Academic journal article Contemporary Economic Policy

Welfare Reform, Earnings, and Incomes: New Evidence from the Survey of Program Dynamics

Academic journal article Contemporary Economic Policy

Welfare Reform, Earnings, and Incomes: New Evidence from the Survey of Program Dynamics

Article excerpt


One of the primary goals of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (PRWORA) is to end welfare dependence (Title I, sec.401(a)(2)). However, implicit (and often explicit) in much of the discussion surrounding welfare reform is a loftier goal: that former welfare recipients will not only become self-sufficient but will rise out of poverty. (1) Families who rely on public assistance often live below the poverty line, so achieving this goal will require more than the simple replacement of public assistance income with earnings; it will require a substantial increase in recipients' total incomes. The key question addressed in this article is whether and how welfare reform has affected the earnings and incomes of families who receive cash assistance.

Although there have been many recent studies of the effects of welfare reform on a variety of outcomes, surprisingly few have used econometric analysis to study the impact on earnings and incomes. In fact, only 3 of the 33 econometric studies included in a recent synthesis of the literature (Grogger, Karoly, and Klerman 2002; hereafter, GKK) estimate the effect of welfare reform on earnings and only four examine the effect on incomes. Bitler et al. (2003a) provide a more recent example of the use of regression analysis with nonexperimental data to analyze this issue.

Given the relatively small number of econometric studies on the subject, most of the information about the effect of welfare reform on earnings and incomes comes from analyses of experimental data generated from random assignments into specific programs. The general consensus from this type of study is that some types of welfare reform did lead to higher earnings and incomes, at least in the short term. However, recipients by and large are not achieving self-sufficiency (GKK). These findings are confirmed by Bitler et al. (2003b), who estimate quantile treatment effects on experimental data from Connecticut's Jobs First program. The information gleaned from these studies is very valuable, but as with any methodology, random assignment experiments have drawbacks as well (see GKK, chap. 3). The authors seek to determine whether the findings from these studies are confirmed by econometric analysis of a relatively new nonexperimental data set, the Survey of Program Dynamics (SPD). (2)

The econometric studies cited by GKK are generally in agreement with the experimental studies. The estimates of the effect of welfare reform on earnings are generally positive but insignificant with a few exceptions. Moffitt (1999) finds a significant increase in earnings for women aged 16-54 with exactly a high school education, whereas Schoeni and Blank (2000) find a similar result for women in the same age group with less than high school. Grogger (2003) separates out the effect of time limits on welfare receipt from other types of reform and finds that time limits have no significant effect but other reforms do have a positive impact on the earnings of female household heads.

Each of the three studies discussed also examines the impact of welfare reform on family incomes, as does a fourth (Bitler et al. 2002). Again, most of the estimates are insignificant. However, two (Schoeni and Blank; Bitler et al.) find a positive effect of pre-PRWORA waivers for women with less than high school. (3) Grogger finds a positive effect of reforms other than time limits for a model measured in logs, but the effect is not significant when measured in levels.

All of these econometric studies rely on pooling cross-sections from the annual March supplements of the Current Population Survey (CPS). Analysis of the rich micro-data available in the CPS is extremely valuable but remains subject to the criticism that unobserved heterogeneity among individuals may bias the results.

The issue of unobserved individual heterogeneity has been addressed by Bitler et al. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.