entire federal penalty amount with their own funds as well.
7. This will not be the case for states whose spending is constrained by the MOE spending level, however, in which case the variation in state spending will be smaller under PRWORA.
8. Because of long lags in data collection, states that rely on Food Stamp changes may find that they qualify for the fund long after a crisis has passed (GAO 1998).
9. This condition is even stricter than it appears, because the types of spending that count toward the MOE amount are more limited when a state is seeking a federal match from the Contingency Fund.
10. However, the effective federal matching rate declines precipitously if the state qualifies to use the fund for only part of the year.
11. An alternative would be to translate the unemployment conditions into relative terms.
12. For example, Chernick (1998) predicts that differentials between state benefits will widen under TANF because of the fact that poorer states face a greater effective increase in the price of welfare. However, he also considers evidence provided by others that states only react to these price effects with long lags (four to five years) to be credible. If so, the assumption here of a zero price elasticity of state spending may not be so unreasonable for considering the fiscal impact of the first PRWORA funding cycle.
13. In principle, TANF is not an entitlement program, although it remains to be seen how great a factor this will be in policymaking. Since the objective here is to investigate only the potential impact of the change in federal funding, states' potential to make policy changes that would make welfare spending more stable can be ignored, as can the impact of new features such as the 60-month lifetime limit on the variability of welfare expenditures.
14. A strong correlation between economic cycles and welfare spending is commonly assumed, although this finding is surprisingly fragile when individual states are examined (e.g., see Powers 1998; Dye and McGuire 1998; Bartik and Eberts 1998). Although this issue is discussed as appropriate, reconciling the findings of this literature is beyond the scope of this report.
15. The coefficient of variation equals the standard deviation divided by the mean. It is a unitless measure of how much a variable varies about its mean.
16. Measuring the complete additional burden of this increased volatility on states is complicated, since one must also consider the covariation of state welfare spending with state tax bases and states' other fiscal responsibilities. Dye and McGuire (1998) attempt to set state welfare expenditures in this context.
17. In 1991, average expenditure growth in qualifying states was 12.9 percent, compared with 12.3 percent in the remaining states. In 1992, the comparable figures are 12.5 percent (for the 18 qualifying states) and 8.6 percent (for the rest). In 1993, the comparable figures are 1.8 percent (for the 15 qualifying states) and 3.3 percent.
18. Bartik and Eberts (1998) find that the relationship between the aggregate unemployment rate and welfare cases is weak in Michigan but that other employment-related variables are good predictors of caseload growth. See also Dye and McGuire (1998), Powers (1998), and GAO (1998).
19. The Federal Loan Fund is not an additional source of federal funds, as states must repay all loans.
20. Note that the Contingency Fund monies can only be elicited with additional state expenditures. These "costs" to states are discussed below.
21. Note that the Contingency Fund would be exhausted even faster if the $2 billion had been deflated to 1987 prices.

---

Appendix A

---

About the Author

Elizabeth T. Powers has been an assistant professor of economics and a faculty member of the Institute of Government and Public Affairs at the University of Illinois at Urbana-Champaign since August 1996. From 1993 to 1996 she was an economist with the Federal Reserve Bank of Cleveland. From 1989 to 1990 she was a junior staff economist with President Bush's Council of Economic Advisers. Ms. Powers holds a Ph.D. in economics from the University of Pennsylvania. Her primary research interests are in the area of welfare policy.

---

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.

Usage, posting and reprint of materials on the UI web site:

Most publications may be downloaded free of charge from the web site in PDF format. This information may be used and copies made for research, academic, policy or other non-commercial purposes. Proper attribution is required.

Copyright of the written materials contained within the Urban Institute website is owned or controlled by the Urban Institute. Posting UI research papers on other websites is permitted subject to prior approval from the Urban Institute—contact paffairs@urban.org.

If you are unable to access or print the PDF document please contact us or call the Publications Office at (202) 261-5687.