Welfare Reform and the Cyclicality of Welfare Programs
by Elizabeth T. Powers
June 1996


Congress, the White House, and the nation’s governors all seem to agree that it’s time for a dramatic overhaul of the U.S. welfare system. What has been harder to come by is a consensus on the fairest and most politically feasible way to implement that goal. President Clinton vetoed the welfare provisions in both the budget reconciliation bill and the Republican-sponsored Personal Responsibility Act of 1995, while the National Governors Association continues to stand solidly behind similar provisions in its own reform initiative.

Although state and federal lawmakers may approach welfare reform from different perspectives and with different agendas, their most recent proposals share at least five characteristics: 1) the potential for huge reductions in expenditures, 2) heightened eligibility standards, 3) increased work requirements, 4) a shift away from entitlement status for benefits, and 5) a transfer of fiscal accountability from the federal government to the states. Any successful reform effort will likely incorporate all of these basic features.

It is important to understand that the nonwelfare elements of our social safety net—unemployment insurance (UI) and Social Security—have relatively stringent work and earnings history requirements. Thus, many households with a weak attachment to the labor force either are not eligible for these programs or qualify only for minimal benefits. Furthermore, UI benefits do not fully replace lost earnings and are time limited, meaning that even households whose workers have strong earnings records may turn to welfare programs during tough economic times.

Current reform bills would do little to address the cyclical variation in caseloads and expenditures. Because welfare would be funded primarily through capped block grants to the states, it would fall to state lawmakers to cope with increased program costs during economic downturns.1 Eliminating the entitlement status of benefits means that states would no longer be obligated to expand programs in times of greater need. Moreover, work incentives would likely be of limited value when unemployment is high.

To understand how potential welfare recipients would be affected by the proposed policy changes and to what extent funding shifts would exacerbate states’ budget problems, it is first necessary to examine the basic features of our current welfare system. This Economic Commentary presents an overview of the safety net’s four main programs, analyzes the impact of business cycle fluctuations on caseloads and expenditures, and reviews some provisions of the most recent reform proposals.

Return to Index

The Programs

The federal government administers a host of means-tested programs that pay cash or provide in-kind assistance. This article focuses on the most significant ones in terms of caseloads and outlays: Aid to Families with Dependent Children (AFDC), Food Stamps, Supplemental Security Income (SSI), and Medicaid. Figures 1 and 2 display historical trends in caseloads and total spending by program.

Figure 1: Welfare Recipients

[Welfare Recipients]

SOURCE: U.S. Department of Health and Human Services.

Figure 2: Welfare Expenditures

[Welfare Expenditures]

SOURCE: U.S. Department of Health and Human Services.

With one important exception, these programs do not even begin to match the level of non-means-tested cash transfers paid out through the Social Security system. Excluding Medicaid, combined state and federal welfare transfers were only 23 percent as large as Social Security payments in 1994.2 The exception, Medicaid, grew enormously in the past 10 years, with expenditures now nearly matching Medicare outlays (each program paid out over $140 billion in 1994).

One would not expect these programs to experience the same cyclical swings, since they target different groups of people. It is possible, however, to make an informed guess about their cyclicality by considering the features of each.

SSI: The SSI program provides cash aid to both elderly and disabled Americans who meet certain income and asset restrictions. Eligibility and benefits are based on federal poverty guidelines. Benefits are indexed for inflation and are the most generous of all the cash benefit programs. In 1994, SSI paid out more than $25 billion.

Participation in SSI expanded from around 4 million people in 1974 to about 6.4 million in 1994. Recent caseload growth stems entirely from an increase in disabled recipients. Since the elderly and disabled are presumably the least active members of the labor force, a reasonable conjecture is that SSI is the least cyclical of the welfare programs.

AFDC3: Since 1935, this program has provided cash assistance to single parents (usually women) meeting certain income and asset eligibility standards. Funding is provided by the federal government and the states, with eligibility and benefit levels varying significantly across state lines. Benefits are not indexed to inflation. In 1994, AFDC served about 5 million families (approximately 14 million people) and paid out almost $23 billion.

Although caseloads (measured as number of persons) expanded 43 percent between 1979 and 1994, real expenditures changed little over this period. AFDC caseloads and outlays are expected to rise during recessions for three reasons: female heads of household frequently participate in the labor force, unemployment appears to contribute to family breakups, and AFDC-UP benefits are contingent on unemployment. However, a sizable fraction of AFDC recipients are known to depend on the program regardless of the economy’s health.

Medicaid: Although the latest reform efforts would change this, AFDC and SSI recipients are now automatically enrolled in the Medicaid program. In most states, other individuals may be granted eligibility if their resources have been nearly exhausted by medical bills. In 1994, after a decade of explosive growth, more than 35 million people were covered by Medicaid for a total program cost of $143.5 billion. Variation in caseloads and expenditures over the business cycle should mimic those of AFDC and SSI (since these recipients account for most of the Medicaid caseload), but may be further influenced by economic conditions in states that extend eligibility to other groups.

Food Stamps: Founded in the early 1970s, the Food Stamp program is the only welfare program whose eligibility requirements are solely financial. Benefits are indexed, funded totally by the federal government, and uniform nationwide. In 1994, about $24 billion worth of benefits was paid out to about 27 million individuals. Because Food Stamps can be collected by all types of families, and because income-eligibility requirements are the least stringent for this program, one would expect caseloads and outlays to have the strongest association with cyclical swings in business activity.

Return to Index

Measuring Welfare Cyclicality

It is reasonable to expect welfare expenditures and caseloads to move in tandem with business cycles, since no program perfectly insures workers and their dependents against recessions. As mentioned above, UI is time limited, does not fully compensate for lost earnings, and excludes workers who do not have qualifying work and earnings histories. In addition, potential welfare recipients may be dependents (such as a child or elderly parent) of an unemployed worker who is either underinsured or lacks insurance altogether.

The strength of cyclical variation in welfare spending depends on several factors. First, Medicaid is an unusual case because many new enrollees come through the AFDC program and are presumably not ill. Therefore, caseloads may grow without a commensurate rise in expenditures. In other programs, it is likely that enrollees who sign on during a recession will receive a below-average payment, which means that spending will grow less quickly than the number of cases. The reason is that people with earnings records (that is, job losers) are more likely to receive other types of income that reduce their welfare benefits (Social Security, UI, alimony, and child support). Finally, at least for programs that are indexed, nominal expenditures must rise with inflation, so the response of nominal benefits also depends on the relationship between inflation and unemployment over the period of study.

Using standard methodology, I estimated the effect of higher unemployment on caseload and expenditure growth for each of the four major welfare programs.4 Although my approach has several limitations, evidence from other studies (discussed below) suggests that, overall, my methodology is conservative in that it tends to understate the relationship between economic conditions and welfare programs.5

First, consider the relationship between unemployment and caseload growth for the various programs. Caseload growth accelerates during recessions and slows during expansions, as expected. My findings confirm the intuition based on the individual programs’ features. A 1-percentage-point rise in unemployment would add 0.39 percentage point to the growth rate of SSI caseloads, 1.88 percentage points to Medicaid, 2.52 percentage points to AFDC, and 4.54 percentage points to the Food Stamp program.6 Put another way, if the 1994 jobless rate had been 1 percentage point higher, more than 1.2 million people would have been added to the Food Stamp rolls, 620,000 to Medicaid, 200,000 to AFDC, and 23,000 to SSI.

The relationship between nominal spending growth and unemployment follows much the same pattern. An additional 1-percentage-point rise in 1994’s jobless rate would have contributed 0.97 percentage point to the SSI expenditure growth rate, 2.05 percentage points to Medicaid, 3.00 percentage points to AFDC, and 4.84 percentage points to Food Stamps. In each case, nominal spending growth seems to respond more strongly to unemployment changes than do caseloads. These figures suggest that an additional percentage point of unemployment would have boosted annual expenditures by more than $3.3 billion in the case of Medicaid, over $1 billion for Food Stamps, nearly $235 million for SSI, and about $0.7 million for AFDC.

Return to Index

Current Proposals

Although President Clinton recently vetoed HR 4, the Republican-sponsored welfare reform bill, its fundamental features have resurfaced in a proposal backed by the nation’s governors.7 Both proposals emphasize capped block grants for the major welfare programs. Although states would have the option to spend more on welfare, they would also be allowed to slash payments up to 25 percent without penalty.

Contingency funds for recessions, emergencies, and population growth are probably unrealistically modest in their scope and funding. The governors’ plan would set aside $2 billion, allocated over seven years, for states experiencing increased unemployment—well below the needs suggested by recent experience. Furthermore, a state would become eligible for this money only after a substantial rise in its jobless rate. Even the very conservative estimates presented here suggest that moderately high but sustained unemployment could exhaust the contingency fund well before its seven-year horizon is reached.

Both proposals also place five-year limits on lifetime cumulative welfare receipts and eliminate entitlement status for groups currently served by Food Stamps and AFDC. Presumably, many who apply for aid during a recession will be turned away, particularly considering the block-grant nature of welfare funding. All of these changes would serve to severely limit access to welfare, exacerbating households’ financial difficulties during economic downturns.

Finally, the proposals emphasize the welfare-to-work philosophy by 1) financially rewarding states that effectively move people off public assistance and onto private payrolls, 2) increasing funds for child care and other work-related expenses, and 3) requiring able-bodied adult recipients to work (in public jobs if need be). However, these incentives and requirements would do little to control caseload growth during recessions, since many of those coming into the system would be unable to find work.

Return to Index


The degree of caseload and funding cyclicality across the nation’s welfare programs varies in ways that can be logically anticipated. SSI, serving the elderly and disabled, is the least sensitive to business cycle conditions, both in program use and outlays. Medicaid is moderately sensitive. AFDC, which mainly serves female-headed households, is somewhat sensitive, and Food Stamps, which bases eligibility primarily on financial need, is highly sensitive. According to the findings of other researchers using state-level data, even SSI participation and expenditures would rise significantly during a recession.8 Thus, the degree of cyclicality suggested in this Commentary is a conservative estimate.

It is safe to say that shifting funding responsibilities from Congress to the nation’s statehouses will impose formidable cyclical burdens on the states. With entitlement status for benefits gone, state lawmakers could react to a recession-induced fiscal squeeze by tightening standards or reducing benefits, which would pass cyclical income risk down to households. Without compensating changes in other programs (most obviously, UI), reduced payments and decreased opportunities for participation will inevitably worsen the impact of a faltering economy on the well-being of the poorest Americans.

Return to Index


1. Block grants are funds distributed to states to spend on certain items as they see fit. Many of the current welfare reform proposals fix (or "cap") the amount of money that will be granted over a five-year horizon.
Back to commentary

2. This is the most recent year for which data are available. More than $73 billion in welfare benefits (AFDC, Food Stamps, and SSI) was paid out, while the Social Security system provided over $312 billion in old age, survivors,’ and disability insurance.
Back to commentary

3. AFDC figures include the relatively recent AFDC-UP ("unemployed parent") program, which allows two-parent families meeting certain requirements to receive cash benefits.
Back to commentary

4. Caseload and expenditure growth rates were regressed on a constant and on the change in the unemployment rate.
Back to commentary

5. These limitations include ignoring nonemployment factors and using national rather than regional data. The former may have important effects on welfare caseloads and expenditures, some of which may be incorrectly attributed to unemployment. Using national data may obscure the relationship between unemployment and welfare expenditures and caseloads, because national trends are not equally distributed on an interregional basis.
Back to commentary

6. Caseloads are measured by the number of persons served, with the exception of AFDC, which refers to the number of families.
Back to commentary

7. The new bill is currently scheduled for consideration by the House Ways and Means Committee.
Back to commentary

8. One recent study estimates a 2 percent increase in SSI awards for every 1 percentage point increase in unemployment. See David Stapleton and K. Dietrich, "The Effects of the Business Cycle on Disability Applications and Awards," Lewin–VHI, mimeo, January 1995.
Back to commentary

Return to Index

Elizabeth T. Powers is an economist at the Federal Reserve Bank of Cleveland. The author thanks Michael Bryan for helpful comments and Kristin Roberts for excellent research assistance.

The views stated herein are those of the author and not necessarily those of the Federal Reserve Bank of Cleveland or of the Board of Governors of the Federal Reserve System.