Welfare Reform at 10: Analyzing Welfare Caseload Fluctuations, 1996-2002

Report Welfare

Welfare Reform at 10: Analyzing Welfare Caseload Fluctuations, 1996-2002

August 17, 2006 About an hour read Download Report
Michael New
Assistant Professor of Political Science at the University of Alabama

Ten years ago, President Bill Clinton signed land­mark welfare reform legislation into law. While pre­vious attempts at reform resulted in only cosmetic changes, the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) of 1996 has had a meaningful and lasting impact on the fed­eral welfare regime. PRWORA ended the entitlement status of Aid to Families with Dependent Children (AFDC) and replaced it with a time-limited assis­tance and work requirement program called Tempo­rary Assistance to Needy Families (TANF).

Most important, however, PRWORA gave states more leeway to structure their welfare administra­tions. Under PRWORA, states receive federal block grant allocations. These allocations allow states to use TANF funding in any manner reasonably calcu­lated to accomplish the purposes of TANF as long as the states maintain historical levels of spending agreed to in "maintenance of effort" plans. To con­tinue receiving their full federal TANF allocations, states must also conform to specific requirements regarding current recipients' work participation rates and length of time on the rolls.[1]

Although PRWORA passed by wide margins in both the House and Senate, it was still politically controversial. The Senate Minority Leader at the time, Tom Daschle (D-SD), opposed the bill, call­ing the work requirements "extremist." Likewise, House Minority Leader Richard Gephardt (D-MO) voted against the bill, citing an Urban Institute study that predicted that welfare reform would force more than 1 million children into poverty. Senator Daniel Patrick Moynihan (D-NY) was even more strident, declaring that the new law "was the most brutal act of social policy since Reconstruc­tion." He predicted, "Those involved will take this disgrace to their graves."[2]

Contrary to these alarming predictions, welfare reform went more smoothly than critics expected. A great deal of evidence demonstrates that welfare reform has been effective. For example:

  • By 1999, overall poverty and child poverty had substantially declined, with 4.2 million fewer people, including 2.3 million children, living in poverty than in 1996.[3]
  • Between 1996 and 2001, welfare caseloads were reduced by 58 percent.[4]
  • Between 1996 and 2002, the rate of increase in out-of-wedlock childbearing was reduced.[5]

Even some opponents of PRWORA have acknowledged the success of welfare reform. Wen­dell Primus, former Deputy Assistant Secretary in the Department of Health and Human Services, who resigned in protest after President Clinton signed the reform bill, remarked in 2001, "In many ways welfare reform is working better than I thought it would." He added, "The sky is not falling anymore. Whatever we have been doing during the past five years we ought to keep doing."[6]

However, a number of welfare reform opponents still stubbornly refuse to acknowledge its progress, crediting instead the economic boom during the late 1990s. Donna Shalala, who as Secretary of Health and Human Services urged President Clin­ton to veto the welfare reform bill, said, "What hap­pened on welfare reform was this combination of an economic boom and a political push to get people off the welfare rolls."[7]

Others who argued that the economy deserved most of the credit for the decline in caseloads, including Marian Wright Edelman of the Children's Defense Fund, expressed concern about what would happen during the most recent economic slow­down.[8] However, their arguments in favor of an eco­nomic explanation of welfare caseload changes do not hold up to empirical scrutiny. While the strength of the economy does affect the number of people receiving welfare, other economic expansions did not generate welfare caseload declines of similar magnitude. For instance, the economy expanded by 10.63 percent between 1993 and 1996, but the number of individuals receiving welfare declined by only 8.8 percent. Moreover, the economic expansion during the 1980s failed to reduce the total number of individuals receiving AFDC.[9] Finally, welfare caseloads increased dramatically during the eco­nomic boom during the mid to late 1960s, largely because benefits became more generous.[10]

Existing Research

If the booming economy is not responsible for the decline in welfare caseloads, what is? A consid­erable amount of research addresses this question. In 1999, the Council of Economic Advisers ana­lyzed the decline in welfare caseloads and con­cluded that the economy was responsible for 10 percent of the decline in registrants between 1996 and 1998. The authors argued that welfare reforms were responsible for approximately one-third of the decline and that the remainder was the conse­quence of other, unnamed factors.[11]

In 1999, The Heritage Foundation released a more detailed study of welfare caseload declines. The authors used multivariate regression analysis to analyze the percentage decline in welfare caseloads in each of the 50 states and the District of Columbia. They found substantial differences among the states in their policies toward welfare recipients who were not performing mandated work activities. In some states, recipients would lose their entire TANF check at the first instance of nonperformance. In other states, recipients could be assured of keeping almost their entire benefit check regardless of their conduct.[12]

The Heritage Foundation analysts found that the strength of state sanctioning policies had a major impact on the size of state welfare caseload declines. In general, the larger caseload reductions occurred in states with more stringent sanctions, and more modest declines took place in states with weaker sanctioning policies. The Heritage study also found that immediate work requirements led to declines in the number of individuals receiving welfare. Inter­estingly, the authors found that the strength of the economy, as measured by each state's average unem­ployment rate, did not have a statistically significant impact on caseload declines.[13]

In the summer of 2001, the Manhattan Institute released a study by June O'Neill and M. Anne Hill entitled "Gaining Ground? Measuring the Impact on Welfare and Work." It differed from most other stud­ies because the authors attempted to explain welfare caseload declines using survey data rather than whole-population data. O'Neill and Hill found that the implementation of the TANF program had a neg­ative and statistically significant effect on the proba­bility that a single woman would receive welfare benefits. They also found that the state waivers that preceded TANF negatively affected welfare partici­pation. The authors concluded that welfare reform is responsible for more than half of the decline in the welfare population since 1996.[14]

However, O'Neill and Hill neglected to consider other factors that likely played a role in the caseload declines. For instance, they did not consider the effect of the relative strength of state sanctions on the number of welfare recipients. In addition, while the authors held benefit levels constant in their regres­sion analysis, they did not elaborate on their findings. They also did not state whether they considered only benefits available through TANF or included benefits available to welfare recipients through other pro­grams, including food stamps, Medicare, and the Women, Infants, and Children program.

Another study that provides useful insights about welfare caseloads is William A. Niskanen's 1996 Cato Journal article "Welfare and the Culture of Poverty." Niskanen used 1992 data to examine the specific impact of welfare benefits on a variety of social pathol­ogies. Holding a variety of demographic, cultural, and economic factors constant, Niskanen found that increases in AFDC benefits led to statistically signifi­cant increases in the numbers of welfare recipients, people in poverty, births to single mothers, abortions, and violent crimes.[15] The article is useful to this anal­ysis because it provides evidence that higher levels of benefits lead to higher welfare caseloads.

A final study that examines welfare caseloads after the passage of welfare reform in 1996 was authored by Michael New and released by the Cato Institute during the summer of 2002. As in the 1999 Heritage Foundation study, the 2002 study found that the strength of sanctioning policies was strongly corre­lated with state welfare caseload declines. Similarly, it found that the strength of the economy had only a marginal impact on reductions in welfare caseloads. However, unlike the Heritage Foundation study, it considered the impact of benefit levels on welfare caseload declines and found statistically significant evidence that states with low levels of cash TANF benefits had larger welfare caseload declines.[16]

The New and Niskanen studies found that states with lower benefit levels had lower welfare case­loads. This is of interest because, historically, bene­fit levels have been a politically salient issue.

In his 1984 book Losing Ground, Charles Murray convincingly argued that increases in welfare bene­fits, which were legislated during the Great Society period, were largely responsible for the welfare caseload expansion that took place during the mid to late 1960s. According to Murray, before the increase in benefits, a woman facing an unplanned pregnancy had three basic choices. She could give the child up for adoption, get married, or fend for herself. However, when welfare benefits were increased, staying on welfare suddenly became an economically viable option for many unwed moth­ers. Not surprisingly, welfare caseloads and the number of single-parent families soared.[17] Since the evidence suggests that high welfare benefits led to an increase in welfare caseloads during the 1960s, it seems reasonable that an analysis of ben­efit levels might help to explain the decline in case­loads during the 1990s.

Revisiting the Topic

Previous and current research has identified three major factors that appear to affect fluctuations in welfare caseloads: the strength of sanctions, the performance of the economy, and the level of ben­efits. Statistical analysis could be useful in deter­mining which of these factors is most responsible for the decline in welfare caseloads since 1996.

Even though both the Heritage Foundation and the Cato Institute have examined this issue in studies released in 1999 and 2002, respectively, the topic is worth revisiting for several reasons. First, the Cato study, which is the more recent of the two, examined caseload declines up to August 2000.[18] Since then, more data on caseload levels have been released. Second, data from the U.S. Department of Health and Human Services and the U.S. General Accounting Office indicate that some states have changed their sanctioning poli­cies since 2000.[19]

In this analysis, I used state-level data to exam­ine the effects of sanctions, the economy, and ben­efits on welfare caseloads. A comparison of the states promised to prove fruitful because states had experienced varying amounts of success in reducing their welfare caseloads during the past 10 years.

For instance, between August 1996 and August 2002, Wyoming reduced its welfare caseload by over 93 percent. Conversely, Indiana's caseload actually increased by 3 percent over the same period. In addition, there were variations in the strength of state economies, the level of state bene­fits, and the stringency of state sanctioning policies. Because different state policies resulted in different outcomes, a proper analysis of these variables across the states should be able to identify the pol­icies that were the most responsible for substan­tially reducing welfare caseloads.

Sanctioning Policies. After the passage of wel­fare reform in 1996, all states adopted one of three types of sanctioning policies:

  1. Full family sanctioning. Some states sanction the entire TANF check at the first instance of nonperformance of required work or other activities. This is the strongest sanction that a state can impose.
  2. Graduated sanctioning. Other states do not sanction the entire TANF check at the first instance of nonperformance but do sanction the full TANF check after multiple infractions.
  3. Partial sanctioning. Some states sanction only the adult portion of the TANF check, even after repeated infractions. This enables recipients to retain the bulk of their TANF benefits even if they fail to perform workfare or other required activities.

Appendix A lists the sanctioning policies of each state and the years when they were in effect.

Analysis. To sort out the individual effects of sanctioning policies, benefit levels, and the econ­omy on declines in state welfare caseloads, two sep­arate sets of regressions were run. Regression analysis makes it possible to sort out the effects of each individual variable by holding constant the effects of all other variables. The first set of regres­sions examines why some states experienced larger welfare caseload declines than others between August 1996 and August 2002. The second set of regressions analyzes seven years of state caseload data to examine why some states have lower TANF caseloads than others.

First Regression Analysis: Caseload Decline 1996-2002

The first set of regressions examines why some states have experienced larger welfare caseload declines than others since the enact­ment of welfare reform. Nationally, the number of families receiving TANF has declined substantially, falling by approximately 60 percent between 1996 and 2002. However, some states have experienced con­siderably larger caseload declines than others. The TANF caseloads of Wyoming, Idaho, and several Mid­western states declined by well over 80 percent. Conversely, Indiana's TANF caseload actually increased slightly after passage of welfare reform. Similarly, Hawaii's caseload increased during the late 1990s until more stringent sanctioning policies were put in place.

Click to view table 1

The question remains: Why did states like Wyoming and Idaho expe­rience larger caseload declines than other states experienced? This first set of regressions attempts to provide some insights by analyzing the three factors identified in the academic lit­erature: the performance of the economy, the strength of sanctions on welfare recipients who are not complying with work activities, and the generosity of welfare benefits.

Dependent Variables. The regressions were run on two separate dependent variables: (1) the per­centage decline in the number of individuals receiving TANF between August 1996 and August 2002 and (2) the percentage decline in the number of families receiving TANF between August 1996 and August 2002.[20]

Independent Variables. The regressions ana­lyzed the effects of five different independent variables.

FullSanction measures the number of years between August 1996 and August 2002 that a state had a full family sanction in force.[21]

GraduatedSanction measures the number of years between August 1996 and August 2002 that a state enforced a graduated sanction.[22]

Income Growth measures the real growth of state per capita personal income between 1996 and 2002. This was designed to capture the relative strength of each state's economy.[23]

Benefits measures the average level of TANF cash benefits as a percentage of state per capita income available to a single mother with two children from 1996 to 2002.[24]

Caseload1996 measures the percentage of the state population (individuals or families, depend­ing on the dependent variable) that was receiving AFDC in August 1996. It seems likely that states with relatively more people on welfare could reduce their caseloads more easily than states with relatively few people on welfare could.

The Results. The results are consistent with other studies that have examined welfare caseload declines. In both regressions, states with full sanc­tions experienced the largest caseload declines. For every year that a state had a full sanction in place, the welfare caseload declined by slightly more than 3 percent compared to a state with a partial sanction. That means that over six years, a state with a full sanction would see its caseload decline by more than 18 percent compared to a state with only a partial sanction. This finding is statistically significant.

Furthermore, for every year that a state had a graduated sanction in place, its caseload declined by slightly more than 2 percent compared to a state with a partial sanction. This finding is also statisti­cally significant. Overall, these findings add to the body of evidence in the policy and social science literature that strong sanctions are correlated with large declines in welfare caseloads.

The only other variable in this set of regressions that reaches statistical significance is the percentage of the population that received AFDC in 1996. States with a high AFDC population in 1996 enjoyed more success in reducing their caseloads than did states with a low AFDC population. This is unsurprising. A state with a low caseload might already have had success in lowering its welfare rolls prior to 1996, and those remaining on the welfare rolls might be those who have a more diffi­cult time making the transition from welfare to work. Conversely, if a state has a high welfare case­load, it seems likely that it has more welfare recipi­ents who could be persuaded more easily to leave welfare and obtain employment.

Finally, some evidence indicates that states with strong economic growth between 1996 and 2002 experienced larger caseload declines; however, this finding failed to achieve statistical significance. States with low TANF benefits between 1996 and 2002 also experienced larger caseload declines than states with high TANF benefits. However, the coefficient is small and fails to reach statistical sig­nificance. It should be noted that this variable mea­sures only cash benefits. Individuals and families receiving TANF are also eligible for a variety of other non-cash benefits including Medicaid, food stamps, and housing subsidies. If the value of these benefits could be included in the regression model, it might show a stronger correlation between low benefits and welfare caseload declines.

Second Regression Analysis: Caseload Levels 1996-2002

To further this analysis, another set of regres­sions was run. In this case, the dependent variables measure caseload levels rather than caseload declines to examine why some states have smaller percentages of people receiving TANF than others have. For instance, in 2002, only 0.18 percent of Idaho residents were receiving TANF compared to over 7 percent of the residents of Washington, D.C. Overall, analyzing the percentage of people receiv­ing TANF should provide additional insights into welfare caseload fluctuations.

Furthermore, this analysis of caseload levels nicely complements this paper's earlier analysis of caseload declines for several reasons. First, simply analyzing caseload declines could be misleading. Some states could have experienced small caseload declines simply because they had relatively few welfare recipients prior to the passage of PRWORA. Similarly, states with large welfare caseloads in 1996 might have experienced large declines but still have caseload levels that are considerably higher than those of other states.

Analyzing caseload levels offers additional advan­tages. We have seven years of data on caseload levels after the passage of welfare reform, so we have more data to analyze. Furthermore, analyzing caseload levels might grant additional insights into the effects of sanctions, benefits, and the economy on main­taining low caseloads after they decline.

In this analysis, two sets of regressions were run. In the first regression, the dependent variable is the percentage of each state's population that was receiving TANF benefits. In the second regression, the dependent variable is the percentage of each state's families that was receiving TANF benefits. The independent variables are similar to the ones used in the first set of regressions. FullSanction, an indicator variable, equals 1 if a state has imple­mented a full family sanction that year and zero otherwise. Similarly, GraduatedSanction is 1 if a state has implemented a graduated sanction that year and zero otherwise. Personal Income Growth measures the growth in state personal income for that year. Finally, TANF Benefit measures the cash benefits welfare available to a single mother with two children as a percentage of state per capita income. The results are presented in Table 2.

Click to View Table 2

This set of regression results provides further evidence that strong sanctioning policies effectively reduce welfare caseloads and keep caseloads low. The findings indicate that states with stronger sanctioning policies have a lower percentage of individuals and families receiving welfare than states with weak sanc­tions have. These findings achieve statistical significance.

There is also statistically signifi­cant evidence that welfare caseloads fluctuate with the strength of the economy. Unsurprisingly, caseloads fall during times of strong economic growth and rise when the economy slows. Finally, there is statistically significant evidence that states with low cash TANF benefits have a lower percentage of people receiving wel­fare than states with high cash TANF benefits have.

Overall, even though welfare ben­efit levels and economic growth had relatively little to do with the large decline in welfare caseloads since 1996, it appears that they do affect year-to-year fluctuations in welfare caseloads. This should be of interest to policymakers who desire to keep welfare case­loads low.

Conclusion

Welfare reform was one of the leading public pol­icy stories of the 1990s. In the 10 years since Con­gress enacted welfare reform in 1996, the number of people receiving welfare has been cut by nearly 60 percent, and both poverty and hunger have declined.[25] This decline in welfare caseloads has attracted a great deal of attention, and many scholars have attempted to explain the large declines in wel­fare caseloads. Some states experienced considerably larger caseload declines than others experienced. As a result, many studies analyzing the success of wel­fare reform have paid close attention to program dif­ferentiation among the states.

Many of those studies have presented a number of important insights into why welfare caseloads declined so sharply after welfare reform. However, shortcomings are evident in much of the research. Prior analyses of welfare reform indicate that three factors influence welfare caseload fluctuations: the strength of sanctions, the level of benefits, and the strength of the economy. However, almost all of the cited studies omit one or more of these factors from their analysis. In addition, since many studies con­sider caseload declines over a limited period of time since the passage of reform, they are unable to distinguish between policies that cause short-term fluctuations and those that lead to long-term declines.

This study breaks new ground in several ways.

First, the use of multivariate regression analysis makes it possible to consider the effects of the economy, sanctions, and TANF benefits simulta­neously and to determine which factors have had the most impact.

Second, although many other studies consider caseload declines for a short period of time after reform, this study tracks caseload declines for six years. Using a longer time frame increases the cer­tainty that the various factors are having a long-term impact on caseloads and are not simply caus­ing a temporary decline.

Finally, this study also analyzes both caseload levels and caseload declines. This provides more data to analyze and offers insights into the effective­ness of sanctions in maintaining and preserving low caseloads levels.

Overall, the most important finding is that the strength of state sanctioning policies had the largest impact on both caseload declines and caseload lev­els between 1996 and 2002. The other variables that were considered, including the strength of the economy and TANF benefit levels, had some effect on year-to-year caseload levels but played only a minor role in the large decline in welfare caseloads between 1996 and 2002.

For example, the regression model estimates that differences in sanctioning policies result in a 20 percentage point difference in caseload declines. Conversely, holding other factors constant, the model estimates that the difference in caseload decline between a state with a strong economy and a state with a weak economy is only about 3 per­centage points.[26] Similarly the difference in case­load decline between a state with high TANF cash benefits and a state with low TANF cash benefits is only about 1 percentage point.[27]

Michael J. New, Ph.D., is Visiting Fellow at The Heritage Foundation and Assistant Professor of Politi­cal Science at the University of Alabama. The author would like to thank Mark Jackson and Calley Means for their help with data collection.

Click to view Appendix



[1] Lisa Oliphant, "Four Years of Welfare Reform: A Progress Report," Cato Institute Policy Analysis No. 378, August 22, 2000, p. 2, at /static/reportimages/C6A190E312DDE01A71589BAE1CB2D922.pdf (August 11, 2006).

[2] Editorial, "Welfare as They Know It," The Wall Street Journal, August 29, 2001, p. A14.

[3]Joseph Dalaker and Bernadette D. Proctor, Poverty in the United States 1999, U.S. Bureau of the Census, Current Popula­tion Reports: Consumer Income, P60-210, September 2000, p. B2, at /static/reportimages/7C643D94B5DE71D76B303FAD2B1B1E44.pdf (August 11, 2006).

[4]U.S. Department of Health and Human Services, Administration for Children and Families, "Statistics," at http://www.acf.dhhs.gov/news/stats/aug-dec.htm (August 11, 2006).

[5]Stephanie J. Ventura and Christine A. Bachrach, "Nonmarital Childbearing in the United States, 1940-99," National Vital Statistics Reports,Vol. 48, No. 16 (October 18, 2000), pp. 1-2, at /static/reportimages/FE85ACCC9BA24CF55C93F887DB6A1E2F.pdf (August 11, 2006).

[6]Quoted in Blaine Harden, "2-Parent Families Rise After Change in Welfare Laws," The New York Times, August 12, 2001, p. A1.

[7]Editorial, "Welfare as They Know It."

[8]Ibid.

[9]In 1983, 10.9 million individuals were receiving AFDC; by 1989, 12.1 million individuals were receiving AFDC. That is a caseload increase of 11 percent. U.S. Bureau of the Census, Statistical Abstract of the United States: 1992 (Washing­ton: U.S. Government Printing Office, 1992).

[10]Editorial, "Welfare as They Know It."

[11]Council of Economic Advisers, "The Effects of Welfare Policy and the Economic Expansion on Welfare Caseloads: An Update," August 3, 1999, executive summary, p. 1, at http://clinton4.nara.gov/WH/EOP/CEA/html/welfare (August 14, 2006).

[12]Robert E. Rector and Sarah E. Youssef, "The Determinants of Welfare Caseload Decline," Heritage Foundation Center for Data Analysis Report No. 99-04, May 11, 1999, pp. 1-3, at http://www.heritage.org/research/welfare/CDA99-04.cfm.

[13] Ibid., p. 6.

[14]June O'Neill and M. Anne Hill, "Gaining Ground: Measuring the Impact of Welfare Reform on Welfare and Work," Man­hattan Institute Civic Report No. 17, July 17, 2001, at /static/reportimages/737FE8240EF04CAD7E0373D2657B2CDF.pdf (August 14, 2006).

[15]William Niskanen, "Welfare and the Culture of Poverty," Cato Journal,Vol. 16, No. 1 (Spring-Summer 1996), at http://www.cato.org/pubs/journal/cj16n1-1.html (August 14, 2006).

[16]Michael J. New, "Welfare Reform That Works: Explaining the Caseload Decline 1996-2000," Cato Institute Policy Analysis No. 435, May 7, 2002, at /static/reportimages/06BEE2B091355B7A931ADFF301FA24D3.pdf (August 14, 2006).

[17]Charles Murray, Losing Ground (New York: Basic Books, 1984), pp. 154-66, 244, and 263.

[18]New, "Welfare Reform That Works."

[19]Gil Crouse, "State Implementation of Major Changes to Welfare Policies, 1992-1998," U.S. Department of Health and Human Services, 1999, Table W-3, at http://aspe.hhs.gov/hsp/Waiver-Policies99/policy_CEA.htm (August 11, 2006); U.S. General Accounting Office, Welfare Reform: State Sanction Policies and Number of Families Affected, GAO/HEHS-00-44, March 2000, pp. 44-47, at /static/reportimages/CD151B127A55FAA06611307104426165.pdf (August 11, 2006); and U.S. Department of Health and Human Services, Administration for Children and Families, Temporary Assistance for Needy Families Sixth Annual Report to Con­gress, November 2004, Chap 12, p. 18, Table 12-8, at /static/reportimages/828A4C665B90A1F84E5C9232E60C82FC.pdf (August 15, 2006).

[20]Caseload data that are exactly four years apart are used to ensure that regional seasonal variation in caseloads does not bias the findings.

[21]Crouse, "State Implementation of Major Changes to Welfare Policies," and U.S. General Accounting Office, Welfare Reform, pp. 44-47.

[22]The names of the categories of sanctions are taken from Crouse, "State Implementation of Major Changes to Welfare Policies"; U.S. General Accounting Office, Welfare Reform; and U.S. Department of Health and Human Services, Tem­porary Assistance for Needy Families Sixth Annual Report to Congress.

[23]Data are from U.S. Department of Commerce, Bureau of Economic Analysis, Regional Economic Accounts, Local Area Personal Income, Table CA1-3, at http://www.bea.gov/bea/regional/reis/default.cfm?catable=CA1-3 (August 15, 2006).

[24]Data on monthly TANF benefits are from Committee on Ways and Means, U.S. House of Representatives, The 2000 Green Book: Background Material and Data on Programs Within the Jurisdiction of the Committee on Ways and Means, 17th ed., October 6, 2000, Section 7, at aspe.hhs.gov/2000gb (August 14, 2006). This variable is in the form of a ratio to account for the differences in the cost of living between states.

[25]However, many of the people who have left the welfare rolls are still dependent on various transfer programs. The challenge of transition to self-sufficiency has not yet been met. See Oliphant, "Four Years of Welfare Reform."

[26]This calculation was made by using the regression results to compare the welfare caseload decline in a state with per­sonal income growth at the 25th percentile to the caseload decline in a state with personal income growth at the 75th percentile, all other factors being equal.

[27]This calculation was made by using the regression results to compare the welfare caseload decline in a state with cash TANF benefits (as a percentage of state per capita income) at the 25th percentile to the caseload decline in a state with cash TANF benefits at the 75th percentile, all other factors being equal.

Ten years ago, President Bill Clinton signed land­mark welfare reform legislation into law. While pre­vious attempts at reform resulted in only cosmetic changes, the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) of 1996 has had a meaningful and lasting impact on the fed­eral welfare regime. PRWORA ended the entitlement status of Aid to Families with Dependent Children (AFDC) and replaced it with a time-limited assis­tance and work requirement program called Tempo­rary Assistance to Needy Families (TANF).

Most important, however, PRWORA gave states more leeway to structure their welfare administra­tions. Under PRWORA, states receive federal block grant allocations. These allocations allow states to use TANF funding in any manner reasonably calcu­lated to accomplish the purposes of TANF as long as the states maintain historical levels of spending agreed to in "maintenance of effort" plans. To con­tinue receiving their full federal TANF allocations, states must also conform to specific requirements regarding current recipients' work participation rates and length of time on the rolls.[1]

Although PRWORA passed by wide margins in both the House and Senate, it was still politically controversial. The Senate Minority Leader at the time, Tom Daschle (D-SD), opposed the bill, call­ing the work requirements "extremist." Likewise, House Minority Leader Richard Gephardt (D-MO) voted against the bill, citing an Urban Institute study that predicted that welfare reform would force more than 1 million children into poverty. Senator Daniel Patrick Moynihan (D-NY) was even more strident, declaring that the new law "was the most brutal act of social policy since Reconstruc­tion." He predicted, "Those involved will take this disgrace to their graves."[2]

Contrary to these alarming predictions, welfare reform went more smoothly than critics expected. A great deal of evidence demonstrates that welfare reform has been effective. For example:

  • By 1999, overall poverty and child poverty had substantially declined, with 4.2 million fewer people, including 2.3 million children, living in poverty than in 1996.[3]
  • Between 1996 and 2001, welfare caseloads were reduced by 58 percent.[4]
  • Between 1996 and 2002, the rate of increase in out-of-wedlock childbearing was reduced.[5]

Even some opponents of PRWORA have acknowledged the success of welfare reform. Wen­dell Primus, former Deputy Assistant Secretary in the Department of Health and Human Services, who resigned in protest after President Clinton signed the reform bill, remarked in 2001, "In many ways welfare reform is working better than I thought it would." He added, "The sky is not falling anymore. Whatever we have been doing during the past five years we ought to keep doing."[6]

However, a number of welfare reform opponents still stubbornly refuse to acknowledge its progress, crediting instead the economic boom during the late 1990s. Donna Shalala, who as Secretary of Health and Human Services urged President Clin­ton to veto the welfare reform bill, said, "What hap­pened on welfare reform was this combination of an economic boom and a political push to get people off the welfare rolls."[7]

Others who argued that the economy deserved most of the credit for the decline in caseloads, including Marian Wright Edelman of the Children's Defense Fund, expressed concern about what would happen during the most recent economic slow­down.[8] However, their arguments in favor of an eco­nomic explanation of welfare caseload changes do not hold up to empirical scrutiny. While the strength of the economy does affect the number of people receiving welfare, other economic expansions did not generate welfare caseload declines of similar magnitude. For instance, the economy expanded by 10.63 percent between 1993 and 1996, but the number of individuals receiving welfare declined by only 8.8 percent. Moreover, the economic expansion during the 1980s failed to reduce the total number of individuals receiving AFDC.[9] Finally, welfare caseloads increased dramatically during the eco­nomic boom during the mid to late 1960s, largely because benefits became more generous.[10]

Existing Research

If the booming economy is not responsible for the decline in welfare caseloads, what is? A consid­erable amount of research addresses this question. In 1999, the Council of Economic Advisers ana­lyzed the decline in welfare caseloads and con­cluded that the economy was responsible for 10 percent of the decline in registrants between 1996 and 1998. The authors argued that welfare reforms were responsible for approximately one-third of the decline and that the remainder was the conse­quence of other, unnamed factors.[11]

In 1999, The Heritage Foundation released a more detailed study of welfare caseload declines. The authors used multivariate regression analysis to analyze the percentage decline in welfare caseloads in each of the 50 states and the District of Columbia. They found substantial differences among the states in their policies toward welfare recipients who were not performing mandated work activities. In some states, recipients would lose their entire TANF check at the first instance of nonperformance. In other states, recipients could be assured of keeping almost their entire benefit check regardless of their conduct.[12]

The Heritage Foundation analysts found that the strength of state sanctioning policies had a major impact on the size of state welfare caseload declines. In general, the larger caseload reductions occurred in states with more stringent sanctions, and more modest declines took place in states with weaker sanctioning policies. The Heritage study also found that immediate work requirements led to declines in the number of individuals receiving welfare. Inter­estingly, the authors found that the strength of the economy, as measured by each state's average unem­ployment rate, did not have a statistically significant impact on caseload declines.[13]

In the summer of 2001, the Manhattan Institute released a study by June O'Neill and M. Anne Hill entitled "Gaining Ground? Measuring the Impact on Welfare and Work." It differed from most other stud­ies because the authors attempted to explain welfare caseload declines using survey data rather than whole-population data. O'Neill and Hill found that the implementation of the TANF program had a neg­ative and statistically significant effect on the proba­bility that a single woman would receive welfare benefits. They also found that the state waivers that preceded TANF negatively affected welfare partici­pation. The authors concluded that welfare reform is responsible for more than half of the decline in the welfare population since 1996.[14]

However, O'Neill and Hill neglected to consider other factors that likely played a role in the caseload declines. For instance, they did not consider the effect of the relative strength of state sanctions on the number of welfare recipients. In addition, while the authors held benefit levels constant in their regres­sion analysis, they did not elaborate on their findings. They also did not state whether they considered only benefits available through TANF or included benefits available to welfare recipients through other pro­grams, including food stamps, Medicare, and the Women, Infants, and Children program.

Another study that provides useful insights about welfare caseloads is William A. Niskanen's 1996 Cato Journal article "Welfare and the Culture of Poverty." Niskanen used 1992 data to examine the specific impact of welfare benefits on a variety of social pathol­ogies. Holding a variety of demographic, cultural, and economic factors constant, Niskanen found that increases in AFDC benefits led to statistically signifi­cant increases in the numbers of welfare recipients, people in poverty, births to single mothers, abortions, and violent crimes.[15] The article is useful to this anal­ysis because it provides evidence that higher levels of benefits lead to higher welfare caseloads.

A final study that examines welfare caseloads after the passage of welfare reform in 1996 was authored by Michael New and released by the Cato Institute during the summer of 2002. As in the 1999 Heritage Foundation study, the 2002 study found that the strength of sanctioning policies was strongly corre­lated with state welfare caseload declines. Similarly, it found that the strength of the economy had only a marginal impact on reductions in welfare caseloads. However, unlike the Heritage Foundation study, it considered the impact of benefit levels on welfare caseload declines and found statistically significant evidence that states with low levels of cash TANF benefits had larger welfare caseload declines.[16]

The New and Niskanen studies found that states with lower benefit levels had lower welfare case­loads. This is of interest because, historically, bene­fit levels have been a politically salient issue.

In his 1984 book Losing Ground, Charles Murray convincingly argued that increases in welfare bene­fits, which were legislated during the Great Society period, were largely responsible for the welfare caseload expansion that took place during the mid to late 1960s. According to Murray, before the increase in benefits, a woman facing an unplanned pregnancy had three basic choices. She could give the child up for adoption, get married, or fend for herself. However, when welfare benefits were increased, staying on welfare suddenly became an economically viable option for many unwed moth­ers. Not surprisingly, welfare caseloads and the number of single-parent families soared.[17] Since the evidence suggests that high welfare benefits led to an increase in welfare caseloads during the 1960s, it seems reasonable that an analysis of ben­efit levels might help to explain the decline in case­loads during the 1990s.

Revisiting the Topic

Previous and current research has identified three major factors that appear to affect fluctuations in welfare caseloads: the strength of sanctions, the performance of the economy, and the level of ben­efits. Statistical analysis could be useful in deter­mining which of these factors is most responsible for the decline in welfare caseloads since 1996.

Even though both the Heritage Foundation and the Cato Institute have examined this issue in studies released in 1999 and 2002, respectively, the topic is worth revisiting for several reasons. First, the Cato study, which is the more recent of the two, examined caseload declines up to August 2000.[18] Since then, more data on caseload levels have been released. Second, data from the U.S. Department of Health and Human Services and the U.S. General Accounting Office indicate that some states have changed their sanctioning poli­cies since 2000.[19]

In this analysis, I used state-level data to exam­ine the effects of sanctions, the economy, and ben­efits on welfare caseloads. A comparison of the states promised to prove fruitful because states had experienced varying amounts of success in reducing their welfare caseloads during the past 10 years.

For instance, between August 1996 and August 2002, Wyoming reduced its welfare caseload by over 93 percent. Conversely, Indiana's caseload actually increased by 3 percent over the same period. In addition, there were variations in the strength of state economies, the level of state bene­fits, and the stringency of state sanctioning policies. Because different state policies resulted in different outcomes, a proper analysis of these variables across the states should be able to identify the pol­icies that were the most responsible for substan­tially reducing welfare caseloads.

Sanctioning Policies. After the passage of wel­fare reform in 1996, all states adopted one of three types of sanctioning policies:

  1. Full family sanctioning. Some states sanction the entire TANF check at the first instance of nonperformance of required work or other activities. This is the strongest sanction that a state can impose.
  2. Graduated sanctioning. Other states do not sanction the entire TANF check at the first instance of nonperformance but do sanction the full TANF check after multiple infractions.
  3. Partial sanctioning. Some states sanction only the adult portion of the TANF check, even after repeated infractions. This enables recipients to retain the bulk of their TANF benefits even if they fail to perform workfare or other required activities.

Appendix A lists the sanctioning policies of each state and the years when they were in effect.

Analysis. To sort out the individual effects of sanctioning policies, benefit levels, and the econ­omy on declines in state welfare caseloads, two sep­arate sets of regressions were run. Regression analysis makes it possible to sort out the effects of each individual variable by holding constant the effects of all other variables. The first set of regres­sions examines why some states experienced larger welfare caseload declines than others between August 1996 and August 2002. The second set of regressions analyzes seven years of state caseload data to examine why some states have lower TANF caseloads than others.

First Regression Analysis: Caseload Decline 1996-2002

The first set of regressions examines why some states have experienced larger welfare caseload declines than others since the enact­ment of welfare reform. Nationally, the number of families receiving TANF has declined substantially, falling by approximately 60 percent between 1996 and 2002. However, some states have experienced con­siderably larger caseload declines than others. The TANF caseloads of Wyoming, Idaho, and several Mid­western states declined by well over 80 percent. Conversely, Indiana's TANF caseload actually increased slightly after passage of welfare reform. Similarly, Hawaii's caseload increased during the late 1990s until more stringent sanctioning policies were put in place.

Click to view table 1

The question remains: Why did states like Wyoming and Idaho expe­rience larger caseload declines than other states experienced? This first set of regressions attempts to provide some insights by analyzing the three factors identified in the academic lit­erature: the performance of the economy, the strength of sanctions on welfare recipients who are not complying with work activities, and the generosity of welfare benefits.

Dependent Variables. The regressions were run on two separate dependent variables: (1) the per­centage decline in the number of individuals receiving TANF between August 1996 and August 2002 and (2) the percentage decline in the number of families receiving TANF between August 1996 and August 2002.[20]

Independent Variables. The regressions ana­lyzed the effects of five different independent variables.

FullSanction measures the number of years between August 1996 and August 2002 that a state had a full family sanction in force.[21]

GraduatedSanction measures the number of years between August 1996 and August 2002 that a state enforced a graduated sanction.[22]

Income Growth measures the real growth of state per capita personal income between 1996 and 2002. This was designed to capture the relative strength of each state's economy.[23]

Benefits measures the average level of TANF cash benefits as a percentage of state per capita income available to a single mother with two children from 1996 to 2002.[24]

Caseload1996 measures the percentage of the state population (individuals or families, depend­ing on the dependent variable) that was receiving AFDC in August 1996. It seems likely that states with relatively more people on welfare could reduce their caseloads more easily than states with relatively few people on welfare could.

The Results. The results are consistent with other studies that have examined welfare caseload declines. In both regressions, states with full sanc­tions experienced the largest caseload declines. For every year that a state had a full sanction in place, the welfare caseload declined by slightly more than 3 percent compared to a state with a partial sanction. That means that over six years, a state with a full sanction would see its caseload decline by more than 18 percent compared to a state with only a partial sanction. This finding is statistically significant.

Furthermore, for every year that a state had a graduated sanction in place, its caseload declined by slightly more than 2 percent compared to a state with a partial sanction. This finding is also statisti­cally significant. Overall, these findings add to the body of evidence in the policy and social science literature that strong sanctions are correlated with large declines in welfare caseloads.

The only other variable in this set of regressions that reaches statistical significance is the percentage of the population that received AFDC in 1996. States with a high AFDC population in 1996 enjoyed more success in reducing their caseloads than did states with a low AFDC population. This is unsurprising. A state with a low caseload might already have had success in lowering its welfare rolls prior to 1996, and those remaining on the welfare rolls might be those who have a more diffi­cult time making the transition from welfare to work. Conversely, if a state has a high welfare case­load, it seems likely that it has more welfare recipi­ents who could be persuaded more easily to leave welfare and obtain employment.

Finally, some evidence indicates that states with strong economic growth between 1996 and 2002 experienced larger caseload declines; however, this finding failed to achieve statistical significance. States with low TANF benefits between 1996 and 2002 also experienced larger caseload declines than states with high TANF benefits. However, the coefficient is small and fails to reach statistical sig­nificance. It should be noted that this variable mea­sures only cash benefits. Individuals and families receiving TANF are also eligible for a variety of other non-cash benefits including Medicaid, food stamps, and housing subsidies. If the value of these benefits could be included in the regression model, it might show a stronger correlation between low benefits and welfare caseload declines.

Second Regression Analysis: Caseload Levels 1996-2002

To further this analysis, another set of regres­sions was run. In this case, the dependent variables measure caseload levels rather than caseload declines to examine why some states have smaller percentages of people receiving TANF than others have. For instance, in 2002, only 0.18 percent of Idaho residents were receiving TANF compared to over 7 percent of the residents of Washington, D.C. Overall, analyzing the percentage of people receiv­ing TANF should provide additional insights into welfare caseload fluctuations.

Furthermore, this analysis of caseload levels nicely complements this paper's earlier analysis of caseload declines for several reasons. First, simply analyzing caseload declines could be misleading. Some states could have experienced small caseload declines simply because they had relatively few welfare recipients prior to the passage of PRWORA. Similarly, states with large welfare caseloads in 1996 might have experienced large declines but still have caseload levels that are considerably higher than those of other states.

Analyzing caseload levels offers additional advan­tages. We have seven years of data on caseload levels after the passage of welfare reform, so we have more data to analyze. Furthermore, analyzing caseload levels might grant additional insights into the effects of sanctions, benefits, and the economy on main­taining low caseloads after they decline.

In this analysis, two sets of regressions were run. In the first regression, the dependent variable is the percentage of each state's population that was receiving TANF benefits. In the second regression, the dependent variable is the percentage of each state's families that was receiving TANF benefits. The independent variables are similar to the ones used in the first set of regressions. FullSanction, an indicator variable, equals 1 if a state has imple­mented a full family sanction that year and zero otherwise. Similarly, GraduatedSanction is 1 if a state has implemented a graduated sanction that year and zero otherwise. Personal Income Growth measures the growth in state personal income for that year. Finally, TANF Benefit measures the cash benefits welfare available to a single mother with two children as a percentage of state per capita income. The results are presented in Table 2.

Click to View Table 2

This set of regression results provides further evidence that strong sanctioning policies effectively reduce welfare caseloads and keep caseloads low. The findings indicate that states with stronger sanctioning policies have a lower percentage of individuals and families receiving welfare than states with weak sanc­tions have. These findings achieve statistical significance.

There is also statistically signifi­cant evidence that welfare caseloads fluctuate with the strength of the economy. Unsurprisingly, caseloads fall during times of strong economic growth and rise when the economy slows. Finally, there is statistically significant evidence that states with low cash TANF benefits have a lower percentage of people receiving wel­fare than states with high cash TANF benefits have.

Overall, even though welfare ben­efit levels and economic growth had relatively little to do with the large decline in welfare caseloads since 1996, it appears that they do affect year-to-year fluctuations in welfare caseloads. This should be of interest to policymakers who desire to keep welfare case­loads low.

Conclusion

Welfare reform was one of the leading public pol­icy stories of the 1990s. In the 10 years since Con­gress enacted welfare reform in 1996, the number of people receiving welfare has been cut by nearly 60 percent, and both poverty and hunger have declined.[25] This decline in welfare caseloads has attracted a great deal of attention, and many scholars have attempted to explain the large declines in wel­fare caseloads. Some states experienced considerably larger caseload declines than others experienced. As a result, many studies analyzing the success of wel­fare reform have paid close attention to program dif­ferentiation among the states.

Many of those studies have presented a number of important insights into why welfare caseloads declined so sharply after welfare reform. However, shortcomings are evident in much of the research. Prior analyses of welfare reform indicate that three factors influence welfare caseload fluctuations: the strength of sanctions, the level of benefits, and the strength of the economy. However, almost all of the cited studies omit one or more of these factors from their analysis. In addition, since many studies con­sider caseload declines over a limited period of time since the passage of reform, they are unable to distinguish between policies that cause short-term fluctuations and those that lead to long-term declines.

This study breaks new ground in several ways.

First, the use of multivariate regression analysis makes it possible to consider the effects of the economy, sanctions, and TANF benefits simulta­neously and to determine which factors have had the most impact.

Second, although many other studies consider caseload declines for a short period of time after reform, this study tracks caseload declines for six years. Using a longer time frame increases the cer­tainty that the various factors are having a long-term impact on caseloads and are not simply caus­ing a temporary decline.

Finally, this study also analyzes both caseload levels and caseload declines. This provides more data to analyze and offers insights into the effective­ness of sanctions in maintaining and preserving low caseloads levels.

Overall, the most important finding is that the strength of state sanctioning policies had the largest impact on both caseload declines and caseload lev­els between 1996 and 2002. The other variables that were considered, including the strength of the economy and TANF benefit levels, had some effect on year-to-year caseload levels but played only a minor role in the large decline in welfare caseloads between 1996 and 2002.

For example, the regression model estimates that differences in sanctioning policies result in a 20 percentage point difference in caseload declines. Conversely, holding other factors constant, the model estimates that the difference in caseload decline between a state with a strong economy and a state with a weak economy is only about 3 per­centage points.[26] Similarly the difference in case­load decline between a state with high TANF cash benefits and a state with low TANF cash benefits is only about 1 percentage point.[27]

Michael J. New, Ph.D., is Visiting Fellow at The Heritage Foundation and Assistant Professor of Politi­cal Science at the University of Alabama. The author would like to thank Mark Jackson and Calley Means for their help with data collection.

Click to view Appendix



[1] Lisa Oliphant, "Four Years of Welfare Reform: A Progress Report," Cato Institute Policy Analysis No. 378, August 22, 2000, p. 2, at /static/reportimages/C6A190E312DDE01A71589BAE1CB2D922.pdf (August 11, 2006).

[2] Editorial, "Welfare as They Know It," The Wall Street Journal, August 29, 2001, p. A14.

[3]Joseph Dalaker and Bernadette D. Proctor, Poverty in the United States 1999, U.S. Bureau of the Census, Current Popula­tion Reports: Consumer Income, P60-210, September 2000, p. B2, at /static/reportimages/7C643D94B5DE71D76B303FAD2B1B1E44.pdf (August 11, 2006).

[4]U.S. Department of Health and Human Services, Administration for Children and Families, "Statistics," at http://www.acf.dhhs.gov/news/stats/aug-dec.htm (August 11, 2006).

[5]Stephanie J. Ventura and Christine A. Bachrach, "Nonmarital Childbearing in the United States, 1940-99," National Vital Statistics Reports,Vol. 48, No. 16 (October 18, 2000), pp. 1-2, at /static/reportimages/FE85ACCC9BA24CF55C93F887DB6A1E2F.pdf (August 11, 2006).

[6]Quoted in Blaine Harden, "2-Parent Families Rise After Change in Welfare Laws," The New York Times, August 12, 2001, p. A1.

[7]Editorial, "Welfare as They Know It."

[8]Ibid.

[9]In 1983, 10.9 million individuals were receiving AFDC; by 1989, 12.1 million individuals were receiving AFDC. That is a caseload increase of 11 percent. U.S. Bureau of the Census, Statistical Abstract of the United States: 1992 (Washing­ton: U.S. Government Printing Office, 1992).

[10]Editorial, "Welfare as They Know It."

[11]Council of Economic Advisers, "The Effects of Welfare Policy and the Economic Expansion on Welfare Caseloads: An Update," August 3, 1999, executive summary, p. 1, at http://clinton4.nara.gov/WH/EOP/CEA/html/welfare (August 14, 2006).

[12]Robert E. Rector and Sarah E. Youssef, "The Determinants of Welfare Caseload Decline," Heritage Foundation Center for Data Analysis Report No. 99-04, May 11, 1999, pp. 1-3, at http://www.heritage.org/research/welfare/CDA99-04.cfm.

[13] Ibid., p. 6.

[14]June O'Neill and M. Anne Hill, "Gaining Ground: Measuring the Impact of Welfare Reform on Welfare and Work," Man­hattan Institute Civic Report No. 17, July 17, 2001, at /static/reportimages/737FE8240EF04CAD7E0373D2657B2CDF.pdf (August 14, 2006).

[15]William Niskanen, "Welfare and the Culture of Poverty," Cato Journal,Vol. 16, No. 1 (Spring-Summer 1996), at http://www.cato.org/pubs/journal/cj16n1-1.html (August 14, 2006).

[16]Michael J. New, "Welfare Reform That Works: Explaining the Caseload Decline 1996-2000," Cato Institute Policy Analysis No. 435, May 7, 2002, at /static/reportimages/06BEE2B091355B7A931ADFF301FA24D3.pdf (August 14, 2006).

[17]Charles Murray, Losing Ground (New York: Basic Books, 1984), pp. 154-66, 244, and 263.

[18]New, "Welfare Reform That Works."

[19]Gil Crouse, "State Implementation of Major Changes to Welfare Policies, 1992-1998," U.S. Department of Health and Human Services, 1999, Table W-3, at http://aspe.hhs.gov/hsp/Waiver-Policies99/policy_CEA.htm (August 11, 2006); U.S. General Accounting Office, Welfare Reform: State Sanction Policies and Number of Families Affected, GAO/HEHS-00-44, March 2000, pp. 44-47, at /static/reportimages/CD151B127A55FAA06611307104426165.pdf (August 11, 2006); and U.S. Department of Health and Human Services, Administration for Children and Families, Temporary Assistance for Needy Families Sixth Annual Report to Con­gress, November 2004, Chap 12, p. 18, Table 12-8, at /static/reportimages/828A4C665B90A1F84E5C9232E60C82FC.pdf (August 15, 2006).

[20]Caseload data that are exactly four years apart are used to ensure that regional seasonal variation in caseloads does not bias the findings.

[21]Crouse, "State Implementation of Major Changes to Welfare Policies," and U.S. General Accounting Office, Welfare Reform, pp. 44-47.

[22]The names of the categories of sanctions are taken from Crouse, "State Implementation of Major Changes to Welfare Policies"; U.S. General Accounting Office, Welfare Reform; and U.S. Department of Health and Human Services, Tem­porary Assistance for Needy Families Sixth Annual Report to Congress.

[23]Data are from U.S. Department of Commerce, Bureau of Economic Analysis, Regional Economic Accounts, Local Area Personal Income, Table CA1-3, at http://www.bea.gov/bea/regional/reis/default.cfm?catable=CA1-3 (August 15, 2006).

[24]Data on monthly TANF benefits are from Committee on Ways and Means, U.S. House of Representatives, The 2000 Green Book: Background Material and Data on Programs Within the Jurisdiction of the Committee on Ways and Means, 17th ed., October 6, 2000, Section 7, at aspe.hhs.gov/2000gb (August 14, 2006). This variable is in the form of a ratio to account for the differences in the cost of living between states.

[25]However, many of the people who have left the welfare rolls are still dependent on various transfer programs. The challenge of transition to self-sufficiency has not yet been met. See Oliphant, "Four Years of Welfare Reform."

[26]This calculation was made by using the regression results to compare the welfare caseload decline in a state with per­sonal income growth at the 25th percentile to the caseload decline in a state with personal income growth at the 75th percentile, all other factors being equal.

[27]This calculation was made by using the regression results to compare the welfare caseload decline in a state with cash TANF benefits (as a percentage of state per capita income) at the 25th percentile to the caseload decline in a state with cash TANF benefits at the 75th percentile, all other factors being equal.

Authors

Michael New

Assistant Professor of Political Science at the University of Alabama