The Changing Role of Welfare in the Lives of Low-Income Families with Children

Transcription

1 The Changing Role of Welfare in the Lives of Low-Income Families with Children Pamela Loprest Sheila Zedlewski The Urban Institute Occasional Paper Number 73 Assessing the New Federalism An Urban Institute Program to Assess Changing Social Policies

2 The Changing Role of Welfare in the Lives of Low-Income Families with Children Pamela Loprest Sheila Zedlewski The Urban Institute Occasional Paper Number 73 Assessing the New Federalism An Urban Institute Program to Assess Changing Social Policies The Urban Institute 2100 M Street, NW Washington, DC Phone: Fax:

3 Copyright August The Urban Institute. All rights reserved. Except for short quotes, no part of this paper may be reproduced in any form or used in any form by any means, electronic or mechanical, including photocopying, recording, or by information storage or retrieval system, without written permission from the Urban Institute. This report is part of the Urban Institute s Assessing the New Federalism project, a multiyear effort to monitor and assess the devolution of social programs from the federal to the state and local levels. Olivia Golden is the project director. The project analyzes changes in income support, social services, and health programs. In collaboration with Child Trends, the project studies child and family well-being. The Assessing the New Federalism project is currently supported by The Annie E. Casey Foundation, The Robert Wood Johnson Foundation, The John D. and Catherine T. MacArthur Foundation, The California Endowment, and the Charles Stewart Mott Foundation. The authors thank Jennifer Holland and Simone Schaner for their excellent research assistance and contributions to the study. Olivia Golden and Gregory Acs provided insightful comments on an earlier draft. 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.

4 Contents Executive Summary v Changing Policies and Knowledge about Their Effects 2 Syntheses of the Effects of Reforms in the 1990s 4 Post-Reform Welfare Leavers and Stayers 5 Some Recent Key Studies 6 Data and Methods 8 Changing Welfare Caseload 10 Family Structure and Demographics 14 Family Structure 14 Demographics 16 Work 19 Employment Outcomes 20 Characteristics of Jobs 24 Barriers to Work 26 Family Income and Economic Well-Being 34 Composition of Income 34 Poverty 36 Use of Other Government Supports 38

5 Family and Child Well-Being 42 Disconnected Families 44 Summary and Implications 47 Appendix A: Sample Selection and Definitions 51 Appendix B: Results for All Three NSAF Years (1997, 1999, and 2002) 57 Notes 73 References 76 About the Authors 79 iv

6 Executive Summary Reforms passed in 1996 ended welfare as we knew it. Individuals no longer have an entitlement to welfare, and states have changed how they administer cash assistance. Most states actively discourage welfare entry and aggressively require new entrants to pursue paid work. Numerous other safety net programs also changed, as the 1996 legislation limited immigrant eligibility for food stamps, scaled back children s eligibility for disability benefits, increased federal money for child care, and placed greater demands on states child support enforcement systems. States were given incentives to encourage marriage and the formation of two-parent families and to reduce out-of-wedlock child bearing. The dramatic decline in welfare caseloads represents the most stunning postreform outcome. Caseloads declined by over 50 percent in just a few years. Numerous studies attempt to explain caseload decline and how families fared. Scholars generally agree that the strong economy, work supports such as the Earned Income Tax Credit, and welfare reform all encouraged a shift from welfare to work, especially among single mothers. Scholars disagree about the relative importance of each of these factors. The National Survey of America s Families (NSAF), conducted by the Urban Institute as part of its Assessing the New Federalism project, documented changes in low-income families circumstances at the national level over the 1996 to 2002 period. This study uses these data to summarize what we learned about families in the welfare system. We describe outcomes for three low-income groups: families currently on welfare, families that recently left welfare, and those that never received welfare. The outcomes discussed include family structure, demographic characteristics, work and barriers to work, income, and well-being. To establish comparability across the three sample groups we limit the sample to low-income families, defined as those with income below 200 percent of the federal poverty level either last year or in the current year and not above 250 percent of the poverty level in either period. 1 These income restrictions confine all three sample groups current welfare recipients, recent welfare leavers, and nonwelfare families to families with consistently low incomes. Changes in welfare policy after reform had the potential to affect who stayed on welfare, who exited welfare, who came on to welfare, and who did not come on. This means that all three groups of low-income families we study could be affected by changes in welfare policy. In addition, all three groups can be affected by the economy and broader societal trends. Over time, changes for one group can influence the composition of other groups of low-income families. Our results, therefore, generally describe outcomes among low-income families most likely to be affected by the evolution of the new welfare policies. Our key findings are described below. THE URBAN INSTITUTE v

7 Family Structure and Demographics States implemented numerous changes in welfare policies that could affect the structure and demographics of families on welfare. Most states expanded eligibility for two-parent families, some no longer increased benefits when a baby was born while on welfare, and all required teen parents to live in supervised settings. Some states adopted federal restrictions on eligibility for immigrant families, while others used their own funds to maintain immigrant eligibility. Generally, we observe increases in cohabitation among low-income families but decreases in marriage, changes in the immigrant composition of welfare caseloads, and changes in the geographic distribution of welfare caseloads. Changes in family structure occurred across all three sample groups. Marriage declined among both recent welfare leavers and nonwelfare families. Cohabitation nearly doubled among families currently on welfare and those never on welfare between 1997 and Given similar changes in marital status among the nonwelfare group and the groups with welfare experience, these changes are likely due to factors beyond welfare policy. The immigrant share of the welfare group increased from 6.6 to 11.3 percent, and their share of the nonwelfare group jumped from 15 to 24.4 percent, between 1997 and The increasing share of immigrants on welfare occurs only among immigrants in the United States for five or more years (who remained eligible for benefits under the new federal rules) and not among newer immigrants. We do not observe a comparable increase in immigrants among recent welfare leavers. Immigrant parents make up a larger share of longer-term welfare stayers in 2002 than in 1997, suggesting that immigrants find it more difficult to leave welfare than nonimmigrants. Comparing 2002 to 1997, a larger share of current welfare recipients live in the West, and a larger share of recent welfare leavers live in the South. The geographic distribution of the nonwelfare group remained fairly constant during this time. These shifts probably reflect geographic differences in welfare policies. Western states with more generous policies (especially California, which dominates the western caseloads) have retained or attracted more low-income families to welfare while southern states with low benefits and stricter work participation policies discourage welfare use more than other regions. Given the simultaneous declines in welfare caseloads across all regions of the country, the result does not mean that families have moved for more generous welfare benefits. Work and Barriers Moving welfare recipients into work was a major focus of welfare reform. Through a mix of incentives and requirements, states implemented various work policies to increase work among recipients and move recipients off welfare into work. Over time, as states endeavored to engage greater shares of the caseload in work, they vi

8 were confronted with recipients multiple barriers to work and began to implement strategies to address these needs while continuing to focus on work. We find that work among welfare recipients increased dramatically, while work among former recipients and the nonwelfare group declined. Work and work-related activity among current welfare recipients increased dramatically from 1997 to In 1997, 20.9 percent of recipients were working, compared with 29.2 percent in In contrast, work declined among recent leavers and nonwelfare families over this period. Work increased among current welfare recipients with employment barriers, including those with less than a high school education, those with a child receiving SSI, and those who are primarily Spanish speaking. Employment of current recipients with multiple barriers more than doubled between 1997 and 2002, from 10.3 to 25.9 percent. Over the same time, the percentage of nonwelfare recipients working with barriers generally declined. This suggests that states stronger work policies had a broad impact on caseloads. We find little evidence to support the concern that an increasing share of current welfare recipients face barriers to work. Between 1997 and 2002, the share of the caseload with multiple barriers remained constant and only one potential barrier to employment, the percent primarily speaking Spanish, increased. In contrast, barriers to employment increased among recent welfare leavers, who reported poor physical and mental health problems and multiple barriers more often in 2002 than in This could be due to an increase in the share with barriers able to work their way off welfare or an increase in recipients with barriers being cut off benefits because they were unable to meet requirements. Between 1997 and 2002, wage rates increased significantly for all three groups. Median wage rates, after accounting for inflation, increased from $5.50 to $7.00 an hour for current recipients, from $6.25 to $7.75 for former recipients, and from $7.00 to $8.15 for low-income women with no welfare history. In percentage terms, the wage increases for the two welfare groups are higher than the increase for the nonwelfare group. Other benefits on the job (such as paid leave and employer health insurance) remained unchanged. Income and Poverty Welfare policies that encourage earnings and discourage receipt of welfare benefits and changes in family composition also can affect cash family income. In addition, changes in work supports including food stamps, child care, and health insurance will affect families noncash income. On average, we see improvements in income for current welfare recipients but little change in income and poverty for recent welfare leavers and the nonwelfare group. However, poverty among single parents living alone, the group most likely affected by welfare policy, changed in different ways across the three study groups. Also, the share of families in the recent and nonwelfare groups receiving work supports increased over time. THE URBAN INSTITUTE vii

9 Among single parents living alone, poverty declined for current welfare recipients. Deep poverty (income below 50 percent of the federal poverty level) dropped by over 16 percentage points between 1997 and 2002 (from 69 to 52.6 percent). New state policies allowing welfare recipients to receive some cash assistance while working for pay, along with the shift in the welfare population away from the South with relatively low benefits and toward the West with higher benefits, contribute to reduced poverty among families on welfare. For single parents living alone, poverty levels remained fairly flat for recent welfare leavers and increased for the nonwelfare families. Deep poverty increased from 19.5 to 25.8 percent between 1997 and 2002 for the nonwelfare group. While this increase in deep poverty could reflect the weaker economy, it did not occur among comparable singles in the welfare groups. Instead, the result could reflect a compositional shift across groups as more very low income single parents remained off the welfare rolls in 2002 than in Countering the trends in poverty, receipt of noncash government work supports increased for recent welfare leavers and nonwelfare families but declined for current welfare families. More families that left welfare received help paying for child care, food stamps, and government health insurance in 2002 than in 1997, reflecting improvements in the delivery of these benefits, more dollars for child care, and the new State Children s Health Insurance Program enacted in Declining noncash supports for welfare families probably reflect new rules restricting immigrant eligibility for some of these benefits. Family and Child Well-Being During the debate over the 1996 reforms, some argued that children would be helped by the new policies while others argued that children would be harmed. Advocates argued that the new policies would increase income and that working mothers would provide better role models for their children. Opponents argued that children would be harmed if mothers lost eligibility for welfare benefits and could not find jobs or if mothers worked but lacked adequate child care. Like other researchers, we find relatively few changes in family or child well-being over time. Families material well-being remained fairly constant over the 1997 to 2002 period, though a few indicators improved for recent welfare leavers and worsened for nonwelfare families. More recent leavers owned homes and fewer experienced food insecurities in 2002 compared with In contrast, home and car ownership decreased and crowding and food insecurity increased among nonwelfare families. The results suggest a better picture of well-being for recent leavers than shown through the income and poverty measures but mirror the deteriorating income picture for some families without welfare experience. Measures of well-being among young children (age 6 to 11) remained fairly constant for the current welfare and recent welfare groups, while school engagement and participation in activities outside of school declined for young children in the nonwelfare group. While few changes in well-being occurred among older chil- viii

10 dren (age 12 to 17), the incidence of school expulsion decreased among children in families on welfare. School engagement and expulsions increased among children in families with no welfare experience. Reports of parent aggravation indicate increased stress among recent welfare leaver families. Three times as many parents with young children reported being aggravated in 2002 than in 1997, but no changes in aggravation occurred among families in the current or nonwelfare groups. Aggravation among parents of teens remained constant for all three groups. The result suggests the struggles that former welfare recipients, usually single parents, may face when working and arranging child care for young children. Disconnected As many states turned their attention to groups of recipients having a harder time making the transition to work recipients with multiple barriers to work, those nearing the time limit, and sanctioned families it became clear that there was a group of families that leave welfare but do not find work and remain disconnected from both the labor market and the welfare system. This group those without cash welfare or disability benefits, current or recent work, or a working spouse represents families with some of the greatest barriers to work and at highest risk for severe economic hardship. While we often think of the disconnected relative to former welfare receipt, many families that have never been on welfare can also be defined as disconnected. In 2002, one in five former welfare recipients was disconnected. These former recipients are significantly more disadvantaged than other leavers, with lower income, a greater number of barriers to work, and a higher likelihood of food insecurity. About 12 percent of families that have never received welfare were disconnected in They face levels of disadvantage similar to disconnected welfare leavers, including similar barriers to work, low incomes, and material hardships. However, fewer disconnected families without welfare experience receive food stamps, housing assistance, and Medicaid than disconnected former welfare families. Lacking any connection (current or past) to a welfare system that might provide assistance, these families may face the greatest risk. Implications The changing circumstances of low-income families most likely affected by new welfare reforms suggest some good news, some concerns. More single parents have jobs, including some on welfare and some with apparent employment barriers such as poor health and limited education. Deep poverty declined among families on THE URBAN INSTITUTE ix

11 welfare (although this result at least partially reflects the steeper declines in welfare participation among those in traditionally low-benefit states). Families that had recently left welfare or that had remained off welfare had other government work supports such as help paying for child care, health insurance, and food stamps more often in 2002 than in Real wage rates improved among workers in all three low-income groups. However, the results also indicate the precarious lives of families in our lowincome sample. Significant declines in employment among recent welfare leavers and nonwelfare families between 1997 and 2002 highlight the sensitivity of the low-wage labor market to economic forces. Substantial shares of families on welfare and those that recently left welfare live in poverty, and many live in deep poverty. Families also report high levels of stress such as difficulty paying for food and other bills. While some improvements occurred in these qualitative measures of material well-being for recent welfare leavers, the percentage of these parents reporting high levels of aggravation tripled between 1997 and 2002, indicating the struggles of mostly single parents attempting to make it without welfare. The results also suggest some perhaps surprising concerns about low-income families outside the welfare system. Cash income deteriorated and deep poverty increased among single parents in the group of families remaining independent of welfare between 1997 and Also, compared with recent welfare leavers, relatively few of these families receive work supports such as food stamps and help paying for child care to supplement their wages. About 1 in 10 of these families remains disconnected from work or cash government assistance. Policymakers need to reach out and deliver work supports to low-income families outside the welfare system and ensure that new welfare policies do not discourage the neediest families from receiving temporary cash assistance. These concerns argue for continued evaluation of this new safety net system. States now face new challenges to meet tougher work participation requirements in the reauthorized welfare program. They must find a way to balance efforts to increase work among current participants with continued efforts to address the needs of the hardest to serve families in their caseloads and to provide work supports to welfare leavers and low-income families not on welfare. Reauthorization likely will define a new chapter in the evolving story of welfare reform. Note 1. As explained in the paper, we further restrict our sample to families with adults who are parents and not receiving Supplemental Security Income (a means-tested disability program). These restrictions exclude families unlikely to be affected by welfare policies. x

12 The Changing Role of Welfare in the Lives of Low-Income Families with Children In February 2006, Congress reauthorized the landmark welfare programs enacted almost a decade earlier. While the new legislation substantially increased work requirements for states welfare caseloads and imposed new restrictions on what counts as work, it otherwise reaffirmed the 1996 law. Congress retained the fixed block grant funding, the five-year lifetime benefit time limit, and the flexibility states have to implement their welfare programs within broad federal guidelines. Reauthorization presents a good time to take stock of what we know about how welfare reform transformed support for low-income families. The original legislation led to considerable speculation about how states would respond to the new rules and how low-income families would be affected. Some speculated that states would reduce their financial commitment to welfare. Others questioned whether mothers on welfare would be able to get jobs, whether the remaining caseload would grow increasingly disadvantaged, and whether time limits would lead to increased child poverty. Various studies addressed many of these core questions over the past decade. Our own studies reported outcomes for families that left welfare and families remaining on welfare using the National Survey of America s Families (NSAF) conducted in 1997, 1999, and This survey was designed to monitor devolution of social policy from the federal to the state governments and included a set of questions specifically focused on families with welfare experience. Loprest (1999) provided the first national picture of work and other outcomes for families that left welfare between 1995 and She provided a comprehensive follow-up status report for families leaving welfare between 1997 and 1999 and some snapshots of outcomes for families that left welfare between 2000 and Similarly, Zedlewski (1999) provided the first national picture of families on welfare in 1997, and subsequent papers updated the information for families in 1999 and This study uses the three rounds of the NSAF to comprehensively review the status of families moving in and out of welfare during the 1997 to 2002 period. We use the data to describe the changing circumstances of families on welfare and those that left welfare during this time frame. We also describe the situations of families with similar incomes but no welfare experience. Changes in welfare policy after reform had the potential to affect who stayed on welfare, who exited welfare, who came on to welfare, and who did not come on. This means that all three groups of low-income families we study could be affected by changes in welfare policy. In addition, all three groups can be affected by the THE URBAN INSTITUTE 1

13 economy and broader societal trends. Over time, changes for one group can influence the composition of other groups of low-income families. Our results, therefore, generally describe outcomes among low-income families most likely to be affected by the evolution of the new welfare policies. Most states implemented their new welfare programs sometime during While some states retained reforms implemented earlier through waivers from the federal Aid to Families with Dependent Children (AFDC) program, we expect that outcomes in 2002 more fully reflect the effects of welfare reform than outcomes in We begin by reviewing the 1996 landmark legislation and the key literature to date on the effects of welfare reform. Then we describe the changing dynamics of welfare and changes in family structure, demographic characteristics, work and barriers to work, income, and family well-being, placing the outcomes in the context of changing state and federal policies that occurred during the period. Our results cannot describe the effects of particular state welfare policies or even link outcomes to welfare reform in general. Instead, the results show how low-income families on and off welfare were doing during this period. The results highlight some of the positives of welfare reform and some of the concerns as states begin their second decade working with this new safety net. Changing Policies and Knowledge about Their Effects The 1996 welfare reform legislation culminated a process begun in the early 1990s. The federal government had awarded many waivers to states to experiment with new rules in their AFDC programs. 1 By the time welfare reform was passed in 1996, 27 states had major waivers in place (Blank 2002). Many states with waivers tried out more stringent work requirements, and some capped benefits for the family size at the time of enrollment so another child did not result in a higher benefit. Some states had waivers to use temporary time limits in order to discourage long-term welfare use, but none had permanent, lifetime time limits. Evaluations of states waiver experiments generally confirmed that the large majority of waivers designed to increase work (mandatory employment services, earnings supplements, and temporary time limits on welfare receipt) boosted employment and reduced welfare receipt. However, most programs did not raise income (Bloom and Michalopoulos 2001). Also, increased parents work participation had either positive or negligible effects on the well-being of children under age 12 (Morris et al. 2001). The one concern raised by the state waiver programs is that parents in mandatory employment programs reported somewhat worse outcomes for their adolescent children (Gennetian et al. 2002). The success of states welfare experiments and the nation s growing dissatisfaction with the current welfare program led Congress to enact sweeping changes to welfare and related programs in The Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) of 1996 replaced the 60-year-old AFDC program with Temporary Assistance to Needy Families (TANF). The key features of 2

14 TANF were the elimination of the entitlement to welfare and devolution of responsibility for welfare to the states through a system of fixed block grants. While states were free to develop their own policies toward work requirements, they were required to meet federally established work participation requirements for their caseloads. States responded quickly to the new legislation, creating programs that vary widely across the country (Rowe and Giannarelli 2006). All states have strengthened work requirements, but states vary in their use of incentives (such as an increase in the amount of earnings that can be retained while still receiving some cash assistance) and penalties (such as sanctions that eliminate benefits for families that do not comply with job search requirements) to accomplish this goal. While all states must comply with the prohibition against using federal dollars to pay benefits beyond five years, some have shorter time limits and others guarantee some benefits beyond five years with state monies. States also vary considerably in the rules governing how families access cash assistance. Some emphasize diversion strategies that require substantial proof of job search before assistance can begin or offer some short-term assistance in lieu of a monthly welfare check. Other states quickly enroll eligible applicants in TANF and job search programs. PRWORA also included major changes in other income support programs for low-income families with children. In addition to TANF, the legislation made major changes to Supplemental Security Income (SSI), food stamps, child support enforcement, and child care. 2 PRWORA eliminated SSI benefits for some children, scaled back food stamp benefits, and eliminated food stamp and SSI eligibility for most legal immigrants. The legislation included provisions designed to reduce illegitimate births and to encourage marriage and the formation of two-parent families. It also strengthened states child support collection systems and increased money for child care. Congress enacted other changes to the safety net over time. The ban for all legal immigrants was gradually softened. Congress quickly restored SSI eligibility for legal immigrants who had been in the United States before PRWORA was enacted and reinstated food stamp eligibility for legal immigrant children and elderly and disabled immigrants who were in the country when PRWORA was enacted. In 2002, Congress restored food stamp eligibility for all legal immigrants after five years in the United States. Congress also enacted the State Children s Health Insurance Program (SCHIP) in 1997 to provide health insurance for low-income children. All these changes and states responses to them potentially affect low-income families. The variability in states safety nets expands beyond TANF. Some states use their own funds to pay for food stamps for legal immigrants not eligible under federal rules. States have enacted different eligibility standards for SCHIP eligibility. And states use their block grant funds to finance assistance for low-income families in various ways (Zedlewski et al. 2002). During this period, the economy also profoundly affected low-income families. Sustained growth in GDP and declining unemployment throughout the mid- to late 1990s provided the backdrop for large increases in employment. Between 1994 and 1999, the labor force participation rate of unmarried women with children increased 10 percentage points (Blank 2002). These changes dramatically affected families on welfare and other low-income families requiring basic cash assistance. The most obvious and publicized outcome THE URBAN INSTITUTE 3

15 of these reforms was a dramatic decline in the welfare rolls. Caseloads were cut in half between 1996 and 2000 and continued to decline thereafter, although at a much slower pace (Zedlewski and Williamson 2006). Caseloads dropped to about 2 million families by 2004 (HHS 2005b). While there is some disagreement about the relative importance of different factors in explaining the dramatic caseload decline, several studies concluded that the strong economy, welfare policy changes, and other federal policy changes designed to make work pay (such as an expanded Earned Income Tax Credit [EITC] and a higher minimum wage) all contributed to caseload decline (Blank 2002; Besherov 2002). Welfare reform generated numerous studies to investigate the effects on families. This national experiment was not accompanied by a major national evaluation of the effects on families. Instead, researchers conducted analyses using different surveys, including some designed to address specific aspects of welfare reform or to examine the effects of reforms in specific cities or counties and some nationally representative household surveys regularly conducted by the Bureau of the Census. Some important major lessons about the effects of welfare reform on families can be drawn from the post-reform studies. Below we briefly review the important findings from this body of work. We begin with an overview of two important syntheses of the effects of reform that were published in Then we summarize the findings of the post-reform welfare leaver and stayer studies conducted for the U.S. Department of Health and Human Services because they provide excellent context for the results we later present from the NSAF. Next we turn to finding from a few key studies published after the 2002 synthesis papers. The more recent studies tend to focus on how particular aspects of welfare reform policies affected welfare dynamics. We restrict our literature review to syntheses and studies that primarily focus on outcomes for families. Additional studies address welfare reform and state program implementation issues (for example, Blank and Haskins 2001 and Weil and Finegold 2002). Syntheses of the Effects of Reforms in the 1990s Grogger, Karoly, and Klerman (2002) provide a comprehensive synthesis of the research on the effects of TANF legislation on several outcomes, including the welfare caseload, employment and earnings, use of other government programs, fertility and marriage, household income and poverty, food security and housing, and child well-being. The synthesis represents a knowledge base built on studies completed through The synthesis points out, as noted above, that many studies conclude that welfare reform had a substantial effect on reducing the caseload, and numerous studies show that welfare reform was at least partially responsible for the increase in work and earnings among single mothers during the 1990s. These authors conclude that welfare reform, at least as it had evolved through 2000, appeared to produce impacts similar to those seen for mandatory work-related activities with weak financial work incentives (similar to the waiver evaluations summarized earlier): a decline in welfare use and use of food stamps and an increase in employment and earnings. For the most part, former recipients overall incomes did not increase, as earnings replaced welfare and food stamp benefits. However, reforms 4

16 with generous financial work incentives, such as large earned income disregards, did show positive impacts on income. The authors also conclude that studies show both positive and negative effects of welfare reform on child well-being. Also, the evidence does not support any firm conclusions on the effects of welfare reform on marriage, the effects of family caps on fertility, or the effects of particular welfare policies such as sanctions or time limits. Blank (2002) critically reviews the economic evidence about the effects of welfare reforms enacted in the 1990s. She points out some key challenges in evaluating the effects of reform, including the difficulties in characterizing the policy environment given the wide state variability in policies and implementation strategies and the limitations of current survey data that fail to record many important aspects of reform. Reviewing the voluminous literature examining the effects of reforms on caseload change published through 2002, she concludes that state waivers and TANF had significant negative effects on the size of welfare caseloads. She highlights the difficulty in separating out the effects of welfare policy from the economic effects and other policy changes, such as the expansion of the EITC, during this time frame. She also concludes that the literature shows some moderate increases in cash income and declines in poverty among less-skilled, single-mother families during the late 1990s. Blank argues that we need to experience economic cycles before economic effects can be separated from welfare policy effects. We also need to investigate whether certain states welfare strategies make their low-income citizens more or less vulnerable than others, and we need to learn more about the effects of reforms on families disposable income and long-term opportunities for income growth. Post-Reform Welfare Leavers and Stayers The U.S. Department of Health and Human Services (HHS) provided grants to 15 states and localities to survey families that left welfare during the period. Survey findings confirmed that the majority of those who left welfare had a job, although many did not have steady employment (Acs and Loprest 2001). The median result showed that 57 percent of leavers were employed in the first quarter after leaving welfare, but only 30 to 40 percent of leavers worked in all four quarters of the year after leaving welfare. The surveys also showed that average earnings for welfare leavers were relatively low and the eight sites that measured total income found that most family incomes remained below the poverty level after leaving welfare. The leaver surveys also documented that roughly half the families received food stamps and Medicaid for the adults in the first quarter after leaving welfare. HHS also sponsored surveys of welfare recipients in five states and the District of Columbia in 2001 to understand the circumstances and prevalence of barriers to work facing these families. Hauan and Douglas (2004) use the combined data from all six surveys to describe the general prevalence of employment liabilities among single TANF parents and to estimate the effects of these liabilities on work. The authors include human capital deficits (lack of a high school diploma, limited work experience), health and well-being deficits (poor physical health, poor mental health, having a child with serious health problems, pregnancy, domestic violence or crimi- THE URBAN INSTITUTE 5

17 nal record issues), and situational difficulties (lack of transportation, lack of child care, housing instability, neighborhood problems) in their analysis. The authors find a high prevalence (defined as occurring in 20 percent of the caseload or more) of human capital deficits, situational difficulties, and certain health challenges (poor physical and mental health, having a child with special needs) among the caseload. However, they also found that only human capital deficits, poor physical health, pregnancy, and child care problems significantly decreased employment. The authors conclude that state welfare programs need to focus on building up employment assets such as work experience and education to increase work among welfare recipients. Some Recent Key Studies Many additional studies have been completed since the major syntheses described earlier. The key studies reviewed below focus on more specific aspects of welfare policy. Several studies use longitudinal data sets to understand changes in welfare dynamics. Others highlight outcomes for disadvantaged groups and the effects of welfare on family structure. Welfare Dynamics Welfare reform has affected families decisions about whether to participate in welfare and when to leave. A report based on data from the Three Cities Study examines how new welfare rules affect welfare entry (Moffitt and Winder 2003). 3 This study confirms that in general eligible nonentrants (composed of low-income families that have never thought about applying for welfare, families that thought about applying but did not, and families that applied but never entered the rolls for various reasons) tend to be better off financially than welfare entrants. In addition, those in poor health or with a functional disability are less likely to enter TANF than those in good health or without a disability, even after controlling for entry into the SSI program for low-income people with disabilities. The study also documents that most (77 percent) welfare applicants experience some type of diversion, including gatekeeping policies and informal practices of social service agencies, kin networks that discourage participation, and self-imposed diversion in response to time limits. Diversion practices discourage less-educated women from entering welfare more than women with more education. A study using data from the Fragile Families Study, which examined the parents of newborns in 15 cities, investigates the connection between state policies and welfare use post PRWORA (Teitler, Reichman, and Nepomnyaschy 2004). Specifically, the authors ask whether mothers living in states with generous welfare payments and less restrictive time limits and work requirements are more likely to rely on welfare than mothers with similar characteristics living in states with less generous benefits and stricter requirements. The authors find that a significant proportion of city-to-city variation in TANF participation can be accounted for by state policies. Social and demographic factors, however, are stronger predictors of individuals 6

18 reliance on TANF than state policies. That is, most of the variability in reliance on TANF is within rather than between cities. Acs, Ross Phillips, and Nelson (2005) use data from the Survey of Income and Program Participation to compare welfare entry behavior in the pre and post welfare reform periods. Using regression models and decomposition techniques, they find that new welfare policies such as time limits, full-family sanctions, and family caps led to a drop in welfare entry rates during the 1990s. Neither changes in the characteristics of low-income single mothers nor improvements in the economy directly accounted for the decline in entries. The authors find that declining entry rates were not accompanied by substantial improvements in the circumstances of low-income single mothers not on welfare. Grogger (2004) examines the role of time limits in the decline in welfare participation. He finds that being subject to time limits, even before exhausting eligibility, leads recipients to reduce their use of welfare by 6 to 7 percent. He finds that time limits account for 12 to 13 percent of the decline in welfare use in the late 1990s. Richburg-Hayes and Freedman (2004) examine the question of welfare cycling through a reanalysis of MDRC evaluation data collected during the mid- to late 1990s. The authors conclude that cyclers (defined as those receiving welfare in three or more discrete spells during a four-year observation period) make up only a small share (8.5 percent) of the caseload. Cyclers are less disadvantaged than longer-term recipients but more disadvantaged than short-term recipients in terms of their employment history and outcomes. The study also documents that the incidence of cycling increased during the years following PRWORA, although the strong economy during this period and new welfare policies both probably contributed to the increased incidence of cycling. Barriers to Employment and Disconnected Families Some recent studies have focused on barriers to employment among welfare recipients and welfare leavers living without welfare, work, or other cash government assistance. The Women s Employment Study (WES) has been tracking welfare outcomes for a group of women in an urban county in Michigan. Danziger, Kalil, and Anderson (2000) report that about 20 percent of women on welfare in their sample have multiple barriers to employment, and those with co-occurring human capital, mental health, and physical health problems have the poorest work outcomes. Heflin (2006) documents much higher levels of hardship across five waves of the WES than in any single year. Reports of telephone disconnection and having unmet medical needs are common. Researchers have also used the WES data to document the share of welfare leavers without visible means of support. Turner, Danziger, and Seefeldt (2005) report that between February 1997 and August 2003, 9 percent of those who left welfare were without both work and welfare for more than one-quarter of the time. Compared with other welfare leavers and those who remained on the rolls, these disconnected women had lower economic status and a greater number of barriers to employment. THE URBAN INSTITUTE 7

19 As part of their evaluation of welfare reform in New Jersey, Wood and Rangarajan (2003) find that about one in four former recipients in that state are not working and off TANF in a given month; these families face substantially more material hardship than working leavers. A smaller subset of this group, roughly 1 in 10 leavers in a given month, does not have a working spouse, recent work experience, or other source of government income such as disability benefits or unemployment insurance. However, people move in and out of this status over time. Among this group, one in four is back on TANF 12 months later and a similar percentage is working. Welfare and Family Structure Various studies document a rise in cohabitation among low-income families subsequent to welfare reform. Acs and Nelson (2004) use data from the 1997 and 1999 NSAF to examine the effects of state welfare policies on living arrangements of lowincome families with children. Results suggest that more effective child support collection and family cap policies correlate with declines in single parenting and increases in dual parenting. The authors caution, however, that their findings apply to a twoyear period early in the TANF era and, at best, represent short-term effects. Angel and colleagues (2002) as part of The Three Cities Study also report an increase in children living in two-adult families between 1999 and The authors find that cohabiting arrangements accounted for nearly all of the increase in two-adult families. There was no change in the share of children living with their biological mother and father. Bitler, Gelbach, and Hoynes (2006) use data from the Current Population Survey to investigate the effect of changing welfare policies on the living arrangements for children from 1989 through They find large effects of welfare reform on living arrangements. Black and Hispanic children more often live with neither parent or both parents and less often live with an unmarried parent since reforms were implemented. The results are more mixed for white children and include an unexpected, negative effect of welfare reform on the probability of living with a married parent. The authors caution, however, that they cannot understand which features of welfare reform led to the measured impacts and that reforms may simply proxy for unmeasured differences across states rather than true policy responses. Data and Methods The data used in this paper come from the National Survey of America s Families, a nationally representative survey of the civilian population under age 65 and its families. Three rounds of interviews were conducted in 1997, 1999, and Each round included a cross-section of about 40,000 families, and each oversampled the low-income population (family income below 200 percent of the federal poverty level) and the population in 13 states. While most interviews were conducted over the phone, in-person interviews were conducted to account for the population without telephones. 4 8

20 For the most part, interviews were conducted with the adult (usually the parent) most knowledgeable about the children in the family (the most knowledgeable adult, or MKA), so our analyses focus on this group. These adults, limited to persons age 18 to 64, were asked key questions about their families history of welfare participation. Respondents also were asked a full battery of questions about their family s structure and demographics, income (with details on sources of income), and employment (both current and last year for the MKA and her/his spouse or partner). MKAs also were asked about their mental and physical health and answered questions designed to assess the well-being of their children. The NSAF also asked MKAs about their use of other government supports such as child care and food stamps. In short, the NSAF data allow us to richly describe the circumstances of low-income families. We separate the sample into groups of families currently on welfare, those that recently left welfare, and those never on welfare despite having incomes similar to those with welfare experience. Changes in the characteristics and circumstances of these three groups in the 1997 to 2002 period may have occurred as a result of shifting welfare policy. Alternatively, changes may reflect changes in the general social and economic environment. Changes in welfare policy may directly affect families never on welfare and lead to compositional changes in the three groups. For example, the strong work message may increase employment among single mothers never on welfare, and diversion and time limit messages may lead some to avoid welfare enrollment. In addition, factors that encourage exits from welfare can lead to a change in the composition of former recipients. Generally, when changes appear in the welfare population but not among families with no welfare history, this suggests some response to the new welfare rules. When changes occur in all three groups, however, associations between outcomes and welfare policy become more complicated. We must be mindful of these effects when interpreting our results. The three groups are defined as follows: 1. The current welfare recipients group includes all individuals that report receiving TANF benefits at the time of their interviews. For some purposes, we further break these current recipients into three subcategories based on their welfare histories: entrants, who first entered welfare in the past two years; cyclers, who first received welfare more than two years ago but have received it intermittently over the past two years; and stayers, who first received welfare more than two years ago and have received it continuously for the past two years. 2. The recent welfare leavers welfare group includes those not receiving TANF at the time of their interview, but who report having left TANF at some point in the last two years The nonwelfare comparison group includes MKAs that report never having received TANF or AFDC benefits in their adult lives. We further restrict all three groups to families with low incomes in which the parent is potentially eligible for TANF. Since much of our analysis focuses on work participation of the MKAs and family circumstances under welfare reform, we want to limit our sample to families where adults are most likely to be affected by welfare THE URBAN INSTITUTE 9

21 policies. We therefore exclude adults who are not parents of the children in the family and parents who report receiving Supplemental Security Income (SSI) for themselves (and are therefore ineligible to receive TANF). In addition, we limit our sample to low-income families to establish greater comparability across groups. Because our nonwelfare comparison group is defined in large part based on income, we wanted our current and former welfare recipient groups to be subject to the same income cut-off. We define low-income as having income below 200 percent of the federal poverty level either last year or in the current year and not above 250 percent of the poverty level in either period. 6 We use current and prior-year income because neither current nor past-year income alone provides us with a consistent definition of low-income across the three groups. Current recipients for the most part meet a current low-income threshold, but they may have had higher incomes in the prior year. On the other hand, former recipients receiving welfare last year would have low incomes in the prior year, but they could have high incomes in the current year. These income restrictions confine all three sample groups to families with consistently low incomes. Table 1 shows the effects of the sample restrictions and final sample sizes for the three comparison groups in the three rounds of the NSAF. As would be expected, the sample restrictions have different effects across the groups. The exclusion of families where the MKA is not the parent of the children or is on SSI had a relatively larger effect on the current welfare groups because these are two important reasons for children s welfare eligibility. About 17 percent of the 1997 current sample and 27 percent of the 2002 current sample were excluded for these reasons. 7 The income screens eliminated about 3 percent of current welfare recipients and 12 percent of recent welfare recipients in 1997 and 9 percent of recipients and 19 percent of recent leavers in The income screen was the primary restriction that established the nonwelfare comparison group. It is clear from this table that the sample restrictions influence the three groups differentially across time. The impact of our income restrictions is to limit the growth in variation in income within each group over time. Thus, the results for changes over time reported here could vary compared to analyses with no income screen on the welfare groups. We completed most analyses in the paper with and without these income screens, however, and found they did not significantly affect the character of the results over time. The main difference in analyses with and without an income screen is the results on the distribution of income and poverty. A subset of results, including income and poverty, for the sample without restrictions is presented in appendix A. Changing Welfare Caseload As noted earlier, the dramatic decline in welfare caseloads generated considerable attention and was used by many as a signal of the success of welfare reform. Caseloads were cut in half between 1996 and 2000 and continued their decline during the 2001 recession, although much more slowly (figure 1). The number of families receiving cash benefits decreased from about 4.6 million in 1996 to roughly 2 mil- 10