May 9, 1997

A Report by the
Council of Economic Advisers


During the first four years of the Clinton Administration, from January 1993 to January 1997, the number of individuals receiving welfare fell by 20 percent, or 2.75 million recipients the largest decline in over 50 years. Three potential explanations for this decline are (1) economic growth, which created 12 million new jobs over the period, (2) Federal waivers, which allowed 43 states to experiment with innovative ideas to help reduce welfare dependency, and (3) other policies affecting work-related incentives, including the 1990 and 1993 expansions of the Earned Income Tax Credit (EITC) and the recent rise in federal and state spending on child care. It is important to determine the causes of this decline in light of the recently enacted welfare reform legislation. If economic growth was the major contributor, then continued growth seems essential for further progress in moving people from welfare to work. If federal policies played a significant role, however, then continued efforts along these lines are likely to lead to additional reductions. A statistical analysis (described in the companion technical paper to this report) shows that over 40 percent of the decline resulted from a falling unemployment rate associated with the economic expansion and almost one-third from statewide welfare reform waivers (Figure 1). Other factors (which might include other policy initiatives, such as the EITC) account for the remainder.

Figure 1
Reasons for the
Decline in Welfare
Caseloads, 1993-1996

[Figure 1]


Welfare caseloads tend to fluctuate over the business cycle, rising when the economy moves into recession and declining once a recovery is underway and the economy is expanding. For example, the proportion of the population receiving welfare fell during the expansion of the late 1970s and rose as the economy went into recession in 1980 (Figure 2). Between 1989 and 1993, the proportion of the population receiving welfare shot up 25 percent, reaching its highest level ever. The recession of 1990-1991 and the weak labor market through 1992 certainly contributed to this increase, hindering the efforts of those welfare recipients seeking work. One might be tempted to argue that the subsequent decline between 1993 and 1996 simply reflected the normal return to work of welfare recipients who were unable to find jobs when the economy was weak.

[Figure 2]

The business cycle alone, however, is unlikely to account for the entire decline in welfare recipiency after 1993. The 1990-1991 recession was relatively mild; the annual unemployment rate peaked at 7.5 percent in 1992, much lower than the peak rates in the 1974-75 and 1981-82 recessions. It seems improbable that a moderate recession would lead to such severe swings in the rate of welfare receipt. Moreover, some states with large reductions in their unemployment rate during this period did not experience big drops in their welfare caseload, while other states saw a big drop in welfare receipt even though their unemployment decline was moderate (see attached map). For that reason it is important to look at other factors, including the possible impact of changes in welfare programs during that time.


Aid to Families with Dependent Children (AFDC) was the Nation's primary welfare program until last year. The AFDC program was administered by the states, subject to Federal requirements. Since 1962, the Secretary of Health and Human Services has had the authority to waive some of these requirements if states proposed experimental or pilot programmatic changes that furthered the goals of the AFDC program. The Bush Administration was the first to use this authority extensively, especially in its final year. But the Clinton Administration expanded the number of waivers dramatically after 1993, granting waivers to a total of 43 states.

Waivers granted to states to implement experimental welfare policies generally contained a number of provisions that varied greatly in scope. Some were pilot programs that could not have had much effect on the size of a state's overall welfare caseload. Others covered a larger share of the state's welfare population but included some relatively minor provisions that probably had little effect on the number of welfare recipients statewide. Six broad categories of waivers that potentially might have had an observable effect in reducing state welfare caseloads are:

The number of states with statewide waivers of these types rose dramatically between 1993 and 1996 (Figure 3). Some states that experienced large drops in welfare receipt are also states that received waivers (see attached map).

[Figure 3]


Several factors besides economic conditions and waivers are likely to affect the rate of welfare receipt. An increase in female-headed families will tend to increase this rate because the welfare system strongly favors single mothers with children. The generosity of welfare benefits also may affect the number of poor individuals who seek benefits. Labor market returns for less-skilled workers, national changes in welfare policy, and cultural attitudes towards welfare receipt, also may play a role. The task of a statistical analysis is to disentangle the separate effects of these factors in order to identify the relationship between each of them and welfare receipt.

The exercise reported here uses state-level data from 1976 through 1996 to estimate the contributions of economic growth (measured by the change in the unemployment rate) and approved state waivers to the recent decline in welfare receipt. The use of state level data allows us to control for changes that affect welfare receipt across the entire country at a point in time, such as national changes in welfare policy. The relationship between, say, economic conditions and the rate of welfare receipt can still be identified because recessions tend to be worse in some parts of the country than in others and could lead to differences across states in patterns of welfare receipt. Using data over several years allows us to control for long-run differences in welfare receipt that exist across states. The relationship between waivers and welfare receipt, for example, can be observed by following changes in welfare receipt within a state before and after the waiver. Using techniques like these, a statistical analysis can estimate the effects of economic activity and waivers on the size of the welfare rolls holding other things that affect welfare receipt constant.

An Example

Figure 4 presents a comparison of Florida and Georgia that is intended to provide some intuition for the statistical methodology and the manner in which the effects of economic activity are estimated separately from other potential confounding factors. It should not be considered a rigorous test. The figure plots the difference between the two states in unemployment rates between 1984 and 1996 and in the share of the population receiving AFDC over the same period. Taking the difference between the two states in each year controls for any differences that affect both states simultaneously. Because neither state received a waiver until late in the 1996 fiscal year, the difference in trends through virtually all of this time period are unaffected by differences in waiver provisions or their effectiveness. Throughout most of the expansion of the middle to late 1980s, unemployment in Georgia had been somewhat higher than in Florida. When the 1990-91 recession hit, unemployment in Florida rose considerably relative to that in Georgia, and the difference has been slow to recede. Subsequently, AFDC receipt shows an increase in Florida relative to Georgia. The full statistical analysis uses this sort of approach to identify the effects of both waivers and economic activity on the rate of welfare receipt in all states over time.

[Figure 4]

The Timing of the Welfare Caseload Response

A number of other tests were conducted to explore more complicated relationships between economic activity, waivers, and the welfare caseload, particularly the possibility that impacts on the rate of welfare receipt might not be contemporaneous with changes in unemployment or implementation of waivers:


The results of this analysis indicate a strong relationship between the welfare caseload and both economic activity and Federal welfare waivers.

These findings say nothing about the outcomes for those individuals who otherwise would have collected benefits had waivers not been granted. Additional research that can determine how individuals fared under the alternative waiver provisions, rather than an aggregate analysis examining the statewide caseload, clearly is desirable to help address this issue.