Table of Contents | Previous | Next |
Chapter 6
How Are Families Faring After Time Limits?
Chapter 5 examined what is known about the effects of time limits on key outcomes such as employment, income, and welfare receipt. But some of the key questions about time limits are descriptive in nature: Many observers want to know how former recipients and their families are faring after benefit termination. Are they working? Are they receiving other forms of public assistance? Do they experience severe hardships such as homelessness, hunger, or removal of their children?
It is too soon to provide definitive answers to these questions, but a series of state and federally funded post-time-limit studies have yielded a wealth of early data to inform policy-makers and administrators. The studies have the same limitation as other studies of welfare leavers — data on the post-welfare circumstances of families do not necessarily provide evidence about the effects of welfare reform — but they are useful nonetheless.
This chapter reviews the results of eight surveys of individuals whose welfare cases were closed because of time limits. Most of the surveys were conducted 6 to 18 months after the respondents left welfare, and all obtained relatively high response rates. All the surveys were conducted in states with time limits of fewer than 60 months: Connecticut (21 months), Florida (24 or 36 months), Massachusetts (24 months), North Carolina (24 months), Ohio (36 months), South Carolina (24 months), Utah (36 months), and Virginia (24 months).1 (See Chapter 2 and Appendix A for more information on the time-limit policies in these states.)
As discussed in Chapter 3, these eight states account for a majority of the families who have lost benefits because of time limits, but the focus on shorter time limits means that the surveys do not provide evidence about the 60-month time limit on federally funded assistance. Because there are no restrictions on the use of federal funds for families who exceed state time limits of fewer than 60 months, state exemption and extension policies might differ at the 60-month point. Also, it is important to note that all the surveys took place during periods of low unemployment, when jobs were plentiful for recipients whose benefits were canceled at a time limit.
The chapter discusses three kinds of comparisons that are used to assess the experiences of time-limit leavers:
-
State-to-state comparisons. One section presents outcomes across the eight states. There is wide variation in the results, making it difficult to draw general conclusions. Many of the differences in outcomes for time-limit leavers can be explained by differences in the states’ welfare policies that shaped the size and characteristics of the group of families whose benefits were canceled. These results show that it is impossible to interpret such surveys without information about the states’ policies and their implementation.
-
Before-after comparisons. Most of the studies compare respondents’ circumstances before and after leaving welfare. These data are suggestive, but it is not possible to attribute any changes to the fact that the families’ welfare grants were terminated. Also, in many states, time limits were introduced as part of a package of reforms that also included expanded earned income disregards (see Chapter 1). A recipient who mixed work and welfare and then had her case closed at a time limit would experience a loss in income but might still end up with similar (or greater) income than she would have had under the old welfare rules.
-
Comparisons across groups of leavers. Several of the studies compare time-limit leavers with individuals who left welfare for other reasons. These comparisons are also useful, but is not possible to determine to what extent differences in post-welfare outcomes are attributable to the exit reason (that is, to the fact that some people were terminated from welfare at the time limit), as opposed to the differing characteristics of people in the groups.
Ultimately, many observers will undoubtedly compare the survey results with their own standards of what “ought to be” — regardless of what role time limits played in producing the outcomes. Some might conclude that the levels of employment, income, or hardship are satisfactory, while others might find them unacceptable.
The chapter begins by presenting the key findings and discussing the characteristics of families whose cases were closed at time limits. It then uses the three types of comparisons described above to focus on employment, receipt of government benefits, hardship, and other topics. Appendix C provides background information on the surveys.
Key Findings
-
Characteristics. Most studies found that individuals who lost benefits because of time limits were more likely to have large families, to live in public or subsidized housing, to lack a high school diploma, and to be African-American, when compared with people who left welfare for other reasons. These characteristics overlap, however, and it is not clear which are independently associated with reaching a time limit or having one’s benefits canceled.
-
Employment. Post-exit employment rates vary widely across states, ranging from less than 50 percent to more than 80 percent. Most of the variation is attributable to state welfare policies that shape who reaches the time limit (for example, sanctioning and earnings disregards) or to state time-limit extension policies. As a consequence, employment rates are higher for time-limit leavers than for other leavers in some states, and they are lower in other states. For the most part, post-exit employment rates are similar to pre-exit employment rates; in other words, there is little evidence that large numbers of people responded to the termination of their benefits by going to work, although the overall rates can hide dynamic employment patterns.
-
Public assistance receipt. Large proportions of time-limit leavers continue to receive Food Stamps, Medicaid, and other assistance after exit. The variation in Food Stamp receipt across states largely tracks the differences in employment rates (that is, the rate of Food Stamp receipt is lowest in states where most time-limit leavers are working). However, time-limit leavers are more likely than other leavers to receive Food Stamps, even in states where their post-exit employment rate is higher.
-
Income. Most time-limit leavers reported low household income. In all states, some time-limit leavers reported that their post-welfare income or standard of living was higher than when they received welfare, while others reported being worse off. The proportions vary, but, in most states, a greater proportion of respondents said that they were worse off after leaving assistance. In general, employed respondents reported higher household income than nonworking respondents. Similarly, in states where time-limit leavers have lower employment rates than other leavers, they also have lower income.
-
Material hardship. Most time-limit leavers were struggling financially, and, in most states, the leavers reported that they experienced more hardships after leaving welfare than before. Homelessness has been quite rare, but levels of food insecurity and other hardships are relatively high. However, there is not a clear association between levels of hardship and employment status, and, in most states, time-limit leavers did not report consistently greater levels of hardship than other leavers.
Who Loses Benefits Because of Time Limits?
Several of the studies compare the demographic characteristics of time-limit leavers with the characteristics of people who were subject to time limits but left welfare before reaching them.2 Such data are important because they may help administrators predict which types of recipients are most likely to reach limits. Also, as noted earlier, the differing characteristics may explain some of the differences across groups of leavers in the post-welfare outcomes discussed later in the chapter.
It is important to note that, in order to gather information quickly, most of the states surveyed the first cohort of recipients to reach the time limit. This group is likely to include mostly people who received benefits continuously until they reached the limit, and often for long periods before becoming subject to the limit. Recipients who reached the time limit after cycling off and back onto welfare perhaps a somewhat less disadvantaged group are probably underrepresented.
Table 6.1 shows selected characteristics of time-limit leavers and other leavers in each state. Some of the patterns are quite consistent across states: Notably, time-limit leavers are more likely to have three or more children and are more likely to be living in public or subsidized housing. In most states, time-limit leavers are less likely to have a high school diploma and are more likely to be African-American.3
Although not shown in the table, several studies have found, not surprisingly, that time-limit leavers are more likely to have long histories of prior welfare receipt. Most (but not all) of the studies found that time-limit leavers are older, on average, than people who left welfare before reaching limits.
In addition to the comparisons shown, the South Carolina study also compared time-limit leavers with individuals who left because of sanctions. Sanctioned leavers were younger, less likely to have completed high school, and less likely to have been receiving housing subsidies.
| State/Evalutation | High School Diplomaa (%) | Three or More Children (%) | African-American (%) | Subsidized Housingb (%) | ||||
|---|---|---|---|---|---|---|---|---|
| TL | No TL | TL | No TL | TL | No TL | TL | No TL | |
| Connecticutc | 69 | 68 | 26 | 19 | 47 | 37 | 42 | 27 |
| Florida FTPd | 53 | 60 | 35 | 25 | 70 | 50 | 35 | 22 |
| Massachusetts | 68 | 76 | 37 | 33 | 18 | 21 | 56 | 50 |
| North Carolinae | 72 | 69 | 16 | 32 | 65 | 64 | 52 | 30 |
| Ohiof | 45 | 57 | 54 | 30 | 83 | 71 | 69 | 38 |
| South Carolina | 52 | 61 | 54 | 32 | 93 | 71 | 35 | 22 |
| Utahg | 55 | 70 | n/a | n/a | 5 | 3 | 44 | 40 |
| Virginiah | 60 | 61 | 35 | 15 | 51 | 48 | 40 | n/a |
| SOURCES: Connecticut: Bloom et al., 2000; Florida
FTP: Bloom et al., 2000, and MDRC calculations; Massachusetts Department
of Transitional Assistance, 2000; North Carolina: Richardson et al.,
1999, and Richardson et al., 2000; Ohio: Bania et al., 2001; South
Carolina: Richardson et al., 2001, and unpublished data; Utah: Taylor
et al., 2000; and Virginia: Gordon et al., 1999.
NOTES: Unless otherwise noted, the data were collected from the follow-up surveys, and the non-time-limit group includes people who left welfare at roughly the same time as the time-limit leavers. aIncludes people with a high school diploma or GED certificate.(back) bIncludes people living in public housing and those receiving housing assistance vouchers.(back) cThe time-limit group includes people who enrolled in the program between January and June 1996, reached the 21-month time limit by March 1998, and had their benefits canceled. The non-time-limit group includes people who received fewer than 21 countable months of benefits during the same period (some of them may have received exemptions that stopped their time-limit clocks). The data were collected at the point people first became subject to the time limit.(back) dThe non-time-limit group includes people who were subject to a 24-month time limit and received fewer than 24 months of benefits in the four years after becoming subject to the limit, and those who were subject to a 36-month limit and received fewer than 36 months of benefits. The data were collected at the point people first became subject to the time limit.(back) eThe time-limit leavers left welfare in August 1998; the non-time-limit leavers left welfare between December 1998 and April 1999 in eight counties. A very small number of those classified as non-time-limit leavers appear to have reached the 24-month time limit.(back) fRace, number of children, and high school diploma status are drawn from administrative records.(back) gFigures for the non-time-limit group were calculated from separate figures for individuals who left because of increased income and people who left for other reasons.(back) hTime-limit figures are for survey cohort 1 only and are drawn from administrative records.(back) |
The demographic characteristics overlap to some extent, and the studies do not make extensive efforts to determine which are independently important in predicting that a recipient will reach a time limit. Having several children is likely to be important both because it limits employability and because recipients with larger families must earn more to lose eligibility for assistance before reaching a time limit.4 Housing subsidies may be correlated with other demographic factors but may also independently affect incentives to work and/or leave welfare. The Connecticut study examined whether African-Americans were more likely to reach the time limit, after controlling for other characteristics. Some, but not all, of the racial disparity disappeared when other factors were held constant.5
The Connecticut study also shows that such analyses can be complicated in situations where many recipients receive time-limit extensions or exemptions; these individuals receive enough months of assistance to reach a time limit, but they do not have their cases closed. In Connecticut, individuals who reached the 21-month time limit and were granted an extension appeared to be more disadvantaged than either the time-limit leavers or the people who did not accumulate 21 months of receipt.
Post-Time Limit Outcomes: State-to-State Comparisons
By examining outcomes across the eight studies, one may be able to draw some general conclusions about the circumstances of families after time limits. This section summarizes the state data on employment, receipt of government benefits, income, and hardship. To avoid comparing “apples and oranges,” the data are drawn from the first follow-up survey conducted in each state. Results from longer-term follow-ups conducted in North Carolina and Virginia are discussed in a later section.
Employment and Job Characteristics
Many people focus on employment as a key outcome for welfare leavers because work is one of the main sources of income for such families. The first column of Table 6.2 shows the percentage of survey respondents who were working when interviewed in each of the eight studies. The figures in parentheses show approximately how many months after exit the interviews took place.
| State/Evalutation | Employed (%) | Job Characteristics | |
|---|---|---|---|
| Average Hourly Wage ($) |
Average Hours per Week |
||
| Connecticut (6)a | 83 | 7.82 | 35 |
| Florida FTP (varies)b | 54 | 6.11 | 32 |
| Massachusetts (10) | 73 | 8.21 | 31 |
| Ohio (6)c | 53 | 7.30 | 33 |
| North Carolina (6) | 63 | 6.51 | 31 |
| South Carolina (12)d | 50 | 6.00 | 34 |
| Utah (2-5)e | 43 | 6.41 | 27 |
| Virginia (6)f | 66 | 6.50 | 34 |
| SOURCES: Connecticut: Hunter-Manns and Bloom,
2000; Florida FTP: Bloom et al., 2000, and MDRC calculations; Massachusetts
Department of Transitional Assistance, 2000; Ohio: Bania et al., 2001;
North Carolina: Richardson et al., 1999, and unpublished data; South
Carolina: Richardson et al., 2001; Utah: Taylor et al., 2000; Virginia:
Gordon et al., 2002.
NOTES: The figures in parentheses show approximately how many months after exit the interviews took place. aHours per week are for all current jobs; hourly wages are for primary current job.(back) bIndividuals were subject to 24- or 36-month time limits, and were interviewed approximately 48 months after random assignment. On average, the interview took place about 20 months after exit.(back) cWages and hours are for current or most recent job.(back) dEmployment rate is for those still off welfare when interviewed.(back) eHourly wages were calculated from monthly earnings and hours per week.(back) fWages and hours are for the current or most recent job.(back) |
The employment rates vary dramatically, from less than 50 percent in Utah to more than 80 percent in Connecticut. Results from before-after comparisons, discussed below, suggest that the employment rates shown in Table 6.2 largely reflect respondents’ employment status when they were still receiving benefits. In other words, these results do not mean that respondents in Connecticut were most successful in finding employment after losing benefits but, rather, that the people whose benefits were canceled because of the time limit in Connecticut were very likely to have been working while on welfare.
Indeed, as discussed in Chapter 4, in Connecticut, a generous earnings disregard allows many people to mix work and welfare, and recipients who reach the time limit without a job (or with very low earnings) almost always receive at least one 6-month benefit extension; those who are earning above the welfare payment standard are not eligible for extensions. As a result, a very high percentage of the people whose cases were closed because of the time limit were already working while on welfare. Table 6.2 shows that most of them continued to work in the six months after losing benefits.
Virginia’s policies regarding work and sanctioning affect which recipients are likely to stay on welfare long enough to reach the time limit. Recipients who become subject to the state’s time limit must begin working (or take an unpaid community work experience position) within 90 days, or else they lose their entire TANF grant. In addition, a substantial proportion of the state caseload is exempt from the time limit. Thus, as the authors of the study point out, most of the people who reach the time limit will be those who have been working and meeting program requirements. Some of the recipients who might have been expected to reach the time limit were probably exempted or left welfare earlier because of full-family sanctions.
Massachusetts has a high post-time-limit employment rate, and the median length of time in the current job was 10 months (not shown). Since the survey was conducted about 10 months after exit, this suggests that most people were already employed while on assistance. The state granted few extensions during the period of the survey, but a large percentage of the state’s welfare caseload is exempt from the time limit (see Appendix B). It may be that the minority of recipients who were subject to the limit were quite likely to find employment before reaching month 24 even though many of them were not subject to a work requirement and the state’s relatively high grant level and generous disregard allowed many of them to continue receiving benefits until they reached the limit. Other recipients may have been subject to full-family sanctions that prevented them from reaching the limit.6
In contrast, the evaluation of Florida’s Family Transition Program (FTP) found that a diverse group of people reached the time limit, in part because there were no full-family sanctions in place at the time. Very few extensions were granted, so the group that left because of the time limit included many recipients who were employed as well as many who were not. The low employment rate in South Carolina may reflect the state’s low TANF grant level and relatively short-term earnings disregard; this combination limits the number of people who mix work and welfare long enough to reach the time limit. Also, few extensions were granted during the early years of the time limit’s implementation (see Chapter 4).
Across all states, most of the recipients who were employed at follow-up were working full time or close to full time. There are large differences in hourly wage rates, which probably reflect the characteristics of local labor markets. The differences may also reflect the personal characteristics of the recipients whose cases were closed due to the time limit, but this is not clear. For example, the percentage of time-limit leavers with at least a high school diploma is higher in South Carolina than in Cuyahoga County, Ohio (and the employment rate is similar), but the average hourly wage is much higher in Ohio.
Finally, several of the studies asked nonworking respondents to identify reasons why they were not working. The responses differed from state to state, but health problems, an inability to find work, and a desire to attend school were frequently mentioned in most studies. In-depth interviews conducted as part of the Florida FTP study found that at least some of the nonworking respondents were not actively seeking work because they were being supported by a parent or partner. It is impossible to say whether the respondents were relying on other supports because they were unable to work or whether the presence of the other supports allowed people not to work when they could have. Later sections discuss whether nonworking respondents appear to be systematically worse off than working respondents.
Receipt of Government Benefits
Table 6.3 shows the percentage of respondents who reported receiving various forms of public assistance when interviewed. A very high percentage of respondents in all states were probably receiving both Food Stamps and Medicaid while on assistance, but there is wide variation, particularly in Food Stamp receipt, at the follow-up points.
Much of the variation in Food Stamp receipt is probably related to the employment and earnings data discussed in the previous section. For example, earned income was highest in Connecticut and Massachusetts (both the employment rates and the earnings were highest), suggesting that fewer respondents were eligible for Food Stamps in those states. Conversely, earned income was lowest in South Carolina and Utah, which have the highest rates of Food Stamp receipt. However, the association is not perfect, which suggests that state and local pre- time-limit procedures may affect the likelihood that eligible individuals will continue to receive Food Stamps after exiting welfare due to time limits.
| State/Evaluation | Food Stamps (%) |
Medicaida (%) |
Subsidized Housing (%) |
SSI/SSDIb (%) |
|---|---|---|---|---|
| Connecticut (6) | 50 | 91 | 49 | 6 |
| Florida FTP (varies) | 74 | 62 | 38 | 16 |
| Massachusetts (10)c | 52 | 84 | 56 | 19 |
| Ohio (6)d | 73 | 91 | 69 | n/a |
| North Carolina (6)e | 71 | 85 | 52 | 26 |
| South Carolina (12)f | 87 | 93 | 35 | 10 |
| Utah (2-5) | 85 | 80 | 44 | 11 |
| Virginia (6)g | 76 | 72 | 36 | 10 |
SOURCES: Connecticut: Hunter-Manns and Bloom, 2000; Florida FTP: Bloom et al., 2000, and MDRC calculations; Massachusetts Department of Transitional Assistance, 2000; Ohio: Bania et al., 2001; North Carolina: Richardson et al., 1999; South Carolina: Richardson et al., 2001; Utah: Taylor et al., 2000; Virginia: Gordon et al., 2002. NOTES: Unless otherwise noted, all data come from
the follow-up surveys. aUnless noted, the figures show the rate of Medicaid coverage for respondents only.(back) bPercentages reflect SSI or SSDI coverage for anyone in the respondent's household.(back) cSSI includes those who received SSI, SSDI, or Social Security since leaving welfare.(back) dFood Stamp data are from administrative records and show the percentage who received Food Stamps all six months after leaving welfare.(back) eThe figure for Medicaid represents the percentage of "families" with Medicaid coverage.(back) fThe figure for Medicaid represents the percentage of respondents who reported that they or someone in their household had Medicaid coverage.(back) |
The Virginia study found that about half the respondents who were not receiving Food Stamps believed that they were not eligible; this proportion was about the same even for respondents with income below 130 percent of the poverty level, the Food Stamp income eligibility cutoff.7
Rates of Medicaid coverage are fairly high in all eight studies. Direct comparisons are difficult because some of the surveys asked about Medicaid coverage for families rather than individuals, but there appears to be some variation in coverage rates across states. Some of the variation may be related to the differences in employment rates: Respondents in states with higher employment rates may be more likely to have coverage through their employer and thus may opt not to continue Medicaid coverage.
However, this does not fully explain the variation because, for example, Utah has the lowest employment rate and one of the lowest Medicaid coverage rates. Although there may be some differences in eligibility criteria, one would expect that the vast majority of respondents in all states were eligible for coverage, either through the transitional Medicaid provision or because they met the criteria for AFDC eligibility that were in place before the 1996 welfare law passed.8 This suggests that some of the variation is likely attributable to state practices for handling cases that exit welfare due to time limits. The Utah study found that many time-limit leavers were not aware that they were eligible for any services or assistance after their cash assistance grant was closed.
A few of the studies separately measured coverage for children under Medicaid or the State Children’s Health Insurance Program (SCHIP). As expected, coverage rates for children were slightly higher than for adults.
The rates of receipt of Supplemental Security Income (SSI/SSDI) and public/subsidized housing probably reflect respondents’ situations before they left welfare (see Section II). It is unlikely that large numbers of families could have started receiving either of these forms of assistance in the relatively short period since their exit.9 The relatively high SSI rates do not necessarily mean that the respondents themselves were receiving this assistance: The surveys asked about SSI receipt for entire families. However, if someone in a respondent’s family receives SSI, the need to care for the disabled person may create a barrier to employment for the respondent.
Several of the surveys examined public assistance receipt by employment status. In most cases, nonworking respondents are substantially more likely to be receiving Food Stamps and to have someone in their household receiving SSI. Differences are smaller with regard to housing assistance and Medicaid coverage.
Table 6.3 does not include information on the receipt of child care subsidies because the surveys use different bases in reporting the percentage using subsidies (for example, some reported subsidy use among those using child care, others among working respondents with children under a certain age, and still others for the full sample). In any case, it is clear that a significant minority of respondents in some states were receiving subsidies. For example, in Vir-ginia, 39 percent of working respondents with a child under age 13 were receiving subsidies at the six-month point (most of those who were not receiving subsidies were aware of their existence). In Ohio, administrative records showed that 40 percent of those with a child under age 14 received subsidies at some point in the six months after exit. In Massachusetts, 39 percent of those with a child in care reported receiving a federal or state subsidy. Also, almost half of those with a child in care reported that they had no out-of-pocket child care costs. This could reflect subsidy receipt or free care provided by family or friends.
Household Income
The respondents’ earnings and public assistance benefits tell only part of the story with regard to their household income. As shown in Table 6.4, a substantial fraction of respondents in all the surveys were living with at least one other adult when surveyed. These other adults — typically the respondent’s spouse, partner, parent, or adult child — contributed substantial amounts of income to some of the households.
Noncustodial parents are another important source of income for some families, though child support receipt is far from universal. The surveys did not measure child support receipt uniformly, but, in most cases, between one-fourth and one-third of respondents reported that they were receiving at least some child support payments when interviewed. These payments may or may not have been received regularly.10
| State/Evalutation | Households with More Than One Adult (%) |
Average Monthly Household Income ($) |
Income Below Poverty Threshold (%) |
Monthly Income Below $1,000 (%) |
|---|---|---|---|---|
| Connecticut (6) | 43 | 1,100 | n/a | 46 |
| Florida FTP (varies) | 44 | 1,129 | n/a | 51 |
| Massachusetts (10) | 39 | n/a | n/a | n/a |
| Ohio (6) | 26 | 1,152 | 76 | n/a |
| North Carolina (6)a | 39 | 893 | n/a | 65 |
| South Carolina (12)b | 22 | 528 | n/a | 81 |
| Utah (2-5) | n/a | 750 | 72 | n/a |
| Virginia (6)c | 35 | 946 | 86 | 54 |
SOURCES: Connecticut: Hunter-Manns and Bloom, 2000; Florida FTP: Bloom et al., 2000, and MDRC calculations; Massachusetts Department of Transitional Assistance, 2000; Ohio: Bania et al., 2001; North Carolina: Richardson et al., 1999, and unpublished data; South Carolina: Richardson et al., 2001, and unpublished data; Utah: Taylor et al., 2000; Virginia: Gordon et al., 1999, and Gordon et al., 2002. NOTES: The figures in parentheses show approximately how many months after exit the interviews took place. aPercentage of households with other adults was collected at the 12-month follow-up, while income data refer to the 6-month follow-up. Income includes only the respondent's earnings and child support.(back) bRespondents reported their household income within ranges; income was calculated by assuming each respondent's income is at the midpoint of the range.(back) cPercentage of households with more than one adult and poverty rate are for cohort 1; average monthly income and income below $1,000 are for cohorts 1 and 2 combined.(back) |
The studies used a variety of approaches to measure and assess household income, so direct comparisons are probably inappropriate (even the studies that measured income in dollar terms did not necessarily do so in the same way). Table 6.4 includes several measures so that at least some data from each study can be reported. Despite the lack of uniformity, it is clear that most families in all the states had quite low household income when they were interviewed. In assessing the average household income figures, it is worth noting that the federal poverty threshold for a family of three is just under $1,200 per month.
Of course, most of these households were also living in deep poverty while they received welfare. In fact, in many cases, their household income at follow-up was higher than a nonworking family could receive from cash assistance and Food Stamps.11 Comparisons of income before and after the time limit are discussed below.
The Florida FTP, North Carolina, and Virginia studies examined household income by employment status. The Virginia study found that average monthly income six months after exit was almost twice as high for employed respondents (the gap was much smaller at the 18-month point); the results were similar in Florida FTP. The North Carolina study did not estimate dollar income but found that only 42 percent of nonworking respondents reported that their income was adequate to meet their needs; the comparable figure was 59 percent for employed respondents.
Material Hardships
As with income, it is difficult to make direct comparisons across states regarding the prevalence of different types of material hardship because the questions were phrased differently in each survey. Nevertheless, some trends are clear:
-
As might be expected given the income results, large proportions of respondents in all states reported that they were struggling financially. For example, in Connecticut, 61 percent reported that they had delayed paying bills in order to make ends meet; in South Carolina, 57 percent agreed that they were “just barely making it from day to day”; and in Virginia, 48 percent reported “money problems.”
-
Relatively few respondents reported experiencing the most serious kinds of housing distress: eviction and homelessness. Almost all the studies reported the percentage of respondents who had been homeless since leaving welfare. Although the definitions vary, all the figures are 2 percent or below. Three studies reported the percentage who had been evicted since leaving welfare:
-
Florida FTP (8 percent), Ohio (8 percent), and Utah (5 percent). Other studies found that relatively few recipients had moved to worse living arrangements since leaving welfare (in fact, respondents who had moved were more likely to have moved to better arrangements). As noted earlier, relatively large proportions of time-limit leavers are living in public or subsidized housing; it is possible that housing subsidies are protecting some families from severe housing distress.
-
Larger proportions of respondents reported food-related hardships, perhaps reflecting the way families prioritize expenditures when money runs short (for example, by skipping meals in order to pay the rent). In North Carolina, 28 percent of respondents reported that there had been occasions since leaving welfare when they could not afford to buy food. In South Carolina, 15 percent reported that they had to cut the size of meals or skip meals because there was not enough money to buy food.
-
Only a few of the studies asked whether any of the respondent’s children had been removed from their custody or had gone to live elsewhere since the family had left welfare. Although the questions were phrased differently in each survey, the percentage of positive responses was 4 percent or below in all cases.
As noted earlier, the studies that compared household income for respondents who were employed and not employed found that employed households reported much higher average income. However, the prevalence of hardships was not as closely associated with employment status. In some cases, the relationship is in the expected direction: In North Carolina, the proportions of respondents reporting that they sometimes or often did not have enough to eat were 23 percent for working respondents and 54 percent for nonworking respondents. On the other hand, in Florida FTP, employed respondents were more likely to report that they had been unable to pay their rent and were more likely to be classified as food insecure with hunger. In Vir-ginia, employed respondents were more likely to report that they had had trouble buying food and paying for housing.
One possible explanation for the unexpected pattern comes from the Florida study, which found that average monthly expenditures on housing were $288 for nonworking respondents and $415 for working respondents. Another explanation comes from the Virginia study, which found that nonworking respondents were more likely to have received assistance from community agencies or religious organizations in the past month. They were also more likely to have received money, phone access, children’s supplies, and a place to stay from family or friends during that period.
Perhaps because of the pattern discussed above, it does not appear that hardship is systematically more prevalent in the states with lower employment rates. Table 6.5 shows the responses to a set of identical questions about food access that were asked in Connecticut, North Carolina, and Utah. The proportion of respondents reporting that they sometimes or often did not have enough to eat was slightly lower in Utah, even though the employment rate was much lower there than in the other two states. Similarly, the same food security scale was administered in Florida FTP (with an employment rate of 54 percent) and Massachusetts (with an employment rate of 73 percent). The percentage of respondents classified as “food insecure with hunger” was somewhat higher in Massachusetts (24 percent versus 16 percent in Florida FTP). Obviously, factors other than employment status — perhaps family support and housing costs — are critical in explaining levels of hardship.
Before-After Comparisons
All eight of the studies assessed how respondents’ circumstances had changed since leaving welfare. As noted earlier, such changes cannot be attributed to the termination of benefits, but these data are suggestive. In almost all cases, the comparison was made by asking respondents to recall their situation while on welfare.12
Employment and Earnings
Table 6.6 shows the employment rates of respondents in their last months on welfare and at the follow-up interview. As discussed earlier, there was a modest increase in employment over time in North Carolina and Virginia but little change in the other states.13
These results suggest that the imposition of a time limit does not necessarily cause large numbers of respondents to start working — even in states like Ohio and South Carolina, where many people were not working when their cases were closed. This is consistent with the results from several random assignment studies presented in Chapter 5, which showed that employment impacts did not change substantially when families began reaching time limits.
However, a detailed analysis in the North Carolina study shows that averages can hide dynamic employment patterns. Although the employment rate was only modestly higher at the six-month follow-up point than in the last month of benefit receipt, about 60 percent of those who were not working when their case closed worked in the following six months, and more than one-half were employed six months later. Conversely, fewer than half of the respondents who were employed when they left welfare were still working in the same job six months later. About one-fourth were still working, but in a different job, and another one-fourth were not employed.
| Outcome | Connecticut | North Carolina |
Utah |
|---|---|---|---|
| Employed (%) | 80 | 63 | 43 |
| Food measures (%): | |||
| Enough and kinds of food we want to eat | 34 | 53 | 28 |
| Enough but not kinds of food we want to eat | 44 | 23 | 53 |
| Sometimes not enough food | 16 | 21 | 13 |
| Often not enough food | 6 | 3 | 6 |
| SOURCES: Connecticut: Hunter-Manns et al., 1998; North Carolina: Richardson et al., 1999; Utah: Taylor et al., 2000. |
The earnings data shown in Table 6.6 suggest that some people may have increased their work hours after leaving welfare — although the results are somewhat difficult to interpret because different people were working at the two points.14
The Florida FTP, North Carolina, and Virginia studies provide additional information because they conducted more than one round of interviews. Thus, it is possible to see how employment rates (and other outcomes) vary over a lengthy post-welfare follow-up period.
| State/Evaluation | Employed (%) | Monthly Earnings ($) | ||
|---|---|---|---|---|
| Before | After | Before | After | |
| Connecticut (6)a | 85 | 83 | 878 | 1,015 |
| Florida FTP (varies)b | 57 | 58 | 410 | 661 |
| Ohio (6) | 51 | 53 | n/a | n/a |
| North Carolina (6) | 54 | 63 | n/a | n/a |
| South Carolina (12)c | 51 | 53 | 375 | 660 |
| Virginia (6)d | 63 | 71 | 848 | 902 |
| SOURCES: Connecticut: Hunter-Manns and Bloom,
1999, and Melton and Bloom, 2000; Florida FTP: Bloom et al., 2000,
and MDRC calculations; Ohio: Bania et al., 2001; North Carolina: Richardson
et al., 1999; South Carolina: Richardson et al., 2001; Virginia: Gordon
et al., 1999.
NOTES: Unless otherwise noted, the data were collected from the follow-up surveys. The figures in parentheses show approximately how many months after exit the interviews took place. aMonthly earnings data come from administrative records in the quarter prior to exit and the second quarter after exit; quarterly earnings were divided by 3.(back) bData are from administrative records for the quarter prior to termination and the fourth quarter after termination; quarterly earnings were divided by 3. (back) cData are from administrative records for the quarter of exit and the fourth quarter after exit.(back) dData are for cohort 1 only; as a result, the employment rate at follow-up does not match the rate in Table 6.2.(back) |
In Virginia, the employment rate increased slightly in the months after case closure and then leveled off: 62 percent were employed in the month of case closure, 71 percent at the six- month follow-up point, and 69 percent at the 18-month interview. Once again, however, the underlying pattern was dynamic: 92 percent of respondents worked at some point in the 18-month period, and 63 percent experienced at least one spell of unemployment. Hourly wages among those employed were 13 percent higher at the 18-month point than at the 6-month point, and average weekly work hours were also somewhat higher (although it is important to note once again that different people were working at the two points). Interestingly, the percentage of respondents working in jobs that offered health benefits stayed largely unchanged, but the proportion who were enrolled in a company plan rose from 33 percent to 51 percent, perhaps reflecting the expiration of transitional Medicaid coverage or respondents remaining in jobs long enough to qualify for employer coverage.
The patterns are similar in North Carolina: The employment rate increased at first and then remained roughly constant (54 percent were employed in the last benefit month, 63 percent at the 6-month point, and 66 percent at the 12-month point). Wages and hours increased between the two interviews. The proportion of employed respondents working in jobs that offered health insurance did not change much, but the proportion who were enrolled in company plans nearly doubled, from 33 percent to 65 percent.
The Florida FTP evaluation studied employment patterns for a small group of leavers who were interviewed four times in the 18 months after they left welfare: Approximately one-third worked steadily throughout the period, one-third worked sporadically, and one-third did not work at all.
Income Before and After Leaving Welfare
Most of the studies asked respondents to compare their monthly income in their last month on welfare with their income in the month prior to the follow-up interview. Only two studies — Florida FTP and Virginia — examined dollar income averages at more than one point. The Florida FTP study did so by interviewing a small sample of time-limit leavers around the time their benefits were canceled and then at 6-month intervals thereafter, while the Virginia study asked respondents to recall their income during their last month on assistance and to report their income in the month prior to the 6- and 18-month interviews.15
In Virginia, average income was slightly lower 6 months after exit than during the last month on assistance. However, average income at the 18-month point was slightly higher than at either earlier point. About 45 percent of respondents reported an increase of more than 10 percent in their income between case closure and the 18-month interview. Almost as many, 41 percent, reported a decrease of 10 percent or more. The proportion of families with very low income (below $1,000 per month) decreased during this period.
The Florida study also found that average income was lower 6 months after exit than during the last month on welfare. About one-fourth of respondents had higher income at the follow-up point, while the rest had lower income. Interestingly, the Florida FTP study found that the respondents most likely to lose income were those who were working in both the last benefit month and the month before the 6-month follow-up interview.
The other studies asked respondents to compare their income or their general well-being at the two points — without trying to obtain detailed dollar amounts for the pre-exit period. Specifically:
-
In South Carolina, 47 percent of respondents agreed that they had more money than while they were on welfare; 53 percent disagreed.
-
In North Carolina, 56 percent agreed that they were a little or a lot better off one year after leaving welfare, and only 15 percent said they were a little or a lot worse off.
-
In Connecticut, a fairly high-grant state with a very generous earnings disregard, only 20 percent of respondents reported that they were more satisfied with their standard of living after leaving welfare.
-
In Massachusetts, about 40 percent reported that they had more income after leaving welfare, while an equal proportion reported that they had less to live on (about 20 percent reported that their income was about the same). About half reported that their expenses were higher, while only 8 percent reported that their expenses were lower.
-
In Utah, 49 percent reported that life in general was worse since their case closed. Thirty percent reported that life was about the same, and only 21 percent reported that life was better.
Although the patterns vary, in most states, a somewhat greater proportion of respondents believed that they were worse off financially than they had been while on welfare. It is important to recall that — particularly in states like Connecticut, Massachusetts, and Virginia — a large proportion of respondents had been mixing work and welfare before reaching the time limit. Upon reaching the limit, they lost their supplemental welfare grant, resulting in a loss of income. The fact that families saw their income decline over time, however, does not necessarily mean that they were worse off than they would have been without welfare reform (without the reforms, they would not have received the expanded disregards).
Hardships Before and After Leaving Welfare
Several of the studies asked respondents to report whether they had experienced specific hardships before and after leaving welfare. Although the patterns vary, it appears that hardships are more common in the post-exit period. This is consistent with the fact that many families had lower income. For example:
-
In North Carolina, 24 percent reported that they sometimes or often did not have enough to eat in the six months after leaving welfare, while only 8 percent reported that they had experienced this hardship in their last six months on welfare.
-
In South Carolina, respondents reported that they were more likely to get behind in paying rent or utility bills, and to have their phone or utilities disconnected, in the year after leaving welfare than in the year before leaving welfare.
-
In Massachusetts, 24 percent were classified as “food insecure with hunger” after leaving welfare, compared with 13 percent who reported that they had experienced this hardship before leaving welfare.
-
In Connecticut, 29 percent reported at follow-up that they sometimes or often relied on low-cost food to feed the children, because they were running out of money. Only 15 percent reported that this had been true in their last month on welfare.
A few of the studies asked questions about the well-being of respondents’ children before and after leaving welfare. There is no evidence that children were doing worse in the post-welfare period.
-
In Massachusetts, 29 percent reported that child-rearing was better after leaving welfare than before; 18 percent reported that child-rearing was worse.
-
In North Carolina, substantial numbers of respondents reported that their children were experiencing school-related problems, but there is no clear pattern of improvement or decay relative to the period before families left welfare.
-
In South Carolina, the proportion of respondents who reported that their children’s behavior and school performance were better than they had been one year earlier was larger than the proportion who reported that behavior and school performance had gotten worse.
-
The Utah study reported some deterioration over time in child-related outcomes, but the results are not reported separately for time-limit leavers.
Time-Limit Leavers Compared with Other Leavers
Seven of the eight studies compared outcomes for time-limit leavers with outcomes for families who left welfare for other reasons. Five of these seven studies (Connecticut, Florida FTP, Massachusetts, Ohio, and Virginia) compared time-limit leavers with “non-time-limit leavers,” a diverse group that may include families who left welfare owing to sanctions, increased income, changes in family status, or other reasons. The other two studies (South Caro-lina and Utah) divide the “other leavers” into two or more categories based on the exit reason.16 As with many of the other topics discussed earlier, each study used a somewhat different approach, making direct comparisons difficult.
Employment and Earnings
The results of the comparisons vary substantially by state, for the reasons discussed at the beginning of this chapter. For example, as shown in Table 6.7, in Connecticut, Massachu-setts, and Virginia (the three states with the highest employment rates for time-limit leavers), individuals who left because of time limits have either similar or higher employment rates than other leavers — although in all three cases, employed time-limit leavers earned less than other employed leavers. In those states, the non-time-limit leavers were a diverse group — including both people who left welfare for work and people who left for other reasons — while the time-limit leavers were mostly employed, for the reasons discussed earlier.
In the other states, time-limit leavers have both lower employment rates and lower earnings than other leavers. The South Carolina study found that sanctioned leavers had a lower post-exit employment rate than any other category of leavers (results not shown).
Receipt of Government Benefits
Table 6.8 examines the receipt of Food Stamps and Medicaid, by exit reason. All the studies found that rates of Food Stamp receipt are much higher for time-limit leavers than for other leavers. Interestingly, this is true even in states like Connecticut and Massachusetts, where the time-limit leavers have similar or higher employment rates than other leavers. As discussed in Chapter 4, one possible explanation for this pattern is that many time-limit leavers attended special exit interviews just prior to case closure; during these meetings, staff may have informed them about their eligibility for other benefits (and even recertified their benefits). In contrast, many of the non-time-limit leavers probably exited from welfare without contacting the welfare office — for example, they simply failed to show up for their next scheduled eligibility review after finding a job — which resulted in closure of their Food Stamp case. It is also possible that time-limit leavers who are predominantly long-term welfare recipients are more likely to know about the eligibility criteria for other public assistance programs.
| Leavers Employed (%) |
Monthly Earnings Among Those Employed ($) |
|||
|---|---|---|---|---|
| State/Evalutation | TL | Non-TL | TL | Non-TL |
| Connecticut (6)a | 83 | 51 | 1,015 | 1,090 |
| Florida FTP (varies) | 54 | 69 | 804 | 1,079 |
| Massachusetts (10) | 73 | 71 | 1,095 | 1,290 |
| North Carolinab | 63 | 69 | 947 | 1,190 |
| Ohio (6) | 53 | 73 | 989 | 1,276 |
| South Carolina (12) | 50 | 62 | 993 | 1,088 |
| Utah | 43 | 63 | 750 | 1,011 |
| Virginia (6)c | 71 | 63 | 902 | 1,192 |
SOURCES: Connecticut: Melton and Bloom, 2000; Florida FTP: Bloom et al., 2000, and MDRC calculations; Massachusetts Department of Transitional Assistance, 2000; Ohio: Bania et al., 2001; North Carolina: Richardson et al., 2000, and Richardson et al., 1999; South Carolina: Richardson et al., 2001, and unpublished data; Utah: Taylor et al., 2000; Virginia: Gordon et al., 1999. NOTES: Unless otherwise noted, the data were collected
from the follow-up surveys, and the non-time-limit group includes
people who left welfare at roughly the same time as the time-limit
leavers, but for reasons other than the time limit. aData are from administrative records for the second quarter after exit.(back) bEarnings for time-limit leavers are median earnings for those working for an employer.(back) |
| State/Evalutation | Food Stamps (%) | Medicaid (%) | ||
|---|---|---|---|---|
| TL | Non-TL | TL | Non-TL | |
| Connecticut (6)a | 52 | 31 | 91 | n/a |
| Florida FTP (varies) | 74 | 32 | 62 | 23 |
| Massachusetts (10) | 52 | 29 | 84 | 80 |
| North Carolina (6) | 71 | 45 | 85 | 67 |
| Ohio (6)b | 73 | 32 | 91 | 72 |
| South Carolina | 87 | 54 | 93 | 77 |
| Utah | 85 | 59 | 80 | 74 |
SOURCES: Connecticut: Melton and Bloom, 2000; Florida FTP: Bloom et al., 2000, and MDRC calculations; Massachusetts Department of Transitional Assistance, 2000; North Carolina: Richardson et al., 1999, and Richardson et al., 2000; Ohio: Bania et al., 2001; South Carolina: Richardson et al., 2001, and unpublished calculations; Utah: Taylor et al., 2000. Notes: Unless otherwise noted, the data were collected from the follow-up surveys, and the non-time-limit group includes people who left welfare at roughly the same time as the time-limit leavers, but for reasons other than the time limit. The figures in parentheses show approximately how many months after exit the interviews took place. aData are from administrative records for the second quarter after exit.(back) bFood Stamp figures are the percentage who received Food Stamps in all 6 months after exit.(back) |
Income and Material Hardships
As discussed earlier, there is little consistency in the way the studies measured income and hardship, so it is difficult to make direct comparisons. As discussed earlier, reported income is closely associated with employment status in most states, with employed respondents reporting much higher income, on average. Thus, in states where time-limit leavers have lower employment rates than other leavers, they also report lower income. A full measure of household income is not available for non-time-limit leavers in Connecticut, where time-limit leavers have a much higher employment rate. However, as expected, administrative records data suggest that time-limit leavers have substantially higher household income.
Material hardships are common among all groups of leavers, and it does not appear that time-limit leavers are systematically worse off. For example:
-
In Ohio, time-limit leavers were more likely than other leavers to have been evicted, but they were less likely to have skipped doctor visits or to have moved to a worse neighborhood since leaving welfare.
-
In South Carolina, time-limit leavers were less likely than other leavers to report that they had cut the size of meals or skipped meals since leaving welfare, but they were more likely to report having experienced some other hardships (such as going without electricity or heat or having their phone cut off).
-
In Florida FTP, time-limit leavers were slightly more likely to have experienced some hardships, but the differences were very small.
-
In Massachusetts, time-limit leavers were more likely than other leavers to report that their financial well-being had worsened since leaving welfare; the percentage classified as “food insecure with hunger” was about the same for the two groups, however.
-
In Utah, time-limit leavers were fairly consistently worse off than individuals who left because of increased income, but not necessarily worse off than people who left for “other” reasons (many of whom had been sanctioned).
In states where time-limit leavers are less likely to be employed than other leavers, they appear to be relying more heavily on both public assistance and community and family resources. However, there is little evidence that time-limit leavers are more likely to be living with other employed adults.
1Several of the surveys were conducted in selected cities or counties, rather than statewide. See Appendix C for more details.(back)
2Each study used a somewhat different approach. In general, the Massachusetts, North Carolina, Ohio, South Carolina, Utah, and Virginia studies compared time-limit leavers with other people who left welfare at around the same time for other reasons — usually before reaching a time limit. The Connecti-cut and Florida FTP studies compared program group members who reached a time limit within the follow-up period for a random assignment study with those who did not (in most cases, because they left welfare before reaching the limit).(back)
3Most of the data in Table 6.1 were collected through the follow-up surveys. In theory, all the characteristics except race could have changed since the respondents left welfare, but this seems unlikely given the relatively short post-welfare follow-up periods. The data from Connecticut and Florida FTP were collected when people entered the program being tested (that is, when they first become subject to the time limit).(back)
4North Carolina is the only state in which time-limit leavers are less likely to have three or more children. This may be because the state’s time limit was originally applied to recipients with no pre-school-age children; these recipients may have fewer children, on average, than other recipients.(back)
5In a related analysis, the Utah study used stepwise regression to predict post-closure earned income. Significant factors included past employment, clinical depression, high school education, and the presence of young children. (back)
6Also, initially (including the time period when the survey sample was drawn), Massachusetts denied extensions to all recipients who were earning above the welfare payment standard when they reached the time limit.(back)
7A family with income below 130 percent of the poverty line may still be ineligible for Food Stamps if, for example, they have too many assets to qualify.(back)
8PRWORA delinked eligibility for Medicaid from eligibility for cash assistance. States are required to provide coverage to families who meet the preexisting AFDC eligibility criteria.(back)
9In fact, the Virginia study found that the rates of both SSI receipt and public/subsidized housing receipt were very similar in the last month of benefit receipt, at the 6-month follow-up point, and at the 18-month follow-up point.(back)
10A 1998 report examined the child support status of families reaching time limits in Connecticut, Florida FTP, and Virginia, finding that only 16 percent to 29 percent had received child support in the year prior to termination. Between 47 percent and 69 percent had no support order in place when they left welfare (U.S. General Accounting Office, 1998).(back)
11Among the eight states that conducted surveys, the combined TANF/Food Stamp benefit for a family of three with no other income ranges from $544 in South Carolina to $926 in Massachusetts.(back)
12It may be
difficult for respondents to recall their situation while on welfare.
Also, their perception of their earlier status may have been colored by
more recent events.(back)
13Of course, respondents could have responded to the time limit by finding a job a few months before their benefits were terminated.(back)
14When earnings data are obtained from administrative records, it is impossible to tell whether an increase over time is due to higher wages, more hours of employment per week, or more weeks of work in a quarter. However, it seems unlikely that growth in hourly wages could have been substantial in such a short follow-up period.(back)
15The income measure in Virginia includes the respondent’s earnings plus household income from TANF, Food Stamps, child support, SSI, and unemployment insurance. It does not include earnings of other household members. Because the percentage of respondents living with another employed adult increased over time, the study authors report that the 18-month income estimate may be understated.(back)
16In preparation for this report, MAXIMUS calculated several specific outcomes for a single non-time-limit leavers group in South Carolina by combining results for those who left owing to sanctions, earned income, and for other reasons.(back)
| Table of Contents | Previous | Next |