No. 17 July 2001
Gaining Ground? Measuring the Impact of Welfare Reform on Welfare and Work
- Pollitt, Katha, 1996, “What We Know,” The New Republic (August).
- Although a number of studies have tried to measure the separate effects of policy changes and the economy on the decline in the welfare caseload during the 1990s, most of those studies have focused on the pre-TANF period of state waivers primarily because the data pertaining to TANF were not yet available. A few have included the initial period of TANF (for example, the report by the Council of Economic Advisers, The Effects of Welfare Policy and the Economic Expansion on Welfare Caseloads: An Update, Technical Report, August 3, 1999 and Robert F. Schoeni and Rebecca M. Blank, 2000, “What Has Welfare Reform Accomplished? Impacts on Welfare Participation, Employment, Incomes, Poverty and Family Structure,” NBER Working Paper 7627). Most of the studies deal exclusively with explaining the decline in welfare. The Schoeni and Blank paper is one of very few that has examined changes in work participation. The various studies typically use either caseload data or data aggregated on a state level and therefore cannot adjust very well for changes in the demographic characteristics of the population. Hill and O’Neill utilize micro data on individual single mothers from the Current Population Survey (CPS) which allow for more detailed and precise adjustment for demographic changes. In addition the analysis is extended to 2000 and therefore includes more years in the post-TANF period. Prior studies are summarized by Rebecca Blank in “Declining Caseloads/Increased Work: What Can We Conclude About the Effects of Welfare Reform?” forthcoming in a special conference volume of the Federal Reserve Bank of New York’s Economic Policy Review.
- See Frank Levy, 1979, “The Labor Supply of Female Heads, or AFDC Work Incentives Don’t Work Too Well,” Journal of Human Resources 14 (Winter); and Robert Moffitt, 1992, “Incentive Effects of the U.S. Welfare System: A Review,” Journal of Economic Literature 30 (March): 1–61.
- See O’Neill, June, 1990, Work and Welfare in Massachusetts: An Evaluation of the ET Program, (Boston: Pioneer Institute for Public Policy Research); and Janice Peskin, J. Topogna, and D. Marcotte, 1992, “How the Economy Affects AFDC Caseloads,” (paper presented at the annual meetings of APPAM, Denver, October); and Janice Peskin, 1993, “Forecasting AFDC Caseloads, with an Emphasis on Economic Factors,” Congressional Budget Office Staff Memorandum, July.
- Among the many empirical studies reaching this conclusion are: June O’Neill, Laurie J. Bassi, and Douglas A. Wolf, 1987, “The Duration of Welfare Spells,” Review of Economics and Statistics 69: 241–249; and Rebecca Blank, 1989, “Analyzing the Length of Welfare Spells,” Journal of Public Economics 39, 3 (August): 245-273; and David T. Ellwood, 1986, “Targeting ‘Would be’ Long-Term Recipients of AFDC,” Mathematica Policy Research.
- Data tabulated by the authors from the National Longitudinal Survey of Youth, 1979 cohort (NLSY79), show large differences between disadvantaged and other women in the likelihood of going on welfare and large differences in the total number of years on the welfare rolls accumulated by those who had ever been on welfare. Among current and former welfare recipients ages 33–37 in 1994, high school graduates had spent an average of 5.5 years on welfare over their lifetimes. But among high school dropouts the average was 7.8 years and among those who had never married it was 8.3 years. One-third of high school dropouts and 38 percent of never-married mothers were on welfare for more than 10 years compared to 18 percent of high school graduates. Also see the studies cited in note 4 above.
- Changes in percentages can be measured in absolute terms—i.e., as percentage point changes, or as percent changes. We use percentage point changes to describe the shifts in welfare and work participation. However, comparisons of percent changes can give a different ranking than percentage point changes.
- Jeff Grogger makes this point. See his “Time Limits and Welfare Use,” National Bureau of Economic Research Working Paper 7709, Cambridge, MA: NBER, 2000.
- A family cap refers to a state policy to determine the family’s welfare benefit based on the number of children born or conceived before the family went on welfare. States without a family cap generally provide an additional child benefit at the birth of a child even if the child was conceived while the mother was on welfare. With a family cap the family’s welfare benefit is not increased (or in some states with a partial cap, it is partially increased) when a child is born after the mother has been on welfare for 10 months or more. New Jersey enacted a full family cap under a state waiver in 1992. Appendix A lists the states that have implemented a family cap.
- Note that the measure of work participation in Figure 2 referred to the question asked in the March CPS concerning whether the person had worked at all in the prior calendar year. It is used there because it is juxtaposed against welfare participation which also refers to the prior calendar year. However, although Figures 8–12 also show work participation, they refer to current participation which is reported for the last week before the date of the March CPS survey interview. Thus, while the most recent data for welfare and work participation in the prior calendar year are for 1999, the data on work participation last week are available for March 2000.
- The percentage of married mothers on welfare is trivial and previously married mothers are much less likely to go on welfare than never-married mothers.
- The work participation of single mothers who are not on welfare is higher than that of all single mothers, a group which includes mothers still on welfare. For example, among single mothers, ages 18–44, who were high school dropouts and received no welfare in 1999, 74 percent worked at some point during the year and 61 percent worked 26 weeks or more. However, panel data are needed to measure directly the percentage of former welfare recipients who work. Data from the NLSY79 show that among those women who were on welfare in 1994 and were off welfare by 1998, 75 percent were working in the 1998 survey week. The percent employed was somewhat lower among recent leavers—those who left in 1997 or 1998 (69 percent). (If they left in 1995 and stayed off through 1998 their work participation was higher—85 percent.) High school dropouts had lower but still impressive employment rates. Of those who were on welfare in 1994 and off by 1998, 64 percent were employed in 1998 and among recent leavers the percentage was almost the same (63 percent). These panel data refer to women ages 29–37 in 1994 and 33–41 in 1998.
- For each state and in each year the wage rate was estimated from annual CPS data, as a weighted average of the full-time weekly wages of high school graduates and of workers with less than a high school education, ages 18 to 45.
- We investigated the effect of specifying the particular state waivers by redefining and expanding the waiver variable in the regression analysis to identify waivers with time limits and waivers without limits. Waivers with time limits have a clear negative effect on welfare participation, although the result is not quite statistically significant at standard levels.
- The response of the older single mothers to waivers actually had a positive sign (meaning that it increased welfare participation) though it was statistically insignificant. It is possible for waivers to increase welfare participation if the waiver had features that attracted welfare recipients such as providing substantial disregards for earnings in determining welfare benefits.
- The years 1996 and 1997 were transition years for TANF. Our analysis assumes that a state waiver continued until the month TANF was implemented, which did not occur in some states until late 1997 (or early 1998 in the case of California). The contribution of welfare reform shown in Tables 4–6 is the sum of the contribution of TANF and waivers in each period.
- The single exception to the pattern of a weaker role for welfare reform in the waiver period is for college-educated mothers for whom reform in the waiver period is estimated to account for 96 percent of the decline in participation (Table 5). However, this observation may be a statistical anomaly, the result of a larger-than-average regression coefficient for the waiver variable and only a 2.2 percentage point decline in welfare participation during the waiver period.
- In a recent paper Robert A. Moffitt and David Stevens use a different technique to examine whether the caseload has become disproportionately disadvantaged during the PRWORA period and conclude that no change has occurred except for a decrease in never-married women, a group that is typically more disadvantaged. See Changing Caseloads: Macro Influences and Micro Composition, a paper presented at the conference “Welfare Reform Four Years Later: Progress and Prospects,” Nov. 17, 2000.
- See Karen Goudreau, Howard Oberheu, and Vaughn Denton, 1984, “An Assessment of the Quality of Survey Reports of Income from the Aid to Families with Dependent Children (AFDC) Program,” Journal of Business & Economic Statistics, 2, 2 (April); and Dave O’Neill and June O’Neill, 1997, Lessons for Welfare Reform: An Analysis of the AFDC Caseload and Past Welfare-to-Work Program, (Kalamazoo, MI: W.E. Upjohn Institute for Employment Research).