Assessing the New Federalism Discussion Paper No. 02-10
Note: The PDF Version includes all charts and tables.
Assessing the New Federalism is a multiyear Urban Institute project designed to analyze the devolution of responsibility
for social programs from the federal government to the states. It focuses primarily on health care, income security, employment and training programs, and social services. Researchers monitor program changes and fiscal developments. Alan Weil is the project director. In collaboration with Child Trends, the project studies changes in family well-being. The project provides timely, nonpartisan information to inform public debate and to help state and local decisionmakers carry out their new responsibilities more effectively.
Key components of the project include a household survey, studies of policies in 13 states, and a database with information on all states and the District of Columbia. Publications and database are available free of charge on the Urban Institute's web site: http://www.urban.org. This paper is one in a series of discussion papers analyzing information from
these and other sources.
This paper received special funding from The Robert Wood Johnson Foundation as part of the Urban Institute's Assessing the New Federalism project. The project received additional funding from The Annie E. Casey Foundation, the W.K. Kellogg Foundation, The Henry J. Kaiser Family Foundation, The Ford Foundation, The John D. and Catherine T. MacArthur Foundation, the Charles Stewart Mott Foundation, The David and Lucile Packard Foundation, The McKnight Foundation, The Commonwealth Fund, the Stuart Foundation, the Weingart Foundation, The Fund for New Jersey, The
Lynde and Harry Bradley Foundation, the Joyce Foundation, and The Rockefeller Foundation.
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.
Data and Methods
As welfare rolls dramatically declined following reform and strong economic growth, much research has focused on families exiting welfare. This research has aimed to describe the characteristics of recipients as well as measure how they were faring once they left welfare (Loprest 1999; Moffitt and Roff 2000a; Isaacs and Lyon 2000). Once recipients leave the system, however, the policy focus shifts from who left to who's left. Families still left on welfare that have had long welfare histories may have the most barriers to self-sufficiency (Zedlewski and Alderson 2000). Few studies have looked at this population.
There are many different ways to measure the well-being of families with long welfare histories. First, the most basic and perhaps important measure is financial well-being and ability to afford basic needs. Second, since welfare reform, welfare agencies must now consider dependency and the ability to work as key outcomes. The addition of work requirements and the dissolution of entitlements in the welfare system increased the population of welfare recipients who combine welfare and work. Given these changes, it is important to consider which long-term welfare recipients are working, the characteristics of their work, and how dependent they are on work versus government assistance for income.
Finally, aspects that might be affected by the financial and vocational situations of families can also be useful to measure a family's well-being. These include physical and mental health, life stresses, social supports, and the well-being of children. Work and income may affect these other forms of well-being, and these other forms of well-being can, in turn, influence parents' ability to obtain and hold jobs, and thus meet the needs of their families.
This study examines a sample of 546 welfare recipients who have long histories of attachment to welfare in the 1990s. The recipients were receiving welfare in either Alameda or Los Angeles
counties in California, a state that, compared with most other states, has high benefit levels, high income eligibility limits, and high child care benefit levels.1 The state also has a relatively lenient sanction policy; it does not impose a full family sanction, meaning that children continue to receive support when their parents do not comply with work requirements (Moreno et al. 1999).
Our first goal in this paper is to examine the work, income, and dependency outcomes of long-term welfare recipient families to determine how much diversity exists in this population. Our second goal is to examine how these families differ in three areas of well-being: health and health care, family stress and hardship, and social and government support, and to ask how work and income might relate to such measures of well-being. We examine three analytic questions:
- How do long-term welfare recipient families vary on work, income, and dependency outcomes?
- When we combine elements of work, income, and dependency, which welfare recipients are faring better and which are faring worse?
- Are the long-term welfare recipients who are working and better off financially also doing better in terms of family and child well-being?
We find that this population of long-term welfare recipients in California was not a homogenous group. Some families achieved self-sufficiency and were out of poverty, others were balancing work and welfare, and the rest were poor and very dependent on welfare. The group was diverse in terms of other measures as well; specifically physical, emotional, and social well-being. For these measures, families that were both working and out of poverty tended to be better off and poor families that were not working were worse off. Families that worked but were still in poverty were doing well relative to the groups on certain measures (stability and family support) but not as well on other measures (physical and mental health and financial hardship).
The Context for Our Analysis
Our study of welfare recipients differs from much of the past research in this field in three important ways. First, our study sample includes only long-term welfare recipients, a population that has not received much attention in the past research. Second, we have available for analysis not only administrative records on welfare history but also survey data on the physical, emotional, and social well-being of our families. Finally, we look within a population of long-term welfare recipients to examine well-studied links between poverty and family and child well-being.
A Focus on Long-Term Recipients
Much research on welfare recipients has focused on recipients who have left the rolls, and many of these "leavers" might have been short-term recipients (Moffitt and Roff 2000b). Researchers have examined the characteristics of leavers and have measured how they were doing since leaving welfare (Loprest 1999; Moffitt and Roff, 2000a; Isaacs and Lyon 2000). This research has shown there is a great deal of work and income diversity among leavers (Moffitt and Roff 2000a). Little research has focused on the population still left on welfare (the "stayers") and most often when this group is studied it is only compared with leavers (Loprest and Zedlewski 1999). Our study focuses on long-term urban welfare recipients with long histories of attachment to welfare in the 1990s. All were receiving welfare in 1998 when they were selected for the study.
Many studies that focus on long-term welfare recipients have been restricted to the use of data on demographics, income, employment, and family structure but do not have the advantage of rich survey data on the well-being of families (Schmidt, Weisner, and Wiley 1998; Boisjoly, Mullan Harris, and Duncan 1998; Moffitt and Wolfe 1993). The few studies that have had such survey data available on family and child well-being (Winston 1999; Morris et al. 2001; National Center for Children in Poverty 2001; Rangarajan and Wood 2000; Furstenberg, Levine, and Brooks-Gunn 1990) did not focus on long-term recipients. In the present study, we examine the overall well-being of long-term welfare recipients, specifically examining measures of family stress and hardship, health, and social support.
Poverty and Child and Family Well-Being in a Long-Term Welfare Context
Much research in the past has focused on the link between poverty and child well-being (Brooks-Gunn and Duncan 1997; McLoyd 1998). Several researchers have identified pathways through which poverty might affect child well-being, such as economic stress and hardship, the home environment, child care quality, social support, parental health, and parenting quality (see Duncan and Brooks-Gunn 2000; Conger et al. 1993; Crnic and Greenberg 1990; Dodge, Petit, and Bates 1994). There is also research that examines the effects of maternal employment on parental health, parenting quality, and child well-being (McLoyd et al. 1994). Just recently this field of research has begun to examine these effects in the context of experimental studies (Hamilton, Freedman, and McGroder 2000; Morris et al. 2001). This research suggests that simply increasing parental employment and decreasing welfare receipt is not sufficient to foster the well-being of parents and children because often the exit from welfare is accompanied by decreases in family income and increases in stress and family instability, which are detrimental to well-being outcomes (Morris et al. 2001). Instead, it may be a combination of work, income, and welfare receipt that matters. We hypothesize that the work, income, and dependency characteristics of the long-term welfare recipients in our study will relate to the aspects of their family and social well-being.
DATA AND METHODS
This study uses data from a telephone survey of 546 long-term welfare recipients in California, collected by the University of California, Berkeley Survey Research Center in 2000 as part of the Urban Institute's Precarious Families survey. Included in the sample were English-, Spanish-, and Vietnamese-speaking welfare recipients who were on welfare in Los Angeles and Alameda counties in 1992 and again (or still) in 1998. This sample is a subset of welfare recipients who participated in California's Work Pays Demonstration Project (WPDP) in the early 1990s.
The sample pool for the WPDP project was all family group (FG) and unemployed parents (U) cases receiving welfare in four California counties in October and December 1992. From this pool, the WPDP sample was selected. As part of the WPDP, telephone interviews were conducted with a subset of the project's participants. The Wave 1 interviews were conducted with 3,560 respondents in 1993 and 1994; the survey response rate was 64 percent. A Wave 2 follow-up survey was conducted with 2,901 Wave 1 respondents in 1995 and 1996; the response rate was 81 percent. For more complete details on the WPDP project see appendix A.
The Precarious Families survey was developed by a team of researchers at the Urban Institute and the University of California, Berkeley to follow up with WPDP survey respondents who were receiving welfare in 1998, the goal being to identify families who were long-term welfare recipients. There were several criteria for selection into the Precarious Families (Wave 3) survey sample:
- The respondents had been surveyed in Wave 1 and 2 (meaning they were on welfare in October and December 1992),
- They had received welfare at some time in 1998 (not including cases only receiving SSI or child-only grants),
- They originally lived in Los Angeles or Alameda county,
- They had received welfare in 1998 in the same county in which they had originally received it in 1992,
- They were a parent or caregiver of a child who would be under age 18 as of January 2000, and
- They spoke English, Spanish, or Vietnamese.
Between February and September 2000, The Survey Research Center at the University of California, Berkeley attempted 734 telephone interviews and conducted 546 interviews for a response rate of 74 percent.
Although we used a proxy for long-term welfare receipt based on whether families were receiving
welfare in both 1992 and 1998, from administrative records we know that our sample has long
histories of welfare receipt in the 1990s. Between December 1992 and October 1997, 94 percent of
our sample had 1 or 2 welfare spells with an average length of 21 or more months (a definition of long-term recipients created by Moffitt and Stevens ). Seventy-seven percent of the sample was on welfare continuously from 1992 to 1997, and the average time on was 4.5 years.
There are three issues to note about the Precarious Families survey data. First, there is intentional bias in the sample resulting from the selection criteria for Wave 3, as well as unintentional bias from nonresponse to each of the three surveys. We explored both of these concerns using logistical regression analyses. Results were mixed. In some ways our sample is biased toward being worse off than the general population of welfare recipients (e.g., more have longer welfare spells and are in poverty and poor health), while in other ways it is better off (e.g., more have a high school degree and are married). For more complete details on selection and nonresponse bias see appendix B.
Second, weights were used on all analyses of the sample. The weighted number of cases is 145,536, an estimate of the number of English-, Spanish-, and Vietnamese-speaking AFDC recipients who were on aid in Alameda and Los Angeles counties in October and December 1992 and also at any time in 1998. A probability weight based on this expansion weight was used in subsequent analyses; this weight maintains the sample size of 546.2
Finally, the over-sampling of Vietnamese speakers yielded a sample size of 142 or 26 percent of the unweighted survey sample. However, when weighted this percentage drops considerably. The
Vietnamese weighted sample size is 18 or 3 percent of the survey sample. All of the analyses
presented in this paper are based on the weighted sample.
Demographics of the Weighted Sample
The sample consisted of 546 respondents the majority of whom were unmarried, Hispanic or African-American women living in Los Angeles County. The average age of the respondents was 39.0 years. This is much older than the average welfare recipient because our sample only included long-term recipients who were in the system in 1992. Ninety-nine percent were female, so throughout this paper we will refer to the respondent as female. Fifty percent of respondents were Hispanic, 30 percent were African-American, 13 percent white, and 7 percent were of another ethnicity (Vietnamese, Native American, Alaskan native, Asian, Pacific Islander, Filipino, or other). Los Angeles was the county of residence for 85 percent of respondents. Twenty-one percent of respondents were married; 30 percent were separated, divorced, or widowed; and 49 percent had never been married. Of those who were unmarried, 15 percent were living in a marriage-like relationship.
Eighty-seven percent of the sample had biological, step, or adopted children living with them who were under the age of 18.3 The average number of children was 2.33. The respondent was asked about one or two of her children. Both a younger focal child (YFC, under age 6) and an older focal child (OFC, between ages 6 and 17) were chosen at random if she had more than one of either, otherwise the one was chosen. Because respondents were older (having been on welfare since 1992), they were less likely to have a child under age 6. Overall, there were 127 families with a YFC (58 percent male; mean age = 2.8 years) and 487 families with an OFC (53 percent male; mean age = 11.5 years). In 93 percent of families the respondent was the biological parent of the YFC, and in 95 percent the respondent was the biological parent of the OFC. Otherwise the respondent was a relative caretaker, an adoptive parent, or a stepparent.
Analysis 1: How Do Long-Term Welfare Recipient Families Vary on Work, Income, and
First, we examine this population of long-term welfare recipient families on three well-being
outcomes of importance to welfare administrators and policymakers: work, income, and dependency.
We find that not all long-term recipients were unemployed, poor, and dependent in 2000. Some are
doing better than others, and some are doing worse. Over half of the families were connected to the workforce. In families where no one is working, many welfare recipients had barriers to work. A substantial number of families were out of poverty and not dependent on welfare. On the other hand, many were living below 100 percent of the federal poverty level (FPL) and dependent on government assistance for over half of their income.4
Over half of the long-term welfare recipients were working in 2000 (see table 1). In 63 percent of families, either the respondent or her spouse or partner was working a regular paying job, and 53 percent of all respondents were working. Working respondents on average had been at their jobs for 2.5 years, working slightly over 35 hours a week, and earning a little over $9 an hour. Many of these working women had overcome several obstacles to hold a job. Thirty-seven percent did not have a high school diploma or GED, 27 percent had a child at home under the age of 6, 28 percent reported being in poor to fair physical health, and 26 percent were in poor mental health. Still, 61 percent were living above the poverty level. Many were combining work and welfare as sources of income; about 40 percent were receiving CalWORKs and 40 percent were receiving food stamps.
In families where neither the respondent nor her spouse or partner was currently working, many respondents had barriers to work. Half had no access to a car, and 24 percent said language was a barrier to getting a job. Of respondents in nonworking households, 62 percent did not have a high school diploma or GED, 28 percent had a child at home under age 6, and 89 percent had no spouse or partner. Thirty-eight percent had a health condition that limited the amount or type of work they could do, 41 percent considered themselves in poor or fair health, and 35 percent were in poor mental health. As would be expected in a population not working, over 80 percent were living below the poverty level, and a substantial number were receiving CalWORKs (68 percent), SSI (18 percent), or food stamps (77 percent).
Income and dependency
Overall, many long-term welfare recipients were out of poverty and nondependent. Thirty-two percent were low-income families (with incomes between 100 and 200 percent of FPL) and 13 percent had incomes at or above 200 percent of FPL (see table 2). Sixty percent of the sample was not dependent on government assistance for income using the definition proposed by the Advisory Board on Welfare Indicators (DHHS 2000). By this definition, welfare dependency is calculated as considered dependent on welfare if more than 50 percent of its total income comes from these sources. About a third of families (34 percent) were not receiving any income from these sources (see table 3). Fifty-nine percent were earning income from the respondent's work and 18 percent from a spouse or partner's work (see table 4). These two sources provided the highest amount of monthly income, averaging $1,170 for respondent's work and $1,440 for the spouse or partner's.
At the other end of the spectrum, a large portion of families were living in poverty and still dependent on government assistance. Fifty-five percent of the long-term welfare recipient families had incomes below the federal poverty level, and 40 percent were dependent on welfare, food stamps, or SSI. Over half (54 percent) of families were receiving CALWORKs, and 57 percent were receiving food stamps. But on average these sources provided much less income than work, around $500 a month for each.
Analysis 2: When We Combine Elements of Work, Income, and Dependency, Which Welfare
Recipients Are Faring Better and Which Are Faring Worse?
It is not sufficient to look at work, income, and dependency outcomes separately. If we only look at one outcomework, for instanceto determine success, we will miss a lot of variation in income and dependency. Although workers in general seem to be doing better than nonworkers, workers are not a homogenous group (table 1). Some workers are off welfare and out of poverty, others combine welfare and work to just reach the poverty level, and still others are highly dependent on welfare and live in extreme poverty despite their connections to the workforce. Nonworkers, as well, are not a homogenous group. Some are living in poverty and are dependent upon CalWORKs for the majority of their income, while others are living above the poverty level, relying on SSI disability assistance or help from friends and relatives. Welfare agencies base a family's success in the system on their level of dependency on government assistance and involvement with the workforce. In addition, a family's overall income will be affected by work and will determine eligibility for welfare. It is the combination of these elementsdependency, work, and incomethat determines which long-term welfare recipient families are faring better and which are faring worse.
To assess the relative success of the families in this sample, we placed them into groups based on 1) whether the respondent or her spouse or partner was working,5 2) family income, and 3) how dependent the family was on welfare, SSI, and food stamps (see table 5). We ended up with three groups:
- Working nonpoor: workers with incomes above 100 percent of FPL,
- Working poor: workers with incomes below 100 percent of FPL, and
- Nonemployed poor: nonworkers with incomes below 100 percent of FPL.
A fourth group, nonworkers with incomes above the poverty level, was very small and was not
included in subsequent analyses.6 These groups differed on several demographic measures, and a discussion of these differences will follow the descriptions of the groups.
Group 1: Nonpoor, nondependent who work
The first group (working nonpoor) consisted of workers who both had incomes above 100 percent of FPL and were not more than 50 percent dependent on welfare for their income. Two hundred and fourteen families (39 percent of the sample) were included in this group. In 93 percent of the families the respondent was currently working at a regular job for pay, in 29 percent a spouse or partner was working, and in 22 percent of these families both were working (see table 6).
The working nonpoor families had a high average monthly income ($2,367) relative to the other groups. Respondent's work pay (mean = $1,538) was by far the most substantial source of income for most working nonpoor families, but 28 percent still took in a large amount of income from a spouse or partner's work pay (see table 6A). The majority (62 percent) of these families were not dependent on CalWORKs, SSI, or food stamps for income. About a third of families were receiving CalWORKs and/or food stamps, but the amount they were receiving was far less than they were earning through work.
Group 2: Poor or dependent but connected to work
The second group (working poor) consisted of 114 families (21 percent of the sample). All families were either working with incomes below 100 percent of FPL or working with incomes above 100 percent of FPL but still greater than 50 percent dependent on government assistance. For simplicity we refer to this group as the working poor throughout the paper. In the vast majority of these families (89 percent) only one adult was working. In most cases it was the respondent (73 percent), but many spouses and partners (38 percent) were working as well.
The mean income for the working poor, at $1,311, was not enough to keep these families either out of poverty or self-sufficient. Half of their income came from work sources, the respondent earned an average of $658 a month and the spouse or partner earned $706. The majority of working poor families were combining income from the respondent's work, CalWORKs, and food stamps, receiving close to equal amounts from work as from government assistance.
Group 3: Poor with no or little connection to work
With 198 families (36 percent of the sample), the third group consisted of nonemployed families with incomes below 100 percent of FPL or those that were working so little that their average monthly work pay was only $232 and their total income was not above 50 percent of FPL. For simplicity we refer to these families as the nonemployed poor. This group had the smallest percentage of families (18 percent) that were working at the time of the interview. Half of the respondents or their spouses or partners reported to have been participating in job or training activities through CalWORKs, however.
Nonemployed poor families had low incomes by definition: all were below 100 percent of FPL (with a mean monthly income of $643) and 62 percent were living in extreme poverty with incomes below 50 percent of FPL. The large majority (73 percent) of families were 75 to 100 percent dependent on government assistance from CalWORKs, SSI, or food stamps for income. CalWORKs was the most substantial source of income for most nonemployed poor families. A large portion of these families' incomes also came from SSI, but only 12 percent received it. The most common source of income was food stamps, but it provided less than half of the income amount of other forms of government assistance.
Demographics of groups
The groups differed by several demographic characteristics (see table 7). The working nonpoor group was the least likely to be Hispanic (39 percent), though Hispanics still made up the largest percentage of each group. The working nonpoor were more likely to be African-American (35 percent) than the working poor (18 percent). The groups did not differ by the age of the respondent or older focal children, or by language of the interview.
The groups were also found to have different family structures. Single-parenting was more prevalent among the nonemployed poor. Twelve percent of this group were married and 7 percent had a partner, which was far fewer than the working nonpoor (26 percent were married, 15 percent had a partner) and the working poor (28 percent and 19 percent). On average, the working poor had more children (mean = 2.7) than both the nonemployed poor (2.1) and the working nonpoor (1.9).
Respondents' level of education also varied across the groups. Fifty-one percent of the working poor and 61 percent of nonemployed poor respondents did not have their high school diploma or GED, far more than in working nonpoor families (32 percent). A similar trend was seen for spouses and partners. This variation across groups should be considered when group comparisons are presented in the next section, as differences found between groups may be caused by many factors.
Analysis 3: Are the Long-Term Welfare Recipients Who Are Working and Better Off
Financially Also Doing Better in Terms of Family and Child Well-Being?
The first two analyses revealed a great deal of variation among families with long-term welfare histories in terms of work, income, and dependency outcomes. Some were working enough to raise their incomes above the poverty level and survive without welfare, while others were working regularly but still in poverty and not self-sufficient. Still others were not working, poor, and dependent on welfare. Although variation in families' economic situations is certainly important, it is also important to examine measures other than financial well-being to more fully understand how families with long-term welfare histories are faring. Perhaps those more successful financially are also better off in other areas of well-being. On the other hand, a family might be working and less dependent on government assistance but doing poorly in terms of social and emotional well-being. By analyzing family stress, health, and social support across our groups of families, we can better understand how income, work, and dependency outcomes are related to outcomes in other areas of well-being.
Family stress, hardship, and instability
Overall we found few associations between financial and vocational success and fa mily stress (see table 8). We did find that working nonpoor families were more likely to have low-risk family environments (61 percent versus 46 percent and 42 percent for working poor and unemployed poor respectively) and less likely to have high-risk family environments (6 percent versus 17 percent for working poor families). Low-risk environments are characterized by high levels of maternal warmth and family routines, low levels of maternal aggravation and family conflict, and good maternal mental health. High-risk environments have the opposite characteristicslow warmth and routines, high conflict and aggravation, and poor maternal mental health. (See Ehrle, Frasch, and Kortenkamp 2001 for a further description of these family environments.) The percentage of children with
behavioral problems, however, did not differ across groups.7 The types of relationships respondents and their children had with noncustodial fathers also did not vary across groups, except on one measure. Working poor mothers were less likely (0.3 percent) to have no relationship with their younger focal children's absent fathers than mothers in the other groups (25 percent for working nonpoor and 14 percent for nonemployed poor). Across the groups we found that large percentages of the older focal children had no contact with their absent fathers (59-70 percent), and their mothers had no relationship with their fathers (33-42 percent).
Measures of hardship and instability revealed many more family differences. As would be expected those families that were poor had more hardships than those that were not poor. The working poor and nonemployed poor families were most likely to have problems paying rent or utilities (37 percent and 33 percent compared with 19 percent of working nonpoor). Working poor families were twice as likely than working nonpoor families to live in crowded housing (41 percent compared with 20 percent). Thirty percent of nonemployed poor families, almost three times more than working nonpoor families, reported not having enough food to eat.
In terms of instability, the families that were not working experienced more difficulties than those that were working. The nonemployed poor were worse off on measures of instability, while the working poor were not statistically worse off on these measures. The nonemployed poor were almost four times more likely to have had no place to live and to have moved two or more times in the year before the interview than the working nonpoor; they were also more likely to have moved than the working poor (11 percent compared with 3 percent).
Health and health care
Respondents in poor families were more likely to have health and health care problems compared with respondents in nonpoor families (see Table 9). Working nonpoor respondents were least likely to report being in fair or poor health or to report a limiting health condition. Thirty-eight percent of working poor respondents reported symptoms of poor mental health according to a five-item scale while only 23 percent of working nonpoor respondents reported symptoms of poor mental health.8 Working poor and nonemployed poor respondents were more than twice as likely as working nonpoor respondents to have been unable to access medical care when they were in need of it.
Overall, work and income did not seem to be related to children's health and health care. Percentages of children in fair or poor health and percentages who could not or did not receive health care were not different across groups. However, children in nonemployed poor families were almost three times as likely to have a limiting condition than those living in working poor families.
While percentages of families with no health insurance did not differ across groups, except for younger focal children, the type of insurance coverage did differ greatly between poor and nonpoor families. Between 20 and 27 percent of respondents and 5 to 10 percent of older focal children did not have health insurance at the time of the survey, and these percentages did not differ by group. However, while only 1 percent of younger focal children from nonemployed poor families were uninsured, 16 percent from working nonpoor families did not have insurance. This was true perhaps because while children in working nonpoor families were by far the most likely to be covered by a private health insurance plan, they were the least likely to be covered by public health insurance. The opposite was true for children in the working poor and nonemployed poor families.
Social and government support
In terms of social support, nonemployed families were less likely than working families to have the support of other adults living in the household (see table 10). In 79 percent of nonemployed poor families the father of the younger and in 91 percent the father of the older focal child did not live in the household. These percentages are higher than in both working nonpoor families (YFC = 56 percent, OFC = 77 percent) and working poor families (YFC = 46 percent, OFC = 69 percent). Also, only 50 percent of nonemployed poor families had an adult living in the household other than the respondent compared with 64 percent of the working nonpoor families.
Working nonpoor families were more likely than the nonemployed poor to have children in child care arrangements, and in particular they were most likely to use nonrelative family day care. Younger focal children from nonemployed poor families were about twice as likely (52 percent) as those from working nonpoor (27 percent) to be in parent care, meaning no other supervised form of care was noted. Younger focal children from working nonpoor families were far more likely to be in nonrelative family day care (22 percent) than younger focal children from other families (3 percent and 0 percent).
Nonemployed poor families were more likely to have received support from private or government
social service agencies than the working nonpoor and in some cases the working poor. Ninety-nine percent of nonemployed poor families who were using child care were receiving some type of assistance for child care costs compared with 70 percent of working nonpoor families. Twenty-eight percent had used a food bank or other meal program in the month before the interview, but only 11 percent of working nonpoor families had used this type of service. Respondents and older focal children in nonemployed poor families were more likely to have received mental health counseling (R = 17 percent, OFC = 15 percent) compared with those in working nonpoor (R = 6 percent, OFC = 5 percent) and working poor (R = 4 percent, OFC = 1 percent) families. Surprisingly, the groups did not differ on measures of receipt and usage of other types of government support such as public housing, Section 8, the Women, Infants, and Children program, subsidized child care, public programs such as Head Start, and child support.
Work, Income, and Dependency
The families in this sample of welfare recipients have a common history. They all have extensive attachments to welfare that date back to at least 1992. They experienced changes brought about by welfare reform in California, within a climate of general economic growth. Many might expect recipients like these to fit a stereotype of a single unemployed mother, poor, and highly dependent on welfare. But results from the 2000 survey indicate this is not the case. It was found that almost one-third of the sample had a spouse or partner, almost two-thirds were working, and over two-fifths were out of poverty. A third received no government income support. After dividing the sample into groups of families based on differences in work, income, and dependency outcomes, we explored how this sample differed in three areas of well-being: family stress, hardship, and instability; health and health care; and social and government support.
Family Stress, Hardship, and Instability
As expected, families out of poverty had fewer financial hardships and families connected to work had less instability. working nonpoor families were the best-off on measures of both financial hardship and instability. The working poor families tended to be worse off on measures of financial hardship compared with the working nonpoor but were not worse off on measures of instability. Measures of instability may be more related to work than to income. Work might enable a family to be more stable and, conversely, a stable household could facilitate a mother's entrance into the workforce. Families not connected to work, the nonemployed poor group, were more likely than the working nonpoor to experience instability indicated by having had no place to live and having moved a lot.
Work alone was not related to the likelihood of having a low-risk family environment, but work coupled with increased income was. More of the working nonpoor had low-risk family environments compared with the working poor and nonemployed poor. This is consistent with evaluations of welfare-to-work programs that suggest increasing employment, but not income, is not sufficient to improve child well-being (Morris et al. 2001). Though family environments varied, measures of child behavior problems did not differ by group. It seems that income and work is most strongly related to family instability and environment and less related directly to child well-being. This would support past research which has found that work and poverty are related to child well-being indirectly through family environment and the stress created by hardship and instability (Dodge et al. 1994; McLoyd et al. 1994; see Duncan and Brooks-Gunn 2000).
Health and Health Care
On measures of health and access to health care, families that were out of poverty tended to be the best off. Respondents in working nonpoor families were faring better than the working poor and the nonemployed poor. Respondents from working nonpoor families were least likely to be in fair or poor health, least likely to have a limiting condition, and most likely to have received care when they needed it. Children in working nonpoor families were most likely to be covered by private health insurance. The one measure on which working nonpoor families looked worse was insurance coverage for the younger focal children. These children were more likely than those in the nonemployed poor group to be uninsured because many nonemployed poor children were covered by public health insurance.
A connection to work, without an increase in income, was not related to improved health of long-term welfare recipients. working poor respondents were more often worse off on measures of health than the working nonpoor. They were just as likely to be in fair or poor health, to report a limiting condition, to have had trouble accessing care when needed, and to have private insurance coverage as the nonemployed poor respondents. working poor respondents were also far more likely to be in poor mental health than working nonpoor respondents.
Because health problems can often be barriers to work, it seems surprising that the health status of the working poor was more like that of other poor families rather than like that of other working families. There are several possible explanations for this. One is that the poorer health status of the working poor might be a barrier to full-time work, and indeed we see that on average working poor respondents were working less than full time at their jobs. Another explanation could be that while a respondent might experience poor health, her spouse would not be prevented from working as a result. The working poor were more likely to have a spouse or partner than the nonemployed poor, and in over a third of working poor families a spouse or partner was working.
Very few differences were found on measures of children's health. Children's health may be more related to mother's health than to work and poverty characteristics (see Duncan and Brooks-Gunn 2000). The only difference found regarding children's health in this study was in the percentage of older focal children with a limiting health condition. The percentage of this occurrence was almost three times higher in the nonemployed poor group compared with the working poor. Having a child with a disability may have prevented some of these mothers from working.
Social and Government Support
Families connected to work were more likely to have another adult besides the mother in the
household. Working families were more likely to have the father of the focal children living in the household. And the working nonpoor group was more likely to have, if not the child's father, at least one other adult in the household. The addition of a father or another adult in the family would increase the likelihood that one adult was able to hold a job.
Long-term welfare recipients who are not connected to work may seem the most vulnerable welfare population and the most in need of services. The nonemployed poor were more likely than the working nonpoor to have gotten free food from a food bank or other program. And they were more likely than working families to have received mental health counseling for the mother or older focal child. Perhaps nonemployed poor families received these counseling services in an attempt to prepare them for work.
It is surprising that the families did not differ on receipt or use of many government supports such as housing assistance, child care subsidies, public child care programs, and child support. The attachment the working nonpoor have had to the welfare system for so long may contribute to them receiving these services at levels similar to poorer families. And indeed some working families may be able to work only because they are receiving help with child care.
In summary, we find a great diversity of outcomes within this population of long-term welfare
recipients. After nearly a decade of attachment to welfare, some families achieved self-sufficiency and were out of poverty, while others were balancing work and welfare, and others were still poor and very dependent on welfare. Policies directed toward long-term welfare populations should, therefore, carefully weigh the progress and hardships of families when determining what the y need.
In addition, there was diversity in the physical, emotional, and social well-being of these families. As expected, families attached to the workforce and out of poverty tended to have better well-being outcomes. Those not working and in poverty tended to have worse outcomes and may require intensive services before being able to find employment. But there were many families between these two extremes. Those families working but still in poverty were doing as well as the nonpoor in some areas, such as stability and family support, but in many areas they were struggling, such as physical and mental health and financial hardship. Perhaps only with additional health, mental health, and job training services will this group be able to complete the trans ition out of poverty.
Boisjoly, Johanne, Kathleen Mullan Harris, and Greg. J. Duncan. 1998. "Trends, Events, and Duration of Initial Welfare Spells." Social Service Review.
Brooks-Gunn, Jeanne, and Greg J. Duncan. 1997. Consequences of Growing up Poor. New York:
Russell Sage Foundation.
Conger, Rand D., K. J. Conger, Glen H. Elder, Jr., Fred O. Lorenz, Ronald L. Simons, and Les B.
Whitbeck. 1993. "Family economic stress and adjustment of early adolescent girls." Developmental Psychology 29 (2): 206-19.
Crnic, Keith A., and Mark T. Greenberg. 1990. "Minor parenting stresses with young children." Child Development 61: 1628-37.
Department of Health and Human Services. March 2000. "Indicators of Welfare Dependence: Annual Report to Congress." Washington, D.C.: Department of Health and Human Services.
Dodge, K. A., G. S. Pettit, and John E. Bates. 1994. "Socialization mediators of the relation between socioeconomic status and child conduct problems." Child Development 65: 649-65.
Duncan, Greg J., and Jeanne Brooks-Gunn. 2000. "Family Poverty, Welfare Reform and Child
Development." Child Development 71: 188-96.
Ehrle, Jennifer, Karie Frasch, and Katherine Kortenkamp. 2001. "Children of Long-Term Welfare Recipients: The Role of Family Environment in Predicting Behavior." Poster presented at the Annual Meeting of the Society for Research in Child Development, Minneapolis, MN.
Furstenberg, Frank F., Judith A. Levine, and Jeanne Brooks-Gunn. 1990. "The Children of Teenage Mothers: Patterns of Early Childbearing in Two Generations." Family Planning Perspectives 22(2):54-61.
Hamilton, Gayle, Stephen Freedman, and Sharon M. McGroder. 2000. National evaluation of
welfare-to-work strategies: Do mandatory welfare-to-work programs affect the well-being of
children? Washington, D.C.: U.S. Department of Health and Human Services.
Isaacs, Julia B., and Matthew R. Lyon. 2000. A Cross-State Examination of Families Leaving
Welfare: Findings from the ASPE-Funded Leavers Studies. Washington, D.C.: U.S. Department of
Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation.
Loprest, Pamela. 1999. Families Who Left Welfare: Who Are They and How Are They Doing? Washington, D.C.: The Urban Institute. Assessing the New Federalism Discussion Paper No. 99-02.
Loprest Pamela, and Sheila Zedlewski. 1999. Current and Former Welfare Recipients: How Do They Differ? Washington, D.C.: The Urban Institute. Assessing the New Federalism Discussion Paper No. 99-17.
McLoyd, Vonnie. 1998. "Socioeconomic disadvantage and child development." American
Psychologist 53 (2): 185-204.
McLoyd, Vonnie, Tobie Epstein Jayaratne, Rosario Ceballo, and Julio Borquez. 1994.
"Unemployment and work interruption among African American single mothers: Effects on
parenting and adolescents." Child Development 65: 562-89.
Moffitt, Robert, and Jennifer Roff. 2000a. "The Diversity of Welfare Leavers." In Welfare, Children, and Families: A Three City Study. Baltimore: Johns Hopkins University.
Moffitt, Robert, and Jennifer Roff. 2000b. Welfare, Children, and Families: A Three City Study (Policy Brief 00-2). Baltimore: Johns Hopkins University.
Moffitt, Robert, and David Stevens. November 2000. "Changing caseloads: Macro influences and micro composition." Paper presented at the conference of the Federal Reserve Bank, New York, NY.
Moffitt, Robert, and Barbara Wolfe. 1993. "Medicaid, welfare dependency, and work: Is there a causal link?" Health Care Financing Review 15.
Moreno, Manuel, Michael Lichter, Elizabeth Gonzalez, John Hedderson, J. Horton, Linda Shaw, and
Jeff Henderson. 1999. Evaluating CalWORKs in Los Angeles County: First in a series of reports on monitoring the effectiveness of CalWORKs implementation in Los Angeles County. Los Angeles, CA: Los Angeles County Department of Public Social Services.
Morris, Pamela, Aletha Huston, Greg J. Duncan, Danielle A. Crosby, and Johannes Bos. 2001. "How Welfare and Work Policies Affect Children: A Synthesis of Research." Oakland, CA: Manpower Demonstration Research Corporation.
National Center for Children in Poverty. 2001. "Why Some Women Fail to Achieve Economic
Security: Low Job Skills and Mental Health Problems Are Key Barriers." The Research Forum on Children, Families, and the New Federalism 4(2).
Rangarajan, Anu, and Robert G. Wood. 2000. Current and Former WFNJ Clients: How Are They
Faring 30 Months Later? Princeton, NJ: Mathematica Policy Research, Inc.
Rowe, Gretchen. 2000. "Welfare Rules Databook State TANF Policies as of July 1999." Washington, D.C.: The Urban Institute. Assessing the New Federalism.
Schmidt, Laura, Constance Weisner, and James A. Wiley. 1998. "Substance abuse and the course of welfare dependency." American Journal of Public Health 88: 11.
U.S. House of Representatives, Committee on Ways and Means. 1999. 1999 Green Book:
Background Material on Data on Programs Within the Jurisdiction of the Committee on Ways and
Means. Washington, D.C.: GPO.
U.S. House of Representatives, Committee on Ways and Means. 2000. 2000 Green Book:
Background Material on Data on Programs Within the Jurisdiction of the Committee on Ways and
Means. Washington, DC: GPO.
Winston, Pamela, Ronald J. Angel, Linda M. Burton, P. Lindsay Chase-Lansdale, Andrew J. Cherlin,
Robert A. Moffitt, and William Julius Wilson. 1999. Welfare, Children and Families: A Three City Study. Johns Hopkins University, Baltimore, MD.
Zedlewski, Sheila R., and Donald Alderson. 2001. Do Families on Welfare in the Post-TANF Era Differ from their Pre-TANF Counterparts? Washington, D.C.: The Urban Institute. Assessing the New Federalism Discussion Paper No. 01-03.
Work Pays Demonstration Project
In 1992, California received a federal waiver that allowed the state to test new approaches to encourage welfare recipients to work, increase their earnings, and decrease their time on public assistance. The California Department of Social Services contracted with the University of California, Berkeley Data Archive and Technical Assistance (UC DATA); the University of
California, Los Angeles Welfare Policy Research Group; and other researchers to evaluate this
demonstration project in four counties: Alameda, Los Angeles, San Joaquin, and San Bernardino.
The sample pool for this project consisted of family group (FG) and unemployed parent (U) cases
receiving welfare in these counties in October 1992 and still receiving in December 1992 (N =
354,476). From this pool the WPDP sample was selected using sampling intervals applied to the
October 1992 caseload based on a target sample (n = 15,000) of 4,000 FG and 2,000 U cases in Los
Angeles county and 2,000 FG and 1,000 U cases in the other three counties. The observed sample (n = 14,537) was less than the target sample due to attrition from October to December and delays in reporting. Only cases that were on in both October and December 1992 were included in the sample. Cases were randomly assigned to experiment or control conditions at a ratio of two to one. Detailed administrative records of the cases' welfare assistance histories from 1987 onward were made available.
As part of the WPDP, a telephone interview (WPDP Survey, Wave 1) was conducted with a subset of the project's participants. These interviews collected information on family characteristics, education and health status, access to child care, housing arrangements, employment history, and income. These data were linked to each case's administrative records. The sample pool for the survey included all WPDP participants whose primary language was one of the six used to conduct the interviews: English, Spanish, Vietnamese, Laotian, Cambodian, and Armenian (n = 13,792). There was an oversampling of Vietnamese, Laotian, Cambodian, and Armenian speakers; cases in Alameda, San Joaquin, and San Bernardino counties; and U cases. The sample size was 5,541 cases. Between October 1993 and September 1994, interviews were conducted with 3560 respondents for a response rate of 64 percent. A follow-up survey (WPDP Survey Wave 2) was attempted with all Wave 1 respondents. Between May 1995 and June 1996, the survey was conducted with 2,901 participants for a response rate of 81 percent.
Selection into Wave 3
In order to obtain a sample of long-term welfare recipients, a number of selection criteria were established to choose the Wave 3 sample. We used logistical regression and data from the Wave 1 survey to analyze the differences between those selected for the Wave 3 sample and those not selected. We expected to find differences given the sampling for only long-term recipients. Those selected for the sample were more likely to be in a minority race category that included Asians, Pacific Islanders, Filipinos, Native Americans and Alaskan Natives, to have more children, to have incomes below 100 percent of FPL, to be less dependent on work for income, and to have moved once in the year prior to Wave 1. They were less likely to have a limiting health condition. These differences can be explained by the fact that Vietnamese speakers were over-sampled, and the selection criteria required the respondent to have a child under age 18, included only those who had received welfare in 1998, and did not include those only receiving SSI.
At each wave of the survey we used logistical regression to predict nonresponse and to determine how the sample might thus be biased. At Wave 1 our nonresponse analysis was restricted by lack of data on non-respondents; we had only the information found in their administrative records. In our logistical regressions we examined race, language, aid code (FG or U), county, age, and time on welfare before Wave 1. Controlling for these factors, we found that respondents compared with nonrespondents were more likely to be Vietnamese and to live in Alameda County. In addition, from survival analyses, we found nonrespondents exited the welfare spell that they were on in December 1992 more quickly than respondents. So our final Wave 1 sample was of welfare recipients with even longer assistance histories than the general population of recipients.
By Wave 2 we had the Wave 1 survey information on nonrespondents to use in analyses. The
independent variables included in the logit covered age first received AFDC, marital status, number of children, education, race, country of birth, health, hardship, employment, and poverty. Controlling for the other factors, Wave 2 respondents were more likely to be married, to have a high school diploma or GED, and to have been born in the United States. This suggests that at Wave 2 our sample was in some ways more well off than the population as a whole.
Respondents looked more like nonrespondents at Wave 3 than at the first two waves. Controlling for the same factors as described for the Wave 2 analysis, Wave 3 respondents were more likely to be in poor health than nonrespondents.
1 California's maximum AFDC grant to families in January 1997 was $565, while the maximum grant for the median state was $377 (U.S. House of Representatives 1999, 416). The average monthly benefit for TANF families in California for fiscal year 1997 was $526.25, while the national average is $318 (U.S. House of Representatives 2000, 382). To be considered eligible for TANF and food stamp benefits in California, the gross income limit for a one-parent family of three persons in January 1997 was $1,360, while the national average for that year was $1,253.91 (U.S. House of Representatives 1999, 416). The statewide limit in California for AFDC and transitional child care in 1996 was $1,068.30 (under age two) and $1,039.20 (over age two). The national averages are $546.00 for children under age two and $696.00 for children over the age of two (U.S. House of Representatives 1999, 694).
2 The weights correct for nonresponse but not for nonresponse bias. They make the respondents representative of the nonrespondents even though these two groups are not the same on several variables.
3 Although it was a part of the selection criteria for the Precarious Families sample that the families have a child who would be under age 18 in 2000, at the time of the interview not all families actually had a child in the household. In addition, 26 families who may not have had children under 18 were mistakenly included in the sample. We decided to keep them in the sample for this analysis since they are still long-term welfare recipients.
4 When income is calculated as percent of the federal poverty level, the cash value of food stamps is not included in total income.
5 Two questions were used to determine whether the respondent or her spouse or partner was working at the time of the interview. Each respondent was asked whether she or her spouse or partner currently had a regular paying job, and if either was working the family was considered to be working. She was also asked how much money from work she and her spouse or partner had earned in the last month. If she reported that either person had earned more than $479 (minimum wage, 20 hours per week) the family was then considered to be working (even if she had claimed they were not working in the previous question). Nine respondents and six spouses or partners were not reported as having a regular paying job but were reported as having work
earnings higher than $440 in the last month.
6 Those families falling into the nonemployed nonpoor group will not be used in the following comparisons because of the small sample size (unweighted n = 17, weighted n = 19) and uniqueness of the group. The majority of families in this group (87 percent) had incomes between 100 and 150 percent of the federal poverty level (the rest had higher incomes), but neither the respondent nor her spouse or partner was working. Most of them were receiving a combination of CalWORKS (81 percent), SSI (77 percent), and food stamps (94 percent) for someone in their family. Sixty-four percent reported a health condition that limited the amount or type of work they could do, and, therefore, they may have been receiving SSI for themselves and may not have been able to work or be self-sufficient.
7 Level of behavioral problems was measured using a 6-item scale. Respondents were asked how often during the past month the child didn't get along with other kids, couldn't concentrate or pay attention for long, and was unhappy, sad, or depressed. The three remaining questions were age specific. Respondents with 6- to 11-year-olds were asked how often during the past month the child felt worthless or inferior; was nervous, high -strung, or tense; and acted too young for his or her age. Respondents of 12- to 17-year-olds were asked how often during the past month the child had trouble sleeping; lied or cheated; and did poorly at schoolwork. Response options were often true, sometimes true, and not true at all. Scale scores range from 6 to 18, and a score of 12 or more indicates high levels of behavior problems.
8 A parent's mental health was measured using a 5-item scale. Respondents were asked how much of the time during the past 30 days they had been a very nervous person, felt calm and peaceful, felt downhearted and blue, been a happy person, felt so down in the dumps that nothing could cheer them up. Response options included all of the time, most of the time, some of the time, and none of the time. Scale scores range from 25 to 100, and a score of 67 or less indicates poor mental health.
See PDF Version for tables.
This survey and study were directly funded by the Stuart Foundation and The David and Lucile
Packard Foundation. This study is part of the Urban Institute's Assessing the New Federalism (ANF) project, a multiyear effort to monitor and assess the devolution of social programs from the federal to the state and local levels. Assessing the New Federalism has received funding from The Annie E. Casey Foundation, the W.K. Kellogg Foundation, The Robert Wood Johnson Foundation, The Henry J. Kaiser Family Foundation, The Ford Foundation, The David and Lucile Packard Foundation, The John D. and Catherine T. MacArthur Foundation, the Charles Stewart Mott Foundation, The McKnight Foundation, The Commonwealth Fund, the Stuart Foundation, the Weingart Foundation, The Fund for New Jersey, The Lynde and Harry Bradley Foundation, the Joyce Foundation, and The Rockefeller Foundation.
The authors would like to thank Greg Acs, Pam Holcomb, Pam Loprest, Stacey Phillips, Tracy Roberts, Fritz Scheuren, and Sheila Zedlewski for their tremendous help at multiple stages of this project.