Explaining Differences in Welfare Caseload Declines Across California’s Counties

Focusing on welfare policy differences at the county level in California, this paper explores the reasons for the differing declines in welfare recipiency rates across the state’s counties. Using a regression framework, it attempts to isolate the effects of differences in counties’ economic characteristics, demographic characteristics, and policy choices on differences in counties’ recipiency rate trends. The paper concludes that two county policies, as well as a few demographic and economic characteristics, may help explain the differing trends in caseloads observed across the state. 1 I would like to thank Professor MaCurdy for his valuable assistance on this project during the last year and a half. I am also grateful to Dana Rapoport and Margaret O’ Brien-Strain for reading preliminary drafts of this paper. All remaining errors are no doubt my fault. Chapter 1: Introduction After the passage of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (PRWORA) on the federal level, states were forced to reform their welfare systems dramatically. One key change among the several federal changes was that PRWORA replaced the federal entitlement to welfare with block grants to the states. Moreover, the new federal law implemented lifetime limits on the receipt of federal funds for welfare benefits and required a minimum level of work participation from many welfare recipients. On August 11, 1997, Governor Wilson signed the California Work Opportunity and Responsibility to Kids (CalWORKs) legislation, which brought California into compliance with the new federal law. While CalWORKs satisfied the new federal requirements, California’s welfare policy remained quite distinctive. Part of the reason for this was simply that California chose to have more generous benefits, and a less strict sanction policy, than many other states. Another aspect of CalWORKs that made it stand out from other states’ welfare programs was that California devolved an unusual amount of responsibility to its counties, allowing them a high degree of flexibility within the framework of CalWORKs. More than four years after CalWORKs was created, it is still unclear how effectively the program has helped welfare recipients attain self-sufficiency. To fully understand the effects of the reforms on the long term well-being of families will take time. However, even now, one feature of welfare reform in California is completely clear: For whatever reason, there have been large declines in welfare caseloads across the state. A convenient measure for the level of welfare caseloads that this paper will use extensively is a recipiency rate, which is defined as the number of welfare cases per thousand women aged 15 to 44. Figure 1.1 displays the steady decline in California’s recipiency rate since 1996, when welfare reform was enacted on the federal level. Moreover, this decline in caseloads is not confined to only particular parts of California. All but one county – Alpine – has had its recipiency rate fall in the period from October 1997 to December 2000, which will be the period on which this paper’s empirical analysis will focus. Seven counties – El Dorado, Mariposa, Napa, Placer, San Diego, San Mateo, and Sonoma – have 2 In calculating the recipiency rate, I choose to use the number of women aged 15 to 44. This is the most relevant segment of the population when considering welfare recipients, since a large proportion of recipients are women in this age group. Some researchers prefer to use the total population as the denominator in the recipiency rate calculation. Though this is also a reasonable measure, it is problematic because it includes groups that are not commonly found on welfare rolls, such as male senior citizens. Figure 1.1: Statewide Recipiency Rates 0 20 40 60 80 100 120 140 Ja n96 M ar -9 6 M ay -9 6 Ju l-9 6 Se p96 No v96 Ja n97 M ar -9 7 M ay -9 7 Ju l-9 7 Se p97 No v97 Ja n98 M ar -9 8 M ay -9 8 Ju l-9 8 Se p98 No v98 Ja n99 M ar -9 9 M ay -9 9 Ju l-9 9 Se p99 No v99 Ja n00 M ar -0 0 M ay -0 0 Ju l-0 0 Se p00 No v00 Month R ec ip ie n cy R at e seen their recipiency rate decline more than 50% over this period. In order to get an overview of how caseloads have fallen in different parts of the state, I examine the trends in caseloads by region. A convenient regional division of California is in “An Examination of Welfare Caseload Dynamics in California Using Administrative Micro-Data,” written by Gritz, Lieberman, Mancuso, and Scroggins. There, California is divided into five regions, each of which contains counties with similar economic and demographic characteristics. Figure 1.2 displays the five regions, which are the Bay Area, Los Angeles, Southern California (excluding Los Angeles), the Farm Belt, and the North and Mountain counties (Gritz, et al., 11). Table 1.1 lists the counties that each region includes. Throughout this paper, I will refer to these regions repeatedly, since they allow us to get a basic understanding of important patterns without resorting to an examination of each individual county. Every one of the five regions has seen an essentially uninterrupted decline in its recipiency rate. In Figure 1.3, I graph the recipiency rate trends in the late 1990s for each region. Table 1.1 Region Counties Included in Region Bay Area Alameda, Contra Costa, Marin, Napa, San Francisco, San Mateo, Santa Clara, Santa Cruz, Solano, Sonoma Los Angeles Los Angeles Other Southern California Orange, Riverside, San Bernardino, San Diego, Santa Barbara, Ventura Farm Belt Colusa, El Dorado, Fresno, Glenn, Imperial, Kern, Kings, Madera, Merced, Monterey, Placer, Sacramento, San Benito, San Joaquin, San Luis Obispo, Stanislaus, Sutter, Tulare, Yolo, Yuba North & Mountain Alpine, Amador, Butte, Calaveras, Del Norte, Humboldt, Inyo, Lake, Lassen, Mariposa, Mendocino, Modoc, Mono, Nevada, Plumas, Shasta, Sierra, Siskiyou, Tehama, Trinity, Tuolumne Figure 1.2 (from Gritz, et al.) Figure 1.3: Regional Recipiency Rates 0 20 40 60 80 100 120 140 160 180 200 Ja n96 M ar -9 6 M ay -9 6 Ju l-9 6 Se p96 No v96 Ja n97 M ar -9 7 M ay -9 7 Ju l-9 7 Se p97 No v97 Ja n98 M ar -9 8 M ay -9 8 Ju l-9 8 Se p98 No v98 Ja n99 M ar -9 9 M ay -9 9 Ju l-9 9 Se p99 No v99 Ja n00 M ar -0 0 M ay -0 0 Ju l-0 0 Se p00 No v00 Month R ec ip ie n cy R at e bayarea socal farmbelt northmtn losang Although the recipiency rate is consistently higher in the Farm Belt, the North and Mountain counties, and Los Angeles than in the Bay Area and Southern California (excluding L.A.), even these regions have seen large declines. In percentage terms, the Bay Area experienced the greatest decline in its recipiency rate between October 1997 and December 2000, a decline of about 44.8%. In contrast, Los Angeles had the smallest decline, a decline of 26.6%. The other three regions fall in between these extremes, as Southern California, the North and Mountain counties, and the Farm Belt had declines of 41.3%, 38.3%, and 31.8%, respectively. In absolute, rather than relative, terms, the North and Mountain region experienced the greatest decline in caseloads. The variability in caseload trends from one region to another emphasizes the fact that different parts of California have had different experiences in the post-reform period. Though most counties witnessed a rapid fall in their total welfare caseloads, some counties experienced greater caseload declines than others, as already mentioned. In fact, the standard deviation of the counties’ percentage change in recipiency rates from October 1997 to December 2000 is 12 percentage points. What explains these differences in caseload changes across California’s counties? On the national level, many researchers have tried to answer a similar question about the differences in caseload changes across the states. Most of their research focuses on disentangling the effects of three sets of factors – economic characteristics, demographic characteristics, and policy choices. To answer the question about differing caseload trends across California’s counties, we must confront the same issues. Just as on the national level, economic and demographic characteristics may help explain the movements in welfare caseloads in the counties. Holding all else equal, counties with stronger economies likely experienced greater caseload reductions than counties with weaker economies. Moreover, demographic characteristics in some counties may have increased the proportion of the population in danger of going onto welfare or have prevented welfare recipients from finding employment easily. Finally, just as there are significant differences in welfare policy from one state to another, there are important differences in policy choices from one California county to another. Unlike counties in many other states, California counties were given much flexibility in designing their welfare programs under the new welfare regime. Though choices made at the county level may not be as significant as choices made at the state level (such as setting the level of benefits), they are important nonetheless. To take an example, CalWORKs required adults on single parent assistance units to participate in work activities, such as unsubsidized employment or on-the-job training, for a minimum of 26 hours per week from July 1, 1998 to July 1, 1999. However, counties had the option of requiring up to 32 hours of participation per week in this time period. Requiring more hours of work per week could have strongly affected welfare caseloads, since some current recipients or potential recipients could have been unable or unwilling to participate this much. This paper will explore the effects of economic, demographic, and policy factors on recipiency rates in California’s counties during a period from late 1997 to late 2000, after the passage of CalWORKs. More specifically, it will try to answer the following research question: 3 The story of California counties’ experience in the post-reform period becomes even more complicated if we disaggregate the total caseload into the three major case types – one-parent, two-parent, and child-only. The appendix contains brief definitions of the three case types, as well as figures that display the recipiency rate trends for each case type in each region. To what degree are differences in caseload declines across California’s counties attributable to policy differences among the counties, as opposed to differences in economic and demographic characteristics? Finding an answer to this question is helpful to not only California’s county policy makers, but also federal policy makers. In the average month in the year 2000, about 22% of the national caseload resided in California, so any policy that reduces the caseload in California can have a large impact on the amount of federal funds spent on TANF in the future (Statistical Abstract of the United States: 2001, 354). The paper proceeds as follows. Chapter 2 indicates that there are meaningful differences in policy choices across California’s counties and discusses the possible effects of these differences on counties’ recipiency rates. Chapter 3 points out the county economic and demographic characteristics that may also help explain the trends in recipiency rates over my observation period. Having covered this background, the paper then discusses in Chapter 4 the methods which previous researchers have used to disentangle the effects of economics, demographics, and policy choices on welfare caseloads. Chapter 5 describes the methodology and data that my empirical analysis uses. Chapter 6 presents the results of the empirical analysis, and Chapter 7 provides the conclusions of the paper. Chapter 2: Counties’ Flexibility in Creating Their CalWORKs Programs Before the passage of PRWORA, federal law prescribed a rigid framework for states’ welfare programs, but states were still left with some crucial decisions on issues such as the grant level. A similar situation has existed in California in recent years, as the state requires county CalWORKs programs to adhere to a large set of rules, but also allows the counties to make several key decisions in the creation of their programs. The first section of this chapter outlines the most important rules prescribed by the state. Covering this background material facilitates the second section’s discussion of counties’ flexibility within the framework provided by state law. Basic Provisions of California State Law Before reviewing the areas in which counties’ welfare programs can vary, it is helpful to first see the ways in which they must be the same. To begin with, CalWORKs fixes the welfare benefit levels for counties. Legislation adopted in the pre-reform era made grant levels in seventeen counties about five percent higher than those in the remaining counties in order to compensate recipients for differences in the cost of living. The seventeen counties that have a higher benefit level are: Alameda, Contra Costa, Los Angeles, Marin, Monterey, Napa, Orange, San Diego, San Francisco, San Luis Obispo, San Mateo, Santa Barbara, Santa Clara, Santa Cruz, Solano, Sonoma, and Ventura (Gritz, et al., 53). In terms of time limits, CalWORKs imposes a five-year cumulative lifetime limit on aid for individuals, as well as an 18/24-month time limit relating to work participation. While the lifetime limit is well understood, it is also important to understand the 18/24-month time limit. After the provisions of CalWORKs took effect in early 1998, new applicants for welfare can receive aid for 18 months. If they fail to find employment at the conclusion of 18 months, then the recipients must begin participating in a community service program run by their county if they wish to continue receiving benefits. Those recipients who were part of the Aid to Families with Dependent Children (AFDC) program, which was replaced by CalWORKs, faced a time limit of 24 months, instead of 18. 4 Although the material covered in this chapter is public knowledge, I should note for the interested reader that the listing of areas of county flexibility draws from RAND’s “Welfare Reform in California: State and County Implementation of CalWORKs in the Second Year,” by Klerman, et al. Furthermore, CalWORKs requires adult welfare recipients to participate in welfare-towork activities. Such activities are defined broadly, and include job search, unsubsidized or subsidized employment, on-the-job training, participation in mental health, substance abuse, or domestic abuse services, and education for the purpose of becoming employed. Job search is usually the initial step of the welfare process for a recipient. After the job search phase, adults in assistance units are required to work a certain number of hours, depending on whether they are part of a one-parent or two-parent case. As of January 1, 1998, single parents were required to work a minimum of 20 hours per week. This minimum requirement for single parents increased to 26 hours per week on July 1, 1998, and increased again to 32 hours per week on July 1, 1999. In contrast, the hourly requirement for two-parent families has not altered since the passage of welfare reform. Since 1998, the parents in a two-parent assistance unit have had to work a combined 35 hours per week. Currently, single parents and parents in a two-parent assistance unit cannot be required to work more than 32 hours per week and 35 hours per week, respectively. There are several caveats to these participation requirements. A number of people, such as individuals who are disabled, are exempt from work requirements. If counties wish for recipients to participate in welfare-to-work activities, then they are expected to provide recipients with supportive services, such as transportation and child care assistance. When a recipient has not met participation requirements and has been sanctioned, the county must continue to make grant payments to the children on the assistance unit. Key Policy Decisions That Are Left to the Counties As seen above, there are many rules that all counties must abide by. Among the most important of these are rules concerning the grant level, time limits, and work requirements. Nevertheless, this framework provides counties with a large amount of flexibility in designing their particular welfare programs. For instance, counties can exempt individuals from the 5-year time limit as long as they find a justifiable reason. If a county exempts more than 20% of its caseload, however, it may face financial sanction if the state exceeds the exemptions allowed by federal law. Therefore, there is a practical limit to how many recipients a county is willing to exempt. Moreover, counties can choose to extend the 18-month time limit to 24 months if an extension is likely to help a recipient find unsubsidized employment, or if jobs are not available in the local area. Some counties interpreted this liberally and extended this time limit to 24 months for all new applicants. Counties also have leeway in their implementation of participation requirements. While less than 32 hours of work per week were required of single parents up until July 1, 1999, counties had the option of requiring up to 32 hours per week as of January 1, 1998. Counties can exempt a mother who gives birth to a first child for anywhere from 3 months to 12 months; for subsequent children, the possible exemption period ranges from 3 months to 6 months. Moreover, counties can exempt any other individual that they believe is unable to participate in work activities; this can occur when a county is unable to provide adequate child care services, if a participant is a victim of domestic violence, etc. Again, a county must use these open-ended exemptions carefully, because the county is liable for half of the federal penalties levied on the state for violation of the PRWORA work requirements. While time limits and participation requirements are the most apparent areas of county flexibility, counties are given the latitude to make important choices about their welfare programs along other dimensions as well. They can determine whether or not to provide a “diversion” payment (a one-time cash payment) to an individual in order to avoid enrolling the individual in CalWORKs. They can make decisions about the staff arrangements. For instance, they can choose whether or not to co-locate their eligibility and employment staff, to combine the jobs of their eligibility and employment workers, to co-locate staff from welfare and mental health and substance abuse agencies, to use interagency teams to provide case management, and more. While all counties must provide services such as job search support and child care, the counties can decide whether or not to contract these services out to a private provider. This flexibility is especially evident in the manner counties personally work with their welfare recipients. For example, counties can decide how best to serve the portion of their caseload that does not speak English. They can determine whether to visit welfare recipients’ homes, and in what instances to make such visits. In one final, important example of county flexibility, counties determine how strictly to enforce their sanction policies. Therefore, essentially every small issue is left to the county’s discretion by CalWORKs. Many of these small choices have potentially large impacts on outcomes for welfare recipients. Though all these decisions could affect the recipiency rate in a county, it is not feasible to examine the effects of all these policy choices in a regression framework. For reasons detailed in