IntroductionIn this article, I suggest an alternative approach to explaining recent changes in welfare caseloads. Most existing research uses pooled or time series methods. These studies can account for change over time, but in some cases they come up with implausible findings. They also do not identify which specific policies caused change, and they ignore the role of governmental quality. These problems seem due largely to an inability to measure state policies accurately. I suggest a simpler, cross-sectional approach that gives up the time dimension but permits measuring policy and governmental influences more accurately. It leads to more plausible findings, specifies more policy details, and highlights the role of governmental quality in successful reform. 1 The national caseload of Aid to Families with Dependent Children (AFDC, since 1996 Temporary Assistance for Needy Families [TANF]) rose 34% between 1989 and 1994, then fell 56% through June 2000 (data from U.S. Administration for Children and Families). These are the sharpest changes in the history of the program. The fall predates the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (PRWORA), the federal law that recast AFDC as TANF, but PRWORA seems to have accelerated it. The research into the causes has been done mostly by economists using either time series or pooled analyses based on state panel data. These studies say something about the causes of change, but they pose several problems. Some authors admit that despite impressive statistical apparatus, they cannot explain the recent changes well (Blank, 1997;Wallace & Blank, 1999). Some also attribute the recent caseload changes mostly to economic conditions (Congressional Budget Office, 1993; Ziliak, Figlio, Davis, & Connelly, 1997;Bell, 2001). A phenomenon closely related to the caseload decline is that work levels among poor parents have risen sharply. In 1993, only 43% of poor female heads of family reported that they worked at all, but in 1999, 60% did so (U.S. Bureau of the Census, 1994Census, -2000. One prominent study gives most of the credit to the Earned Income Tax Credit (EITC), a subsidy for the working poor that has risen recently (Meyer & Rosenbaum, 1999).However, these findings conflict with what close observers of welfare reform say about the causes of change. Journalists tend to think welfare reform is at least as important as economic conditions in driving the rolls decline, and perhaps more so (e.g., DeParle, 1997). State and local officials carrying out the reform think they have mostly caused the decline, although they also credit the economy (Nathan & Gais, 2001). In a study that asked local Wisconsin officials about the reasons for caseload fall, welfare reform was mentioned most often, then the economy; no one mentioned the EITC (Mead, 1999, pp. 601-604). One reason to doubt that the economy is the leading cause of the decline is that tight labor markets in the late 1980s did not have the same effect on the rolls.
2Some economic studies credit t...