Table of Contents | Previous | Next |
Chapter Nine
OTHER MEASURES OF WELL-BEING
9.1. BACKGROUND
In seeking to understand the full impact of welfare reform, policymakers are interested in broader measures of well-being beyond the outcomes typically considered, such as welfare use, employment and earnings, or income and poverty. For example, income may not fully capture a family’s command over resources if it has savings available to draw on during periods of low income, if it is able to borrow money from family or friends, or if it can incur debt to pay for unexpected costs. In addition, in the transition from welfare to work, there may be an increase in work-related expenses (e.g., child care, transportation, and clothing) that will affect the ability to consume other goods and services but that will not be reflected in income. For this reason, consumption or expenditures are often considered a better gauge of a family’s well-being, because this measure reflects the value of what a family actually consumes in total or for specific categories of purchases such as food. The importance of food consumption for overall nutrition and health status has led to the development of specific measures of food insecurity to capture problems with having enough money to buy food or experiencing periods when there is not enough food or meals are skipped. Likewise, access to health care is considered by some to be a key measure of well-being. Health care coverage for adult and child family members, whether from public or private sources, is one indicator of whether a family can afford visits to medical professionals for preventative care or to treat acute or chronic conditions. Housing conditions and the quality of the neighborhood are other indicators of the circumstances under which families live.
These various indicators of well-being may be affected by welfare reform through either direct or indirect mechanisms. Some features of state waivers or state TANF plans, for example, affect health care coverage directly by providing transitional medical assistance. In other cases, leaving welfare may mean a loss in public health insurance coverage (either due to changes in eligibility or administrative problems in obtaining coverage when eligible) that may not be replaced by employment-based health insurance. The influences may also be more indirect, with changes in welfare use and employment leading to changes in family income that, in turn, affect decisions about consumption and savings or about residential location. For example, if incomes rise or income flows become more stable as a result of welfare reform, we would expect food insecurity to be less of a problem. Moreover, having more income may allow a family to move to a better quality home or safer neighborhood, to put money in a savings account, or to purchase or maintain an automobile. Then again, if incomes fall or remain the same on average but with more fluctuation, there may be an increase in food or housing insecurity, and assets may decline or a family may incur new debt.
Recent descriptive analyses provide some perspective on how former welfare recipients and the low-income population more generally are faring, as captured by various measures of well-being. A recent summary of 15 state "leaver" studies funded by USDHHS reveals that former welfare recipients are at risk of various forms of material hardship (USDHHS, 2001a). For example, 7—45 percent of adult leavers have no health insurance. The comparable range for children in households of leavers is 8—33 percent. Various measures of food insecurity indicate that one-fifth to one-half of former recipients experience problems with having enough money to buy food, running out of food, skipping meals, and other forms of food insecurity and hunger.75 The prevalence of forms of housing insecurity and medical hardships was somewhat lower. In five of the state studies, 13—30 percent of single-parent leavers reported they were worse off financially overall after leaving welfare, while 46—68 percent reported they were better off.
While the collection of leaver studies is informative, these studies are not designed to capture the causal impact of welfare reform as a whole, or of specific policy components in particular, on other measures of well-being. Like the preceding chapters, the remainder of this chapter is devoted to a summary of the causal studies from the experimental and econometric literatures. In particular, we focus on the following domains of well-being: material hardship and food insecurity, health insurance coverage, housing hardships and neighborhood quality, and asset ownership. For these four outcome domains, all the evidence derives from the experimental studies summarized in Tables 3.5 and 3.6; we are not aware of any econometric studies that use the DoD methodology to assess the causal impact of waivers or TANF as a bundle or specific policy reforms on these measures of well-being.76,77,78
All the experimental studies we review in this chapter base their impact analysis on survey-based measures of well-being; administrative data simply cannot fully capture these broader measures of a family’s circumstances.79 For example, material hardship is typically captured by recipients’ perceptions of financial strain, or experiences with specific problems affecting their housing conditions (e.g., a leaky roof or ceiling), neighborhood (e.g., crime, assault, or burglaries), material needs (e.g., could not pay rent or mortgage), and food security (e.g., did not have enough food to eat or used a food bank).
The reliance on survey data raises a few methodological concerns. First, in many cases, the sample sizes available for analysis are smaller than those shown in Table 3.5. Survey samples in demonstration studies are often smaller than the overall study population by design, and survey nonresponse further reduces the sample of respondents. Smaller samples will reduce the statistical power of the study for detecting small- to moderate-sized effects, as well as differences for subgroups.
Second, compared with the outcomes reviewed in prior chapters, a larger number of the experimental studies do not include any broader measures of well-being in their impact analyses (or their analyses to date). Of the studies we consider, CWPDP, SSP Plus, SSP Applicants, IMPACT, TSMF, FIP, VIP/VIEW, AWWDP, FDP, PPI, PIP, and ABC do not assess measures in the domains we list above. For those studies that do focus on other measures of well-being, many include only a few of the various measures that could be collected. For example, the inclusion of a measure of health insurance coverage for adults and children is quite common, while only a handful of the studies collect data on one or more asset measures, such as a savings account or car ownership. Because all measures are not available for all studies, it is more difficult to draw solid conclusions about the impact of the welfare reform policy or policies being evaluated in each demonstration study.
Third, many of these measures of well-being can be conceptualized and measured in a number of different ways, and there is no assurance of uniformity across studies in the measures actually used. Health insurance coverage is at one end of the spectrum, with most demonstration studies measuring whether the respondents or their children have any form of public or private health care coverage at the time of follow-up. The amount of savings or whether the respondent owns a vehicle is also measured in a similar way across the few studies that include such indicators. At the other extreme, food insecurity is measured in a different way in almost every study that includes one or more measures.80 In some cases, a specific question is asked (Did you use a food bank in the last three months? Has your family had enough to eat in the last month?); other studies report a multi-item scale of food insecurity, such as the one developed by the U.S. Department of Agriculture (USDA).81 As the questions above suggest, for some measures, a positive impact would indicate a favorable outcome, whereas for other measures, a negative impact would indicate a favorable outcome. All these factors make comparisons across studies within and between the classifications we have defined more problematic. We will revisit these issues when we synthesize the findings across the various studies at the end of this chapter.
We turn to the experimental literature in the next section, summarizing the findings from the studies that include other measures of well-being. The results from the available studies are synthesized in the third section. The final section concludes the chapter. The limited amount of information regarding subgroup differences in the outcomes covered in this chapter is summarized in Appendix A.
9.2. RANDOM ASSIGNMENT STUDIES OF THE EFFECT OF WELFARE REFORM ON OTHER MEASURES OF WELL-BEING
Tables 9.1, 9.2, 9.3, and 9.4 summarize the results for the random assignment studies in terms of the following measures of well-being: material hardship and food insecurity (9.1); health insurance coverage (9.2); residential moves, housing hardships, and neighborhood quality (9.3); and assets (savings and vehicle ownership) (9.4). In each case, the tables report the specific measure or measures available for the population served, the control group mean, and the impact estimate and its statistical significance.
In this section, we organize our discussion by the four outcome domains. In the section that follows, we synthesize these results by the reform policy or policies being evaluated, using the classification scheme outlined in Chapter 3 (see Table 3.5), considering all the well-being measures within each class of studies. For our six-way classification scheme, the studies in the fifth group (shown in Panel E) do not include any of the measures of well-being covered in this chapter.
9.2.1. Material Hardship and Food Insecurity
As seen in Table 9.1, measures of material hardship and food insecurity are available for at least one program in Panels A through D and F, with follow-up periods that range from 18 months to four years. Of the two programs that focus on financial work incentives (Panel A), only MFIP-IO provides measures of material hardship. Neither of these two measures are statistically significant with the exception of a significant favorable impact on the number of material hardships in the last year as of the three-year follow-up for MFIP applicants. With one exception, MFIP-IO also had no statistically significant impact on two measures of food insecurity. Again, for applicants in the study, MFIP-IO had a statistically significant favorable impact on whether the family had enough to eat in the last month, a measure of food insecurity.82 WRP-IO had no statistically significant effects on the two measures of food insecurity included in the 42-month follow-up survey. Although most of the effects for WRP-IO and MFIP-IO are statistically insignificant, with one exception (a measure of meals skipped by children for MFIP-IO long-term recipients), they are all in the favorable direction.
| Material hardship | Food insecurity | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Name | Cases served | Data | Measure | Control mean | Impact | % | Measure | Control mean | Impact | % |
| A. Programs that focus on financial work incentives | ||||||||||
| WRP-IO | Single-parent recipients and applicants | S | Sometimes or often not enough food in past 12 mos at 42-mo FU (%) | 22.9 | -2.6 | -11.4% | ||||
| S | Often true in last 12 mos that food bought didn't last and didn't have money to get more at 42-mo FU (%) | 18.0 | -1.9 | -10.6% | ||||||
| MFIP-IO | Urban single parents recipients | S | Perception of financial strain at 36-mo FU (range of 1=least strain to 4=most strain) | 3.0 | -0.1 | -3.3% | Family had enough to eat in last month at 36-mo FU (for those with child aged 5 to 12 at FU) (%) | 80.1 | 4.8 | 6.0% |
| S | Number of material hardships (7 items) during past 12 mos at 36-mo FU | 1.5 | -0.1 | -6.7% | In last month, any children skip a meal because not enough money for food at 36-mo FU (for those with child aged 5 to 12 at FU) (%) | 3.9 | 1.1 | 28.2% | ||
| Urban single parents applicants | S | Perception of financial strain at 36-mo FU (range of 1=least strain to 4=most strain) | 2.8 | 0.0 | -1.4% | Family had enough to eat in last month at 36-mo FU (for those with child aged 5 to 12 at FU) (%) | 85.6 | 7.5** | 8.8% | |
| S | Number of material hardships (7 items) during past 12 mos at 36-mo FU | 1.5 | -0.3* | -16.6% | In last month, any children skip a meal because not enough money for food at 36-mo FU (for those with child aged 5 to 12 at FU) (%) | 4.1 | -2.2 | -53.7% | ||
| B. Programs that focus on financial work incentives tied to hours of work | ||||||||||
| New Hope | Poor families employed FT at RA | S | Number of material hardships at 2-year FU | 1.9 | 0.0 | -1.6% | Insufficient food in last month at 2-year FU (%) | 7.7 | 1.2 | 15.6% |
| Poor families not employed FT at RA | S | Number of material hardships at 2-year FU | 2.4 | -0.3*** | -12.5% | Insufficient food in last month at 2-year FU (%) | 13.8 | -1.6 | -11.6% | |
| SSP | Single-parent recipients | S | Used food bank in last 3 months at 18-mo FU (%) | 21.1 | -2.0* | -9.5% | ||||
| S | Used food bank in last 3 months at 36-mo FU (%) | 18.8 | -1.0 | -5.3% | ||||||
| S | Could not get groceries at 36-mo FU (%) | 34.4 | -4.2*** | -12.2% | ||||||
| C. Programs that focus on mandatory work-related activities | ||||||||||
| LA Jobs-1st GAIN | Single-parent recipients and applicants | S | Experienced food insecurity (2 or more problems) at yr 2 (%) | 48.6 | 4.5 | 9.3% | ||||
| S | Experienced food insecurity with hunger (5 or more problems) at yr 2 (%) | 13.3 | 5.5** | 41.4% | ||||||
| D. Programs that focus on financial work incentives and mandatory work-related activities | ||||||||||
| WRP | Single-parent recipients and applicants | S | Sometimes or often not enough food in past 12 mos at 42-mo FU (%) | 22.9 | -1.2 | -5.2% | ||||
| S | Often true in last 12 mos that food bought didn't last and didn't have money to get more at 42-mo FU (%) | 18.0 | -1.5 | -8.3% | ||||||
| MFIP | Urban single-parent recipients | S | Perception of financial strain at 36-mo FU (range of 1=least strain to 4=most strain) | 3.0 | -0.1* | -3.3% | Family had enough to eat in last month at 36-mo FU (for those with child aged 5 to 12 at FU) (%) | 80.1 | -0.3 | -0.4% |
| S | Number of material hardships (7 items) during past 12 mos at 36-mo FU | 1.5 | 0.1 | 6.7% | In last month, any children skip a meal because not enough money for food at 42-mo FU (for those with child aged 5 to 12 at FU) (%) | 3.9 | 2.0 | 51.3% | ||
| Urban single-parent applicants | S | Perception of financial strain at 36-mo FU (range of 1=least strain to 4=most strain) | 2.8 | -0.1* | -3.2% | Family had enough to eat in last month at 36-mo FU (for those with child aged 5 to 12 at FU) (%) | 85.6 | 4.5 | 5.3% | |
| S | Number of material hardships (7 items) during past 12 mos at 36-mo FU | 1.5 | -0.2* | -10.6% | In last month, any children skip a meal because not enough money for food at 42-mo FU (for those with child aged 5 to 12 at FU) (%) | 4.1 | 0.2 | 4.9% | ||
| E. Programs that focus on other individual reforms | ||||||||||
| F. Programs that focus on TANF-like bundle of reforms (time limits with financial incentives, work-related activities, or both) | ||||||||||
| EMPOWER (a) | Recipients | S | Used food bank or soup kitchen since RA as of 30-mo FU (%) | 30.9 | -1.2 | -3.9% | ||||
| FTP | Recipients and applicants | S | Four or more material hardships at 4-year FU (%) | 19.9 | -1.7 | -8.5% | Food secure (USDA 6-item scale) at 4-year FU (%) | 64.2 | 1.8 | 2.7% |
| S | Three or more "severe" hardships at 4-year FU (%) | 14.1 | -5.3*** | -37.6% | Food insecure (USDA 6-item scale) at 4-year FU (%) | 18.8 | -0.5 | -2.7% | ||
| S | Usually has enough money at end of mo. at 4-year FU (%) | 63.0 | 6.0*** | 9.5% | Food insecure with hunger (USDA 6-item scale) at 4-year FU (%) | 17.0 | -1.3 | -7.4% | ||
| Jobs First | Recipients and applicants | S | Four or more material hardships at 3-year FU (%) | 16.9 | -0.8 | -5.0% | Food secure (USDA 6-item scale) at 3-year FU (%) | 59.8 | 1.5 | 2.5% |
| S | Three or more "severe" hardships at 3-year FU (%) | 11.8 | 1.1 | 9.2% | Food insecure (USDA 6-item scale) at 3-year FU (%) | 18.3 | -1.2 | -6.7% | ||
| S | Food insecure with hunger (USDA 6-item scale) at 3-year FU (%) | 21.8 | -0.3 | -1.1% | ||||||
| NOTES: For full program names and citations, see Table
3.4. Abbreviations: S=survey data; FU=follow-up. * = statistically significant at the 10 percent level; ** = statistically significant at the 5 percent level; *** = statistically significant at the 1 percent level. (a) Phoenix site only, cash assistance. |
Both New Hope and SSP (Panel B), in their evaluations of financial work incentives tied to hours of work, consider food insecurity, while New Hope also includes a measure of material hardship. For New Hope, the only significant impact is a favorable effect on the number of material hardships at the two-year follow-up interview for poor families not employed full-time at random assignment. For the food insecurity measures, SSP shows significant favorable effects for two of the three measures assessed at the 18- and 26-month follow-up interviews.
Among the evaluations of programs that focus on mandatory work-related activities (Panel C), L.A. Jobs-First GAIN is the only one to include measures in this domain. That study shows an unfavorable effect as of the two-year follow-up on the multi-item measures of food insecurity and food insecurity with hunger, with a statistically significant impact on the latter outcome. In this case, 13.3 percent of the single-parent recipient control group report five or more problems associated with food insecurity and hunger, compared with 18.8 percent for the treatment group, a 41 percent difference.
For the studies that focus on financial work incentives and mandatory work-related activities (Panel D), only WRP and MFIP assess measures in these domains. In terms of material hardship, MFIP produces significant favorable effects on at least one measure of financial strain and/or the number of material hardships for both recipients and applicants. None of the food insecurity measures for WRP or MFIP are statistically significant, and the signs are mixed.
Of the studies that focus on TANF-like bundles of reforms (Panel F), both FTP and Jobs First consider material hardship and food insecurity using similar measures, while EMPOWER includes only a measure of food insecurity. In the case of FTP, the three hardship measures show favorable effects as of the four-year follow-up, two of them statistically significant. In particular, the fraction with three or more "severe" hardships was 8.8 percent in the treatment group compared with 14.1 percent for the control group, mostly because of the reduction in housing and neighborhood problems (discussed below).83 Sixty-nine percent of the treatment group reported at the four-year follow-up that they usually had enough money at the end of the month compared with 63 percent of the control group. The two hardship measures reported for Jobs First as of the three-year follow-up are not statistically significant and mixed in sign. There is no impact of FTP or Jobs First on food insecurity, which is measured three ways.84 The impact on the use of food banks and soup kitchens is also insignificant in the EMPOWER evaluation.
9.2.2. Health Insurance Coverage
Measures of health insurance coverage for adults and children, reported in Table 9.2, are included in a larger number of studies, with a follow-up period as long as five years. Of the programs reporting impacts in this domain, only WRP and Jobs First provide transitional health benefits that exceed what is available to control group members. Lower rates of coverage for treatment group members may indicate a loss of eligibility for public programs (e.g., Medicaid) that is not replaced by private programs (e.g., employer-provided coverage), or it may result from administrative problems obtaining or maintaining coverage for public programs for which the individual is eligible.
| Recipient health insurance coverage | Children health insurance coverage | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Name | Cases served | Data | Measure | Control mean | Impact | % | Measure | Control mean | Impact | % |
| A. Programs that focus on financial work incentives | ||||||||||
| WRP-IOMFIP-IO | Single-parent recipients and applicants | S | Respondent covered by any HI at 42-mo FU (%) | 81.6 | 0.1 | 0.1% | Child covered by any HI at 42-mo FU (%) | 84.2 | -1.5 | -1.8% |
| Urban single parents recipients | S | Respondent has HCC at 36-mo FU (%) | 83.9 | 1.6 | 1.9% | |||||
| S | Respondent had HCC continuously for 36 mos. after RA (%) | 61.3 | 13.6*** | 22.2% | Children had HCC continuously for 36 mos after RA (for those with child aged 5 to 12 at FU) (%) | 67.0 | 11.7*** | 17.5% | ||
| Urban single parents applicants | S | Respondent has HCC at 36-mo FU (%) | 73.9 | 5.0 | 6.8% | |||||
| S | Respondent had HCC continuously for 36 mos. after RA (%) | 50.0 | 17.9*** | 35.8% | Children had HCC continuously for 36 mos after RA (for those with child aged 5 to 12 at FU) (%) | 62.7 | 13.3** | 21.2% | ||
| B. Programs that focus on financial work incentives tied to hours of work | ||||||||||
| New Hope | Poor families employed FT at RA | S | Any periods without HI at 2-year FU (%) | 55.2 | -8.5 | -15.4% | ||||
| Poor families not employed FT at RA | S | Any periods without HI at 2-year FU (%) | 60.5 | -11.3*** | -18.7% | |||||
| C. Programs that focus on mandatory work-related activities | ||||||||||
| LA Jobs-1st GAINAtlanta LFA | Single-parent recipients and applicants | S | Respondent has HCC at end of 2-year FU (%) | 93.4 | -1.3 | -1.4% | Children have HCC at end of 2-year FU (%) | 92.9 | -0.3 | -0.3% |
| Recipients and applicants | S | Respondent has HCC at end of 2-year FU (%) | 86.0 | -2.4 | -2.8% | All dependent children have HCC at end of 2-year FU (%) | 85.6 | 0.5 | 0.6% | |
| S | Respondent has HCC at end of 5-year FU (%) | 72.4 | -1.4 | -1.9% | All dependent children have HCC at end of 5-year FU (%) | 84.5 | 0.6 | 0.7% | ||
| Grand Rapids LFA | Recipients and applicants | S | Respondent has HCC at end of 2-year FU (%) | 86.0 | -3.3 | -3.8% | All dependent children have HCC at end of 2-year FU (%) | 85.7 | -1.4 | -1.6% |
| S | Respondent has HCC at end of 5-year FU (%) | 77.7 | -2.6 | -3.3% | All dependent children have HCC at end of 5-year FU (%) | 81.8 | -3.0 | -3.7% | ||
| Riverside LFA | Recipients and applicants | S | Respondent has HCC at end of 2-year FU (%) | 87.3 | -1.8 | -2.1% | All dependent children have HCC at end of 2-year FU (%) | 88.4 | -3.3** | -3.7% |
| S | Respondent has HCC at end of 5-year FU (%) | 80.3 | -2.0 | -2.5% | All dependent children have HCC at end of 5-year FU (%) | 83.2 | -1.3 | -1.6% | ||
| Portland | Recipients and applicants; no cases with substantial barriers | S | Respondent has HCC at end of 2-year FU (%) | 90.4 | -3.3 | -3.7% | All dependent children have HCC at end of 2-year FU (%) | 88.6 | -4.8 | -5.4% |
| S | Respondent has HCC at end of 5-year FU (%) | 80.6 | -6.0 | -7.4% | All dependent children have HCC at end of 5-year FU (%) | 80.2 | -4.7 | -5.9% | ||
| Atlanta HCD | Recipients and applicants | S | Respondent has HCC at end of 2-year FU (%) | 86.0 | -2.2 | -2.6% | All dependent children have HCC at end of 2-year FU (%) | 85.6 | -0.8 | -0.9% |
| S | Respondent has HCC at end of 5-year FU (%) | 72.4 | 1.6 | 2.2% | All dependent children have HCC at end of 5-year FU (%) | 84.5 | -1.1 | -1.3% | ||
| Grand Rapids HCD | Recipients and applicants | S | Respondent has HCC at end of 2-year FU (%) | 86.0 | -1.7 | -2.0% | All dependent children have HCC at end of 2-year FU (%) | 85.7 | 0.5 | 0.6% |
| S | Respondent has HCC at end of 5-year FU (%) | 77.7 | 0.1 | 0.1% | All dependent children have HCC at end of 5-year FU (%) | 81.8 | -2.5 | -3.1% | ||
| Riverside HCD | Recipients and applicants | S | Respondent has HCC at end of 2-year FU (%) | 87.5 | -0.8 | -0.9% | All dependent children have HCC at end of 2-year FU (%) | 88.8 | -0.7 | -0.8% |
| S | Respondent has HCC at end of 5-year FU (%) | 80.0 | 0.3 | 0.4% | All dependent children have HCC at end of 5-year FU (%) | 82.1 | 3.2 | 3.9% | ||
| Columbus Integrated | Recipients and applicants | S | Respondent has HCC at end of 2-year FU (%) | 85.0 | -5.2* | -6.1% | All dependent children have HCC at end of 2-year FU (%) | 86.3 | -6.3** | -7.3% |
| Columbus Traditional | Recipients and applicants | S | Respondent has HCC at end of 2-year FU (%) | 85.0 | 0.8 | 0.9% | All dependent children have HCC at end of 2-year FU (%) | 86.3 | 0.2 | 0.2% |
| Detroit | Recipients and applicants | S | Respondent has HCC at end of 2-year FU (%) | 92.0 | -0.9 | -1.0% | All dependent children have HCC at end of 2-year FU (%) | 90.9 | -0.6 | -0.7% |
| Oklahoma City | Applicants | S | Respondent has HCC at end of 2-year FU (%) | 70.9 | -3.3 | -4.7% | All dependent children have HCC at end of 2-year FU (%) | 72.5 | -9.0** | -12.4% |
| D. Programs that focus on financial work incentives and mandatory work-related activities | ||||||||||
| WRPMFIP | Single-parent recipients and applicants | S | Respondent covered by any HI at 42-mo FU (%) | 81.6 | -2.3 | -2.8% | Child covered by any HI at 42-mo FU (%) | 84.2 | -4.2* | -5.0% |
| Urban single-parent recipients | S | Respondent has HCC at 36-mo FU (%) | 83.9 | 1.6 | 1.9% | |||||
| S | Respondent had HCC continuously for 36 mos. after RA (%) | 61.3 | 7.9** | 12.9% | Children had HCC continuously for 36 mos after RA (for those with child aged 5 to 12 at FU) (%) | 67.0 | 8.5** | 12.7% | ||
| Urban single-parent applicants | S | Respondent has HCC at 36-mo FU (%) | 73.9 | 4.4 | 6.0% | |||||
| S | Respondent had HCC continuously for 36 mos. after RA (%) | 50.0 | 12.9*** | 25.8% | Children had HCC continuously for 36 mos after RA (for those with child aged 5 to 12 at FU) (%) | 62.7 | 7.2* | 11.5% | ||
| E. Programs that focus on other individual reforms | ||||||||||
| F. Programs that focus on TANF-like bundle of reforms (time limits with financial incentives, work-related activities, or both) | ||||||||||
| FTP | Recipients and applicants | S | Respondent has no HI at 4-year FU (%) | 38.4 | 0.9 | 2.3% | Children have no HI at 4-year FU (%) | 15.7 | 1.2 | 7.6% |
| Jobs First | Recipients and applicants | S | Respondent has no HI at 3-year FU (%) | 18.4 | -4.4*** | -24.2% | Children have no HI at 3-year FU (%) | 4.6 | -0.7 | -14.5% |
| NOTES: For full program names and citations, see
Table 3.4. Abbreviations: S=survey data; FU=follow-up; RA=random assignment.
* = statistically significant at the 10 percent level; ** = statistically significant at the 5 percent level; *** = statistically significant at the 1 percent level. |
In terms of programs that focus on financial work incentives (Panel A), WRP-IO had a very small and statistically insignificant impact on health insurance coverage as of the 42-month follow-up for the respondent and the child. This is despite the fact that WRP-IO provided three years of transitional Medicaid benefits for recipients leaving welfare for work in contrast to only one year for AFDC recipients in the control group. However, since Vermont has an array of health insurance programs available for low-income families, recipients in the control group were able to obtain coverage from other sources, so there is little treatment-control difference in coverage rates for adults or children.
MFIP-IO also shows no significant impact on coverage at the time of the 36-month follow-up for the respondent. (A comparable measure is not available for children.) In contrast, the fraction of adults and children with continuous health care coverage during the entire 36-month period following random assignment is higher by 12—18 percentage points, differences that are statistically significant. This effect is consistent with the higher rates of welfare usage, which automatically qualified the family for Medicaid coverage among MFIP-IO participants at the time MFIP was in place, as discussed in Chapter 4.
New Hope is the only study in Panel B (programs that focus on financial work incentives tied to hours of work) that reports impacts on health insurance coverage. Poor families not employed full-time at the time of random assignment into New Hope were less likely by 11 percentage points to experience any periods without health insurance in the two-year interval since random assignment compared with the control group. The impact estimate for the group employed full-time at random assignment is nearly as large but not statistically significant. New Hope provided subsidized health insurance, which is attributed with reducing gaps in coverage. Even so, a sizeable fraction (47—49 percent) of the treatment group experienced one or more gaps in health insurance coverage over two years.
Among the programs that focus on work mandates (Panel C), measures of health care coverage for adults and children are available for L.A. Jobs-First GAIN two years after randomization and all 11 NEWWS programs up to five years after randomization. None of these programs provided transitional Medicaid coverage for the treatment group that differed from the comparison group. Almost all the point estimates are negative, indicating that programs that require mandatory work activities tend to reduce the probability of health insurance coverage after two years for both adults and children. However, all the effects are very small and only 4 of the 38 estimates are statistically significant. For the 7 NEWWS programs with five-year follow-up impacts, there is little change between years two and five. The reductions in health insurance coverage are consistent with the move off welfare to employment associated with mandatory work requirements. Medicaid coverage received while on welfare is not entirely made up by transitional Medicaid coverage, coverage under the poverty-related Medicaid expansions, or transitions to employment-based coverage.
Among the programs that combine financial work incentives and mandatory work-related activities (Panel D), MFIP has a favorable effect on health insurance coverage, while the reverse is true for WRP, the program with less generous financial work incentives. For example, urban single-parent applicants in MFIP were 13 percentage points more likely to have had continuous health care coverage in the three years since random assignment compared with the control group. As noted earlier, the favorable MFIP impact is probably the result of the increase in welfare use, which automatically qualified the family for Medicaid coverage at the time MFIP was in place. Even though WRP is the only program of this group to offer transitional Medicaid benefits, the fact that it did not result in a more favorable impact is attributable to the availability of public insurance coverage through other programs in the state.
For the programs that evaluate TANF-like bundles of reforms, health insurance coverage is reported only for FTP and Jobs First (Panel F). The former program shows statistically insignificant effects as of the four-year follow-up. In contrast, Jobs First raises health insurance coverage rates as of the three-year follow-up for children and adults respectively, with an effect that is statistically significant only for the adult recipient. The difference may be due to the fact that Jobs First provided two years of transitional Medicaid coverage compared with only one year for the control group. FTP did not offer any additional transitional Medicaid benefits for program participants compared with the controls.
9.2.3. Residential Moves, Housing Hardships, and Neighborhood Quality
Table 9.3 reports impact estimates for residential moves, housing hardships, and neighborhood quality using various measures. Among the studies that focus on financial work incentives, only MFIP-IO reports results for measures of residential moves and neighborhood quality, with each reported only for the sample with a child age 5—12 at the 36-month follow-up. For both recipients and applicants, MFIP-IO reduces the number of moves that take place following random assignment for families with primary-school-age children, with an effect for the applicant group that is just under one-half a move and statistically significant. Fewer moves may be indicative of reduced housing instability. Alternatively, it may indicate a diminished ability to upgrade housing or neighborhood quality. On this point, reported neighborhood safety does not appear to be affected much for either the recipients or applicants, although the point estimates are opposite in sign for the two groups. When financial work incentives are combined with work mandates (Panel D), MFIP shows no statistically significant impacts for either residential moves or neighborhood quality.
| Residential moves | Housing hardships | Neighborhood quality | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Name | Cases served | Data | Measure | Control mean | Impact | % | Measure | Control mean | Impact | % | Measure | Control mean | Impact | % |
| A. Programs that focus on financial work incentives | ||||||||||||||
| MFIP-IO | Urban single parents recipients | S | Number of moves since RA at 36-mo FU (for those with child aged 5 to 12 at FU) | 1.7 | -0.1 | -5.9% | Live in a safe neighborhood at 36-mo FU (for those with child aged 5 to 12 at FU) (%) | 74.0 | 2.5 | 3.4% | ||||
| Urban single parents applicants | S | Number of moves since RA at 36-mo FU (for those with child aged 5 to 12 at FU) | 1.6 | -0.4** | -25.0% | Live in a safe neighborhood at 36-mo FU (for those with child aged 5 to 12 at FU) (%) | 83.1 | -2.0 | -2.4% | |||||
| B. Programs that focus on financial work incentives tied to hours of work | ||||||||||||||
| New Hope | Poor families employed FT at RA | S | Living in an overcrowded dwelling at 2-year FU (%) | 16.7 | -4.3 | -25.7% | ||||||||
| S | Having had utilities cut off at 2-year FU (%) | 34.3 | 1.3 | 3.8% | ||||||||||
| S | Experiencing one or more housing defects at 2-year FU (%) | 37.1 | 6.4 | 17.3% | ||||||||||
| Poor families not employed FT at RA | S | Living in an overcrowded dwelling at 2-year FU (%) | 15.2 | -1.4 | -9.2% | |||||||||
| S | Having had utilities cut off at 2-year FU (%) | 43.0 | -1.1 | -2.6% | ||||||||||
| S | Experiencing one or more housing defects at 2-year FU (%) | 49.7 | -3.7 | -7.4% | ||||||||||
| SSP | Single-parent recipients | S | Any residential moves since RA at 36-mo FU (for those with child aged 3 to 5 at FU) (%) | 75.0 | 4.4 | 5.8% | Structural problems in house at 36-mo FU (%) | 12.3 | -2.0** | -16.3% | Good neighborhood quality at 36-mo FU (for those with child aged 3 to 5 at FU) (%) | 76.4 | 0.3 | 0.4% |
| S | Any residential moves since RA at 36-mo FU (for those with child aged 6 to 11 at FU) (%) | 63.4 | 4.5** | 7.2% | Things not working properly in house at 36-mo FU (%) | 12.7 | -1.2 | -9.4% | Good neighborhood quality at 36-mo FU (for those with child aged 6 to 11 at FU) (%) | 75.3 | 0.3 | 0.4% | ||
| S | Any residential moves since RA at 36-mo FU (for those with child aged 12 to 18 at FU) (%) | 51.1 | 2.9 | 5.8% | Good neighborhood quality at 36-mo FU (for those with child aged 12 to 18 at FU) (%) | 78.6 | -5.9** | -7.6% | ||||||
| C. Programs that focus on mandatory work-related activities | ||||||||||||||
| LA Jobs-1st GAIN | Single-parent recipients and applicants | S | Neighborhood is bad place to raise children at yr 2 (%) | 32.1 | -4.4 | -13.7% | ||||||||
| S | Neighborhood is unsafe for children to play outside at yr 2 (%) | 27.0 | -3.1 | -11.5% | ||||||||||
| Atlanta LFA | S | Any residential moves since RA at 5-yr FU (%) | 66.2 | 0.8 | 1.2% | |||||||||
| Grand Rapids LFA | S | Any residential moves since RA at 5-yr FU (%) | 78.0 | 7.6*** | 9.7% | |||||||||
| Riverside LFA | S | Any residential moves since RA at 5-yr FU (%) | 84.0 | 2.4 | 2.9% | |||||||||
| Portland | S | Any residential moves since RA at 5-yr FU (%) | 85.7 | -0.9 | -1.1% | |||||||||
| Atlanta HCD | S | Any residential moves since RA at 5-yr FU (%) | 66.2 | 1.5 | 2.3% | |||||||||
| Grand Rapids HCD | S | Any residential moves since RA at 5-yr FU (%) | 78.0 | 5.4** | 6.9% | |||||||||
| Riverside HCD | S | Any residential moves since RA at 5-yr FU (%) | 81.8 | 0.0 | 0.0% | |||||||||
| D. Programs that focus on financial work incentives and mandatory work-related activities | ||||||||||||||
| MFIP | Urban single-parent recipients | S | Number of moves since RA at 36-mo FU (for those with child aged 5 to 12 at FU) | 1.7 | 0.2 | 11.8% | Live in a safe neighborhood at 36-mo FU (for those with child aged 5 to 12 at FU) (%) | 74.0 | -0.6 | -0.8% | ||||
| Urban single-parent applicants | S | Number of moves since RA at 36-mo FU (for those with child aged 5 to 12 at FU) | 1.6 | 0.1 | 6.3% | Live in a safe neighborhood at 36-mo FU (for those with child aged 5 to 12 at FU) (%) | 83.1 | 0.1 | 0.1% | |||||
| E. Programs that focus on other individual reforms | ||||||||||||||
| F. Programs that focus on TANF-like bundle of reforms (time limits with financial incentives, work-related activities, or both) | ||||||||||||||
| EMPOWER (a) | Recipients | S | Used emergency shelter since RA as of 30-mo FU (%) | 1.0 | -0.1 | -10.0% | ||||||||
| FTP | Recipients and applicants | S | Any residential moves since RA at 4-year FU (%) | 69.6 | 2.9 | 4.2% | 2 or more housing problems at 4-year FU (%) | 18.4 | -4.3** | -23.4% | 4 or more neighbor-hood problems at 4-year FU (%) | 21.0 | -3.8* | -18.1% |
| S | Crowding (more than 1 person per room) at 4-year FU (%) | 13.8 | 0.7 | 5.3% | ||||||||||
| Jobs First | Recipients and applicants | S | Any residential moves since RA at 3-year FU (%) | 65.4 | -0.1 | -0.2% | 2 or more housing problems at 3-year FU (%) | 18.1 | -0.4 | -2.1% | 1 ore more neighbor-hood problems at 3-year FU (%) | 70.6 | -6.1*** | -8.6% |
| S | Number of moves since RA at 3-year FU | 1.4 | 0.0 | -3.4% | Ever homeless and living on street in last year at 3-year FU (%) | 1.5 | 1.1* | 76.2% | ||||||
| S | Lived in homeless, emergency or DV shelter in last year at 3-year FU (%) | 3.2 | -0.4 | -13.5% | ||||||||||
| Residential moves | Housing hardships | Neighborhood quality | ||||||||||||
| NOTES: For full program names and citations, see
Table 3.4. Abbreviations: S=survey data; FU=follow-up; DV=domestic
violence. * = statistically significant at the 10 percent level; ** = statistically significant at the 5 percent level; *** = statistically significant at the 1 percent level. (a) Phoenix site only, cash assistance. |
Of the studies classified under Panel B (those that focus on financial work incentives tied to hours of work), impacts on residential moves, housing hardships, and neighborhood quality are reported for New Hope and SSP. For SSP, the number of moves is higher for the treatment group, but the impact is statistically significant only for SSP families with a child age 6—11 at follow-up. Again, whether the larger number of moves is favorable or unfavorable depends upon the motivation for the change in residence. Measures of housing hardships are reported in SSP, with favorable effects for two different measures (one of which is statistically significant). On the other hand, neighborhood quality as measured by SSP is no different for families with children in the first two age categories (3—5 and 6—11 at follow-up), and it is negatively affected for families with the oldest children (age 12 —18 at follow-up). Thus, it would appear that the higher number of moves associated with SSP are not leading to improved neighborhood conditions (and they may actually be worse for families with teenagers), but housing quality may be somewhat better. Finally, New Hope considers only housing hardships and finds impacts that are statistically insignificant and mixed in sign.
Measures in this domain are available for a few of the studies that focus on mandatory work-related activities (Panel C). In the case of L.A. Jobs-First GAIN, there is no statistically significant difference in neighborhood quality after two years. Impact estimates on the number of residential moves as of the five-year follow-up are reported for 7 of the NEWWS programs. With one exception, the impact estimates are all positive–indicating a higher fraction of the treatment group made any move since random assignment–and two of the impacts are statistically significant. Again, it is unclear whether this increased mobility is desirable or not.
A few measures of residential moves, housing hardships, and neighborhood quality are reported in EMPOWER, FTP, and Jobs First (Panel F), three studies that focus on TANF-like bundles of reforms. Four years after random assignment, FTP reduces the incidence of two or more housing problems and four or more neighborhood problems, with effects that are significant at the 5 and 10 percent level, respectively. There is no statistically significant effect of FTP on crowding (defined as more than one person per room) or on the prevalence of making one or more residential moves. Jobs First also lowers the number of neighborhood problems after three years, but it results in a statistically significant increase in the incidence of homelessness in the last year. Use of an emergency shelter is rare, even for the EMPOWER control group (about 1 percent). The treatment group is lower by one-tenth of one percent, but the difference is not significant.
9.2.4. Assets
Table 9.4 records various measures of the level and distribution of financial assets, and ownership of physical property such as a house or automobile. These measures are reported for an even smaller number of studies. WRP-IO is the only program that focuses on financial work incentives with outcomes in this domain. One feature of WRP-IO was that the asset limit (specifically the vehicle value) that determines welfare eligibility was increased, although it is difficult to attribute changes in outcomes to this particular program features in itself. While the effect on the average level of savings for more than three years after random assignment is small and insignificant, the fraction with savings over $500 increases by more than half, from 9.2 to 14.4 percent, an effect that is significant at the 5 percent level. The impact on the fraction owning a vehicle is not statistically significant. When financial work incentives are combined with work mandates as for WRP (Panel D), the impact estimates for these measures have the same sign but are smaller, and hence none are statistically significant.85