The nonpartisan Urban Institute publishes studies, reports, and books on timely topics worthy of public consideration. The views expressed are those of the authors and should not be attributed to the Urban Institute, its trustees, or its funders.
This report was prepared for the U. S. Department of Health and Human Services' Office of the Assistant Secretary for Planning and Evaluation under Contract No. HHS-100-99-0003 Task Order No. 06. The views expressed are those of the authors and should not be attributed to the U.S. Department of Health and Human Services or to the Urban Institute, its trustees, or its funders.
Note: The full report is available in its entirety in the Portable Document Format (PDF). The accompanying appendices for this report are also available in the Portable Document Format (PDF) in their entirety as a separate document.
Acknowledgements
This study was commissioned by the U.S. Department of Health and Human Services (DHHS), Assistant Secretary for Planning and Evaluation (ASPE). Oversight and review were provided by Elizabeth Lower-Basch, our Project Officer. We thank Elizabeth, Julie Isaacs, and Canta Pian from ASPE for their guidance and helpful comments. We also thank Ann Burek and Alan Yaffe from the DHHS Administration for Children and Families for helpful comments.
The study benefited from the experience and advice provided by members of a technical work group, convened for a day of discussion in Washington, DC, and the comments many of them continued to provide after the discussion. The technical work group members included Marc Bendick, Peter Germanis, Jeff Grogger, Irene Lurie, Marcia Meyers, Robert Moffitt, Joe Soss, Alan Weil, and Martha Zaslow.
We gratefully acknowledge the advice and comments from Urban Institute researchers Greg Acs, Linda Giannarelli, Caroline Ratcliffe, Doug Wissoker, and Janine Zweig. Special gratitude goes to Gretchen Rowe for her oversight of the Welfare Rules Databse, and her patient and frequent interpretations of data we extracted from it. We also thank Neal Parikh and Shinta Herwantoro for excellent research assistance.
The opinions and conclusions expressed are solely those of the authors and should not be construed as representing the opinions or policy of any agency of the Federal Government or the Urban Institute.
Table of Contents
Chapter I — Introduction to the Study
Chapter II — Preliminary Typologies of State TANF and Related Policies: Development and Rationale
Chapter III — Database Description
Chapter IV — Analysis of Welfare Policies Hypothesized to Affect Recipient Job Entry
Chapter V — Conclusions and Recommendations for Future Work
Appendix I — Full Annotated Bibliography
Appendix II — Summary of Technical Work Group Meeting
Appendix III — TANF Typologies Database Data Document
Appendix IV — TANF Typologies Database Data Dictionary
Appendix V — Deferred Variables
Chapter I — Introduction to the Study
Table of Contents
- Origins of the Study
- Methodology
- Literature Review Findings
- Technical Work Group
- Preliminary Typologies
- Direct vs. Indirect and Short-run vs. Long-run Effects
- Summary Variables
- Deferred Variables
- Organization of this Report
Origins of the Study
Since the early 1990s, researchers and policymakers have had an increasingly difficult time understanding how states and sub-state areas operate their cash assistance programs for needy families. Under Aid to Families with Dependent Children (AFDC), the federal government determined most program rules, though states established their own rules in some areas such as income eligibility limits, benefit levels, and eligibility requirements for two-parent families. While the federal government offered states this finite number of options, those options fit within a program structure that did not vary substantially across the states. Importantly, any differences in AFDC programs across states could be identified with reasonable accuracy using the AFDC State Plans, as well as documents prepared by the Administration for Children and Families (ACF) in the Department of Health and Human Services (DHHS).
By the mid-1990s, however, the complexity of welfare programs' policies and rules began increasing, as more states were granted waivers to experiment with new policies, such as family caps and time limits on benefit receipt. This increase in state waivers began shifting the control of cash assistance programs from the federal government to the states. While the waiver policies were described in the Waiver Terms and Conditions agreed upon by the state and the DHHS, these were often incomplete. Moreover, changes in implementation details and schedules often occurred after agreement. Further, the waivers were not included in the State Plans. And finally, most waivers were for pilot programs in sub-state areas—not statewide—which increased the variation within states and the difficulties in tracking program rules.
This devolution culminated in the passage of the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) of August 1996, which replaced the AFDC program with Temporary Assistance for Needy Families (TANF). The legislation produced further variation in state welfare programs and increased difficulty in tracking program rules and understanding program outcomes across states. Overall, the level of detail in TANF State Plans varies greatly across states and generally offers insufficient detail on eligibility, benefit computation, requirements imposed on clients, and incentives offered to move them toward independence from cash assistance. This kind of information can only be obtained from state regulations, caseworker manuals, and/or interviews with state, and especially local, policy officials and practitioners. Without it, researchers are highly challenged to analyze the effects of policies on outcomes.
Given these difficulties associated with identifying state TANF policies, it became desirable to create a method of organizing the information in a way that would serve many purposes, primarily to organize, describe, and analyze the complex policy choices states have made. The Urban Institute developed the Welfare Rules Database (WRD) to provide a single location where information on program rules regarding eligibility, benefits, time limits, work requirements, sanctions, and more could be researched across states and across time, without the need to consult multiple documents. The State Policy Documentation Project (SPDP) is another source for state TANF policy, though less detailed and longitudinal than the WRD.
The next logical step is to develop a method for categorizing state TANF and related policies into typologies that facilitate analysis of links between them and outcomes, and that facilitate analysis that clusters states along broad policy dimensions. The goal of this current study is to take that next step—to lay the groundwork for facilitating research across states linking state TANF policies to outcomes, by grouping TANF and related policies into selected summary and independent variables thought to affect a selected outcome. In this study we develop six typologies of TANF and related policies, build a public-use database based on the typologies, and analyze one of the typologies using factor and cluster analysis.
Methodology
To initiate the work, we conducted an extensive literature review searching for other researchers' typologies and their analysis clustering states or linking typologies to outcomes. We then convened a Technical Work Group of researchers across the country with expertise in welfare research and forays into categorizing TANF policies.
Literature Review Findings
We found fewer existing typologies than we had hoped, but did uncover several promising typologies—such as those by the Committee for Economic Development (2000), U.S. General Accounting Office (2000), Meyers, Gornick, and Peck (2001), Soss et al. (2000), and a conceptual model by Weil (2002). These and other relevant articles are summarized in our annotated bibliography, located in Appendix I.A.
Briefly stated, the CED authors' work seemed to have potential for adaptation in our typology development. They identify 24 welfare policies that either Require Work or Limit Eligibility, create Financial Incentives to Work, or Provide Services Supporting Work. They categorize states, then, by whether they do not use at all, use moderately, or make extensive use of each of those policies. Finally, they create indices of policies requiring work and supporting work, by state. The indices are derived from the proportion of policies a state uses under the "requiring work" rubric, and the proportion of policies they use under the combined rubrics of "financial incentives to work" and "services supporting work." Each state is assigned a zero for no use of the policy, one point for moderate use, and two points for extensive use. The policy data were drawn from a 1998 National Governors' Association summary of selected state policy elements. We considered using the typology and substituting policy data from the Urban Institute's Welfare Rules Database.
The GAO's (2000) categorization of state sanction and conciliation policies were considered for combining, for example, with Rector and Youssef's (1999) immediate work requirement variable to test a link with job entry rates. The GAO sanction data time period is not perfect, as they record sanction policies as of September 1999 but number of families sanctioned each month of calendar year 1998. The GAO data are complete in that, for the most part, they cover 50 states and DC.
The Meyers, Gornick and Peck (2000) typology was studied for its measure of how state policies affect a child poverty rate outcome. The authors used calendar year 1994 for administrative data (a combination of data from the Welfare Rules Database, the House Committee on Ways and Means "Green Book," and DHHS reports), and data on policy characteristics from sources including the National Governors' Association, the Children's Defense Fund, and the Center on Budget and Policy Priorities. The authors have since updated their paper with more recent data. We considered using the typology and their updated data, if available, or building a similar typology using the Welfare Rules Database.
For a more complex typology, the Soss et al. (2000) "Index of Policy Severity," was a candidate for adoption. It includes whether or not a state has: a time limit to a work requirement of less than 24 months; a lifetime limit of less than 60 months; a family cap; and whether the severity of its sanctions are weak, moderate, or strong. This typology is a compelling one for analyzing its effect on caseload and job entry rate. It uses data from calendar year 1997 for most of its policy variables. Finally, we were drawn to a model developed by Weil (2002) for a typology of key welfare rules. He develops a 12-cell grid containing a three-point scale of barriers to continued TANF receipt (without work) and a 4-point scale of incentives for families to leave TANF or combine TANF with work.
Technical Work Group
We consulted researchers and experts across the country, most of whom had already developed selected typologies of some TANF and related rules for their individual research studies. They came together for a one-day session of brainstorming and recommendations on how we could most productively take the next steps toward TANF policy typologies development.
The Technical Work Group made dozens of recommendations that are reflected in the decisions made during our typology development, and they suggested several guiding principles that ASPE endorsed and that prioritized our work. First, they recommended that we not adopt or refine any typologies they had previously developed, because their typologies were not specific enough for key outcome variables of interest and because few of the studies used data from 1998 or beyond, a period when TANF was in full operation. Rather, they recommended that we develop new and more expanded typologies of our own. Second, they emphasized that besides the new typologies linked to outcomes, the most valuable contribution we could make to the research community would be a public use database containing typologies of TANF policies (quantified to the maximum extent practicable), outcomes, and contextual demographic and economic data for each, for use by the wider research community. That is what you will find described in this report, and what you will find on our Web-based database. For more detail on the Group's recommendations and the composition of the group, please refer to Appendix II.
Thus, we have developed six new typologies and built a database based on them, the TANF Typologies Database, using the Urban Institute's Welfare Rules Database (WRD) (2000) and its accompanying Welfare Rules Databook (Rowe 2000), as the primary sources for TANF policy. There are, though, a few policies whose data source is other than the Welfare Rules Database and Databook. The specific data source for each TANF policy, outcome, and economic and demographic data element is noted in the on-line Data Dictionary and documentation.1 All policies are measured for 1999 only, though the database has the capacity for adding prior and subsequent years, in the future. These policies measure the rules in effect in each state in 1999 (typically July 1999). The policy variables measure the "majority rule" in the state—the policy that affects the majority of the caseload in the state.2
Preliminary Typologies
Our typologies are organized around six outcomes and two sub-outcomes that are of interest to Congress, DHHS, and the research community. These are (1) job entry rate, (2) job retention rate, (3) earnings gain rate, (4) caseload change, (5) child poverty rate, and (6) out-of-wedlock birth rate. While other outcomes are also of interest, we focus on these six because they are listed as outcomes of interest in PRWORA legislation, regulations, and bonuses. The first three reflect the work goals articulated in the PRWORA legislation; caseload declines, to the extent they occur, reflect attainment of the goal of reducing dependency; states are required to report what corrective actions they will take if child poverty increases five percent; and states can compete for bonus money to reward reductions in out-of-wedlock births.3
We organize the typologies around outcomes, because we think that the best way to think about state policies is to focus on the outcomes they might affect. Policies that have one effect on one outcome may have an entirely different effect on another outcome. For example, generous earned income disregards for benefit computation may reduce poverty but increase caseloads. Outcomes also present a convenient way for organizing welfare policies because not all policies are hypothesized to affect all outcomes. Organizing policy typologies around outcomes thus reduces the complexity of welfare policies by limiting the number of policies in a typology. However, organizing typologies around outcomes does mean that some policies will appear in multiple typologies. For example, time limit policy variables are included in both the recipient job entry and caseload typologies because we hypothesize that they might affect both outcomes.
We further subdivide two of the six outcomes into two sub-outcomes each, in recognition that different policies are known to different sub-populations affected by TANF at different times, or that different policies affect sub-populations in different ways. Thus, job entry rate is subdivided into that for applicants and that for recipients, on the theory that applicants and recipients face different incentives and information that might make them enter employment. Similarly, caseload change is subdivided into (because it is affected by) those entering the rolls and those exiting the rolls, and different information and incentives will affect those two behaviors.
Direct vs. Indirect and Short-run vs. Long-run Effects
To limit the number of policies included in each typology, we consider only those policies hypothesized to most directly affect each outcome. For example, recipient work requirements should directly affect whether or not a welfare recipient enters a job, and so are included in the recipient job entry rate typology. One might hypothesize that family caps affect job entry through their effect on fertility and fertility's effect on employment, but we consider this an indirect effect and so do not include family caps in the job entry rate typology.
To narrow our hypotheses of the possible effects that a policy may have on an outcome, we largely focus on a policy's short-run direct effect on an outcome, rather than its long-run and less direct effect via the composition of the caseload. For example, states that allow recipients to receive TANF benefits for a longer period of time (longer time limits) may have lower job entry rates in the short-run, but higher job entry rates in the long-run. This could happen if states with shorter time limits have the most employable recipients leave their caseload quickly, resulting in a caseload with a higher concentration of "disadvantaged" recipients (relative to states with longer time limits) in the long-run. While many researchers and practitioners expect that the most disadvantaged recipients will be left on the caseloads in the longer run, to date there is little evidence in national Current Population Survey data that welfare reform has affected the composition of the caseload in its labor market skill distribution (Moffitt and Stevens 2001). Similarly, our hypotheses regarding the effects of time limits, earned income disregards and other variables focus on the direct effect of these policies, rather than their potential long-run, indirect effects via the caseload. Differences in many of the policies' effects over time and differences in indirect versus direct effects remain among the many reasons why the hypotheses described on the following pages may not hold true.
Summary Variables
These typologies consist of a combination of one or more "summary variables" and, in some cases, additional independent policy variables. We hypothesize that a given outcome is affected by certain policy domains which we call a summary variable. A summary variable is a conceptual method we used to combine groups of similar policies that we hypothesize to have a common, underlying policy concept. As presented later in this report, factor analysis is used to reduce the size of a group—a summary variable—into 2 or 3 key variables that explained the most significant relationship with an outcome. To illustrate, the policy domain or summary variable, Recipient Work Requirements, is comprised of seven related policy variables that include:
- what is an allowable work activity to meet the requirement;
- the minimum hours of work required in order to count;
- how quickly the requirement kicks in, in relation to the onset of cash benefits;
- who can be exempt from a work requirement (due to two selected criteria);
- the dollar amount of the most severe sanction;
- the duration of the sanction; and
- the dollar amount of the initial sanction.
Notes for this section
1. The Typologies Database and associated on-line Data Dictionary and documentation are found at http://afdc.urban.org/outcomes.
2. Massachusetts is an exception. All relevant Massachusetts policy variables are for non-exempt TANF clients. According to the Massachusetts Department of Transitional Assistance, non-exempt clients account for a minority of the caseload (http://www.state.ma.us/dta/dtatoday/facts/tafdc.htm, accessed November 2001).
3. At the time we developed our typologies and the database, the President had not yet proposed elimination of this bonus and its associated performance data reporting.
Note: The full report is available in its entirety in the Portable Document Format (PDF). The accompanying appendices for this report are also available in the Portable Document Format (PDF) in their entirety as a separate document.
Latest Research
Browse by Topic:
PDF
Printer-friendly summary
