پر کردن سوراخ در شبکه امنیت؟ مشکلات مواد و بهزیستی ذهنی در میان متقاضیان مزایای از کار افتادگی و دریافت کنندگان پس از اصلاحات رفاهی 1996

Abstract Some of the rapid recent growth in disability income receipt in the United States is attributable to single mothers post-welfare reform. Yet, we know little about how disability benefit receipt affects the economic well-being of single mother families, or how unsuccessful disability applicants fare. We compare disability recipients to unsuccessful applicants and those who never applied among current and former welfare recipients, and examine how application and receipt affect material hardships and subjective measures of well-being. We then examine whether alternative ways of making ends meet mediate differences in well-being. After controlling for alternative sources of support, no significant differences in overall actual hardships or difficulty living on current income remained between the three groups. However, even after controlling for these strategies, unsuccessful applicants were significantly more likely to report that they expected hardships in the next two months. Our results suggest a pervasive level of economic insecurity among unsuccessful applicants.

. Introduction The past 30 years have seen dramatic increases in the number of recipients of the Social Security Administration’s two disability programs. In 2009, the Supplemental Security Income (SSI) program provided means-tested income support to 4.5 million blind and disabled low-income adults who met its disability criteria, and to another 1.2 million blind and disabled children under the age of 18. These numbers represent an 82% increase in the non-elderly adult SSI caseload since 1990, and a 289% increase in the child SSI caseload. The Social Security Disability Income Program (SSDI) is a social insurance program which provides benefits to disabled workers who have a sufficient work history to qualify, independent of income and assets tests. In 2009, 7.8 million disabled workers received benefits through SSDI, up 158% since 1990.1 Some of this caseload growth, particularly with respect to SSI, stems from increased receipt by single mothers, especially since passage of the Personal Responsibility and Work Reconciliation Act (PRWORA) and the introduction of the Temporary Assistance for Needy Families Program (TANF) in 1996. Welfare reform had both direct and indirect effects on SSI participation (Schmidt, 2004). There is also evidence that SSI and SSDI are serving as a “back-up” safety net now that it is much more difficult to qualify for and continue to receive cash assistance. Wamhoff and Wiseman (2005/06) conclude that “a significant proportion of each year’s SSI awards to disabled non-elderly people [now] go to TANF recipients.” Although there is evidence that the disabled have lower economic well-being and higher rates of material hardship than the non-disabled (She and Livermore, 2007 and Parish et al., 2009), we know little about the well-being of disability benefit recipients. If disability benefits significantly reduce the incidence of material hardships, SSI and SSDI would represent an alternative safety net for disabled former cash welfare recipients who are able to navigate the complex disability determination process. We know even less about how unsuccessful applicants for disability benefits fare in terms of well-being, which is important given that less than half of these applicants are approved in their first attempt at the application process (Benitez-Silva et al., 1999). In the pre-welfare reform era, an unsuccessful applicant could remain on Aid to Families with Dependent Children (AFDC) as long as she met the program’s eligibility rules; post-1996, qualifying for cash welfare is much more difficult and TANF receipt is limited to no more than 60 months. If disabled single mothers who were previously welfare recipients have great difficulty qualifying for SSI/SSDI, this could lead to higher levels of material hardship for a vulnerable population. In this paper, we analyze data from the Women’s Employment Study, a 6½ year panel with detailed measures of health conditions and material hardship, and compare disability benefit recipients to unsuccessful applicants and to those who never applied for benefits among a sample of current and former welfare recipients. We examine how application and receipt status affect the extent of material hardships, as well as subjective measures of economic well-being. We then examine the extent to which differences in well-being among these three groups can be explained by observable differences in health conditions, as well as by unobserved heterogeneity across respondents. Finally, we analyze whether alternative ways of making ends meet mediate differences in well-being across application and receipt status. We find that while both successful and unsuccessful applicants have higher levels of overall material hardship than those who never applied, there are no significant differences in these measures between successful and unsuccessful applicants. However, unsuccessful applicants are significantly more likely to report that they expect future hardships than either of the other two groups. After controlling for alternative sources of support, no significant differences in difficulty living on current income remained between the three groups. However, even after controlling for these strategies, unsuccessful applicants were significantly more likely to report that they expected hardships in the next two months. Our results suggest a pervasive level of economic insecurity among unsuccessful applicants.

Results 5.1. Characteristics associated with disability benefit application and receipt Table 1 provides sample means for WES respondents, classified into one of three mutually-exclusive categories of disability benefit application and receipt status—(i) those who received disability benefits between 1997 and 2003; (ii) those who applied unsuccessfully for disability benefits between 1997 and 2003, and (iii) those who never applied. In general, unsuccessful applicants (column 1) are similar along many dimensions to those who received disability benefits at some time during the panel (column 2). Both groups are significantly more likely to lack a high school degree than those who never applied (column 3). Recipients are significantly more likely to have a human capital barrier (defined as having worked less than 20% of the years between age 18 and the Fall 1997 interview or having performed 4 or fewer out of 9 specific skills on a job prior to Fall 1997) than unsuccessful applicants, and the unsuccessful applicants are more likely to have a human capital barrier than those who never applied. Table 1. Demographic and socioeconomic characteristics of WES respondents, by disability benefit application/receipt status. (1) (2) (3) Applied for, but did not receive, disability benefits between 1997 and 2003 Received disability benefits between 1997 and 2003 Never applied for or received disability benefits between 1997 and 2003 African American (%) 60.5 43.2⁎ 54.6 Age in 2003 38.2 41.2⁎ 35.2⁎⁎⁎ Married in 2003 (%) 15.8 13.5 22.4 Cohabiting in 2003 (%) 13.2 13.5 17.2 Number of children in household in 2003 1.83 1.46 1.96 No high school degree (1997) (%) 38.6 43.2 25.9⁎⁎⁎ Human capital barrier (low work exp. or skills, 1997) (%) 33.6 54.1⁎⁎ 21.2⁎⁎⁎ Percent of years on welfare, from age 18 until 1997 interview (%) 59.7 64.1 58.6 Mean monthly income-to-needs in 2003 1.08 1.07 1.45⁎⁎⁎ Percent of months worked, 1997–2003 (%) 52.0 19.1⁎⁎⁎ 77.0⁎⁎⁎ Percent working in 2003 (%) 39.5 16.2⁎⁎⁎ 79.2⁎⁎⁎ Average hours worked, 2003 14.3 4.5⁎⁎⁎ 28.3⁎⁎⁎ Number of observations 114 37 379 Notes: Significance stars in Columns 2 and 3 indicate whether that value is statistically different from the value in Column 1. For all variables where the means are statistically different between the never applied (Column 3) and the unsuccessful applicants (Column 1), the means are also statistically different between the never applied and the successful recipients (Column 2). For all variables where the means are not statistically different between Columns 1 and 3, they are also not statistically different between Columns 2 and 3, with the exception of Number of Children in the Household. ⁎ Significant at 10%. ⁎⁎ Significant at 5%. ⁎⁎⁎ Significant at 1%. Table options Recipients are also significantly less likely to be African American (43.2%) than the unsuccessful applicants (60.5%). This is consistent with findings from earlier research on racial differences in acceptance rates (General Accounting Office, 1992 and Lahiri et al., 1995). Recipients are the least likely to be working at the time of the 2003 interview (16.2%), but this is clearly endogenous to benefit receipt. The employment rate of unsuccessful applicants (39.5%) is higher than that of the recipients, but is half the rate for those who never applied (79.2%). Table 2 provides information on the health and mental health conditions of the three groups of respondents. Recipients and unsuccessful applicants are again more similar to each other than to mothers who never applied. About 2/3 of recipients (column 2) and half of unsuccessful applicants (column 1) report a physical health problem that limits activities, compared to only 16.1% of those who never applied (column 3).29 Similar differentials exist for specific health problems that limit activities – arthritis, rheumatism or bone problems, back or neck problem, breathing problem, heart problem, hypertension or high blood pressure, and diabetes. Both recipients and unsuccessful applicants have a significantly higher likelihood for each of the health problems than those who never applied. Recipients report a higher number of days in the last month in which they were unable to work due to health problems (17.7) than do unsuccessful applicants (9.77), although this is likely to be endogenous to disability benefit receipt. Table 2. Health and mental health conditions of WES respondents, by disability benefit application/receipt status. (1) (1) (3) Applied for, but did not receive, disability benefits between 1997 and 2003 Received disability benefits between 1997 and 2003 Never applied for or received disability benefits between 1997 and 2003 Physical health problem in 2003 (%) 51.3 67.6∗ 16.1∗∗∗ Health problem limits activities (2003) Arthritis, rheumatism, or bone problems (%) 43.9 43.2 10.6∗∗∗ Back or neck problem (%) 43.4 50.0 14.1∗∗∗ Breathing problem (%) 28.1 35.1 8.5∗∗∗ Heart problem (%) 17.7 8.1 1.1∗∗∗ Hypertension or high blood pressure (%) 16.7 24.3 1.9∗∗∗ Diabetes or sugar (%) 5.3 5.4 2.1∗ Mean number of health problems (of 6) 1.56 1.64 0.38∗∗∗ Days unable to work due to health problems 9.77 17.7∗∗∗ 0.74∗∗∗ Health limits activity a lot (2003) Vigorous activities (%) 51.8 64.9 13.5∗∗∗ Moderate activities (%) 21.9 24.3 3.4∗∗∗ Carrying groceries (%) 21.9 40.5∗∗ 3.7∗∗∗ Climbing one flight of stairs (%) 20.2 29.7 2.6∗∗∗ Bending or stooping (%) 27.2 43.2∗ 8.2∗∗∗ Walking several blocks (%) 31.6 51.4∗∗ 7.1∗∗∗ Any mental health problem in 2003 (%) 52.6 51.4 29.3∗∗∗ Depression in 2003 (%) 35.1 40.5 15.6∗∗∗ Generalized anxiety disorder in 2003 (%) 23.7 29.7 11.4∗∗∗ Post-traumatic stress disorder in 2003 (%) 29.5 10.8∗∗ 13.3∗∗∗ Social phobia in 2003 (%) 12.4 24.3∗ 5.6∗∗ Hard drug use in 2003 (%) 7.9 5.4 1.8∗∗∗ Child with a health problem in 2003 (%) 21.9 18.9 13.2∗∗ Notes: See notes for Table 1. For all variables where the means are statistically different between the never applied (Column 3) and the unsuccessful applicants (Column 1), the means are also statistically different between the never applied and the successful recipients (Column 2), with the exception of the following: Diabetes/Sugar, Post-Traumatic Stress Disorder, Hard drug use, Child health problem. For all variables where the means are not statistically different between Columns 1 and 3, they are also not statistically different between Columns 2 and 3. Table options Similar patterns are found for limitations in carrying out activities of daily living. Both successful and unsuccessful applicants are significantly more likely than those who never applied to have limitations in carrying groceries, climbing one flight of stairs, bending or stooping, and walking several blocks. The differences between the unsuccessful and successful applicants for these variables are smaller and less statistically significant. Recipients and unsuccessful applicants have similar rates of both mental health conditions and hard drug use (although unsuccessful applicants are more likely to report post-traumatic stress disorder and successful applicants are more likely to report social phobia), and both groups have more problems than those who never applied for disability benefits. In addition, both groups of applicants are significantly more likely to have a child with a health problem than mothers who never applied. For most health and mental health conditions, there are no statistically significant differences between unsuccessful applicants and recipients (columns 1 versus 2). However, the differences in the prevalence of health and mental conditions between the unsuccessful applicants and respondents who never applied for benefits are statistically significant and large in magnitude (columns 1 versus 3). 5.2. Disability benefit application and receipt and well-being We next examine differences in material hardships and subjective well-being among the three groups of respondents at the 2003 interview. The top panel of Table 3 presents means of 11 indicators of material hardship experiences, by application and receipt status over the sample period. Mothers who never applied for disability benefits between 1997 and 2003 are better off in 2003 along several dimensions than both the recipients and the unsuccessful applicants. Recipients (column 2) were significantly more likely to have experienced food insufficiency (40.5%) than unsuccessful applicants (25.4%); but, unsuccessful applicants were significantly more likely to have experienced food insufficiency than those who never applied (13.7%). Respondents who never applied for benefits were significantly less likely to have had their telephone cut off (24.8%) than either the recipients (43.2%) or the unsuccessful (47.4%) applicants. For medical hardships, the recipients were significantly less likely to report having no health insurance (0.0%), than either the unsuccessful applicants (17.5%) or those who never applied (21.1%). Even though none of the recipients reported not having health insurance, 21.6% reported not receiving needed medical or dental care. The health insurance provided through SSI and SSDI appears to reduce, but not eliminate, medical hardships. Table 3. Material hardships and subjective well-being among WES respondents in 2003, by disability benefit application/receipt status. (1) (2) (3) Applied for, but did not receive, disability benefits between 1997 and 2003 Received disability benefits between 1997 and 2003 Never applied for or received disability benefits between 1997 and 2003 Food insufficiency (%) 25.4 40.5⁎ 13.7⁎⁎⁎ No health insurance (mother) (%) 17.5 0.0⁎⁎⁎ 21.1 Mother did not receive needed medical or dental care (%) 34.2 21.6 34.0 No health insurance (child)a (%) 2.1 3.4 4.2 Child did not receive needed medical carea (%) 4.3 3.4 3.3 5 or more housing upkeep problems (%) 7.9 8.1 3.7⁎ Utilities cut off (%) 14.9 8.1 10.8 Eviction (%) 7.0 13.5 6.3 Homeless (%) 7.9 2.7 5.5 Telephone cut off (%) 47.4 43.2 24.8⁎⁎⁎ Moved in with others to share expenses (%) 22.8 21.6 17.7 All hardships Any hardship (%) 72.8 86.5⁎ 62.3⁎⁎ Mean number of hardships 1.90 1.65 1.42⁎⁎⁎ Subjective well-being measures Difficulty living on current income (%) 46.5 48.6 24.5⁎⁎⁎ Expect to experience hardships in next 2 months (%) 30.7 24.3 9.8⁎⁎⁎ Notes: Significance stars in Columns 2 and 3 indicate whether that value is statistically different from the value in Column 1. For all variables where the means are statistically different between the never applied (Column 3) and the unsuccessful applicants (Column 1), the means are also statistically different between the never applied and the successful recipients (Column 2), with the exception of 5 or more housing problems and the mean number of hardships. For all variables where the means are not statistically different between Columns 1 and 3, they are also not statistically different between Columns 2 and 3, with the exception of no health insurance for the mother. a These two questions were asked only to respondents with children under age 18 in the household. All other hardship questions were asked to all respondents. ⁎ Significant at 10%. ⁎⁎ Significant at 5%. ⁎⁎⁎ Significant at 1%. Table options Recipients were significantly more likely to report any material hardship (86.5%) than those who never applied (72.8%). The mean number of hardships did not statistically differ between the unsuccessful applicants (1.90) and the recipients (1.65), but the mean for unsuccessful applicants was significantly higher than for those who never applied (1.42). Note, however, that although the applicants were significantly more likely to report any hardship than those who never applied, 62.3% of the nonapplicants experienced at least one hardship. This suggests a high level of disadvantage among current and former welfare recipients, even among the nonapplicants. In the bottom panel of Table 3, almost half of applicants reported that they had difficulty living on their current income – 48.6% of recipients, and 46.5% of unsuccessful applicants, as compared to 24.5% of those who never applied. When asked if they expected to experience hardships in the next two months, unsuccessful applicants were over 6 percentage points more likely to report “pretty much” or “a great deal” than disability recipients (30.7.% versus 24.3%), although these means are not statistically different from each other. Both types of applicants were significantly more likely to report that they expected to experience hardships than those who never applied (9.8%). Table 4 and Table 5 present results from multivariate regressions that control for varying respondent attributes and examine the extent to which disability benefit receipt and application status affect material hardships and subjective well-being. In Table 4, the dependent variables in Columns 1–3 reflect reports of experiencing any material hardship. Columns 4–6 examine whether the respondent reported having difficulty living on her current household income, while Columns 7–9 analyze the subjective expectation of having hardship in the next two months. For each dependent variable, the first column presents results for regressions that do not include observable health conditions. The second column adds controls for the presence of a mental health problem, the number of maternal health problems, the four daily limitation variables, and an indicator for hard drug use. The third column presents results from individual fixed-effects regressions, and as a result control for any unobservable differences across respondents that are time-invariant. Table 4. Effects of disability benefit application/receipt status on material hardships and subjective well-being in Wave 5. Ever hardship Difficulty living on current income Expect hardship in next 2 months (1) (2) (3) (4) (5) (6) (7) (8) (9) Received disability benefits 0.226⁎⁎⁎ (0.083) 0.150⁎ (0.089) 0.128 (0.102) 0.188⁎⁎ (0.080) 0.090 (0.084) −0.064 (0.091) 0.092 (0.062) 0.019 (0.065) 0.093 (0.071) Applied for disability benefits unsuccessfully 0.105⁎⁎ (0.051) 0.036 (0.056) −0.048 (0.053) 0.180⁎⁎⁎ (0.049) 0.075 (0.053) 0.074 (0.048) 0.187⁎⁎⁎ (0.038) 0.131⁎⁎⁎ (0.041) 0.072⁎ (0.037) Child health problem −0.030 (0.057) −0.055 (0.058) 0.014 (0.035) 0.021 (0.055) −0.020 (0.055) 0.026 (0.031) 0.075⁎ (0.042) 0.058 (0.042) 0.040⁎ (0.024) Married/living with husband −0.154⁎⁎⁎ (0.054) −0.131⁎⁎ (0.054) −0.106⁎⁎(0.050) −0.136⁎⁎⁎ (0.052) −0.108⁎⁎ (0.051) −0.215⁎⁎⁎ (0.045) −0.039 (0.040) −0.034 (0.040) −0.063⁎ (0.035) Cohabiting as unmarried partner −0.070 (0.058) −0.062 (0.058) −0.002 (0.036) −0.140⁎⁎ (0.056) −0.119⁎⁎ (0.054) −0.184⁎⁎⁎ (0.032) −0.024 (0.043) −0.024 (0.042) −0.030 (0.025) # of children in household −0.005 (0.016) −0.004 (0.016) 0.001 (0.016) −0.022 (0.015) −0.014 (0.016) −0.018 (0.015) −0.022⁎ (0.012) −0.021⁎ (0.012) −0.013 (0.011) Less than high school education 0.106⁎⁎ (0.046) 0.115⁎⁎ (0.046) – 0.107⁎⁎ (0.045) 0.124⁎⁎⁎ (0.044) – 0.057 (0.035) 0.067⁎⁎ (0.034) – Low work experience or skills −0.049 (0.049) −0.048 (0.049) – −0.059 (0.047) −0.060 (0.046) – −0.019 (0.036) −0.008 (0.036) – African American −0.164⁎⁎⁎ (0.043) −0.152⁎⁎⁎ (0.043) – −0.018 (0.041) −0.004 (0.040) – −0.060⁎ (0.032) −0.051 (0.031) – Respondent age in 1997 0.0003 (0.0222) −0.0038 (0.0221) – −0.004 (0.021) −0.012 (0.021) – 0.041⁎⁎ (0.017) 0.034⁎⁎ (0.016) – Age-squared −0.0001 (0.0003) −0.0000 (0.0003) – 0.0001 (0.0003) 0.0002 (0.0003) – −0.0005⁎⁎ (0.0002) −0.0004⁎ (0.0002) – Mental health problem – 0.196⁎⁎⁎ (0.045) 0.070⁎⁎ (0.027) – 0.206⁎⁎⁎ (0.042) 0.079⁎⁎⁎ (0.024) – 0.156⁎⁎⁎ (0.033) 0.086⁎⁎⁎ (0.019) Number of maternal health problems – 0.005 (0.023) 0.010 (0.016) – 0.028 (0.022) 0.042⁎⁎⁎ (0.014) – 0.011 (0.017) 0.018∗ (0.011) Limitation – carrying groceries – 0.060 (0.086) 0.040 (0.046) – 0.113 (0.082) 0.123⁎⁎⁎ (0.041) – 0.022 (0.063) −0.010 (0.032) Limitation – climbing one flight of stairs – 0.082 (0.090) −0.019 (0.056) – 0.082 (0.085) −0.010 (0.050) – −0.032 (0.066) 0.023 (0.039) Limitation – bending – 0.020 (0.072) 0.031 (0.042) – 0.027 (0.069) −0.012 (0.038) – 0.097⁎ (0.053) 0.003 (0.030) Limitation – walking several blocks – −0.046 (0.074) −0.054 (0.045) – −0.110 (0.070) −0.013 (0.040) – −0.084 (0.054) 0.003 (0.031) Hard drug use – 0.039 (0.115) 0.039 (0.069) – 0.215⁎⁎ (0.109) 0.099 (0.062) – 0.034 (0.084) −0.026 (0.048) Fixed effects N N Y N N Y N N Y χ2 test of Receive = Apply 1.85 1.61 3.61⁎ 0.01 0.03 2.76⁎ 2.05 2.87⁎ 0.10 ⁎ Significant at 10%. ⁎⁎ Significant at 5%. ⁎⁎⁎ Significant at 1%. Table options Table 5. Effects of disability benefit application/receipt status on individual hardships in Wave 5. Food insufficiency No health insurance (mother) Mother did not receive needed medical care (1) (2) (3) (4) (5) (6) (7) (8) (9) Received disability benefits 0.255⁎⁎⁎ (0.068) 0.216⁎⁎⁎ (0.074) 0.238⁎⁎⁎ (0.078) −0.265⁎⁎⁎ (0.068) −0.245⁎⁎⁎ (0.073) −0.096 (0.084) −0.160∗ (0.083) −0.248⁎⁎⁎ (0.088) −0.116 (0.088) Applied for disability benefits unsuccessfully 0.103⁎⁎ (0.042) 0.064 (0.047) 0.004 (0.041) −0.046 (0.042) −0.008 (0.047) −0.054 (0.044) 0.004 (0.051) −0.085 (0.056) −0.062 (0.046) Child health problem 0.104⁎⁎ (0.046) 0.087⁎ (0.048) 0.037 (0.026) −0.066 (0.046) −0.040 (0.048) −0.032 (0.028) −0.040 (0.056) −0.082 (0.057) −0.014 (0.030) Married/living with husband −0.059 (0.044) −0.047 (0.044) −0.022 (0.038) −0.014 (0.044) −0.016 (0.044) −0.009 (0.041) 0.003 (0.054) 0.025 (0.054) −0.053 (0.044) Cohabiting as unmarried partner 0.001 (0.047) 0.000 (0.045) −0.045 (0.027) −0.011 (0.047) −0.007 (0.047) −0.017 (0.029) −0.052 (0.058) −0.029 (0.057) 0.024 (0.031) # of children in household 0.002 (0.013) 0.006 (0.014) 0.002 (0.012) −0.053⁎⁎⁎ (0.013) −0.060⁎⁎⁎ (0.014) −0.047⁎⁎⁎ (0.013) −0.060⁎⁎⁎ (0.016) −0.048⁎⁎⁎ (0.016) −0.022 (0.014) Less than high school education 0.040 (0.038) 0.048 (0.038) – 0.061 (0.038) 0.068∗ (0.038) – 0.062 (0.046) 0.067 (0.046) – Low work experience or skills 0.014 (0.040) 0.021 (0.040) – 0.055 (0.040) 0.050 (0.040) – 0.023 (0.049) 0.008 (0.048) – African American 0.012 (0.035) 0.013 (0.035) – −0.110⁎⁎⁎ (0.035) −0.108⁎⁎⁎ (0.035) – −0.155⁎⁎⁎ (0.043) −0.146⁎⁎⁎ (0.042) – Respondent age in 1997 0.027 (0.018) 0.026 (0.018) – 0.009 (0.018) 0.010 (0.018) – 0.003 (0.022) −0.006 (0.022) – Age-squared −0.0003 (0.0002) −0.0003 (0.0002) – −0.0001 (0.0002) −0.0001 (0.0002) – −0.0001 (0.0003) 0.000 (0.0003) – Mental health problem – 0.119⁎⁎⁎ (0.037) 0.080⁎⁎⁎ (0.021) – 0.024 (0.037) 0.005 (0.022) – 0.135⁎⁎⁎ (0.044) 0.026 (0.024) Number of maternal health problems – 0.017 (0.019) −0.008 (0.012) – −0.047⁎⁎ (0.019) −0.026⁎⁎ (0.013) – 0.035 (0.023) 0.013 (0.014) Limitation – carrying groceries – −0.053 (0.071) 0.022 (0.035) – 0.086 (0.071) 0.037 (0.037) – −0.046 (0.086) −0.042 (0.040) Limitation – climbing one flight of stairs – 0.043 (0.074) −0.001 (0.043) – −0.027 (0.074) 0.047 (0.046) – 0.047 (0.089) −0.002 (0.049) Limitation – bending – −0.057 (0.060) −0.007 (0.032) – 0.038 (0.060) 0.032 (0.035) – 0.172⁎⁎ (0.072) 0.074⁎⁎ (0.037) Limitation – walking several blocks – −0.008 (0.061) −0.044 (0.034) – −0.019 (0.061) −0.021 (0.037) – −0.034 (0.073) −0.015 (0.039) Hard drug use – 0.064 (0.095) −0.028 (0.052) – 0.026 (0.095) 0.079 (0.056) – 0.095 (0.114) 0.118⁎⁎ (0.060) Fixed effects N N Y N N Y N N Y χ2 test of Receive = Apply 4.38⁎⁎ 4.18⁎⁎ 10.98⁎⁎⁎ 9.15⁎⁎⁎ 10.16⁎⁎⁎ 0.30 3.46⁎ 3.33∗ 0.46 ⁎ Significant at 10%. ⁎⁎ Significant at 5%. ⁎⁎⁎ Significant at 1%. Table options The results in Column 1 of Table 4 indicate that recipients were 22.6 percentage points more likely to have reported at least one material hardship than those who never applied. This is twice the magnitude of the coefficient for unsuccessful applicants, who are 10.5 percentage points more likely to have reported any material hardship than those who never applied. These coefficients are each statistically significant at the 5 percent-level or higher. However, as evidenced by the F-statistic from the Wald test reported at the bottom of the table, they are not statistically different from each other. Results in Column 4 show that both types of applicants are significantly more likely to report having difficulty living on current income than those who never applied (18.8 percentage points for recipients and 18.0 points for unsuccessful applicants). When we analyze expectations of experiencing hardships in the next two months (Column 7), unsuccessful applicants are 18.7 percentage points more likely to expect future hardships than those who never applied, and are almost 10 percentage points (18.7 less 9.2) more likely to expect future hardships than disability recipients. Both of these differences are statistically significant. The coefficients on the demographic variables are generally in the expected direction and are sometimes statistically different from zero. Those with less than a high school education are significantly more likely to report any hardship and more likely to report difficulty living on current income, while those who are married and living with their husband are significantly less likely to report any hardship. The estimated coefficients for African Americans in all regressions in Table 4 are negative, and are statistically significant in many, suggesting that African American mothers report significantly fewer hardships than white mothers, all else constant. Where statistically significant, these results are robust to inclusion of variables for a number of different categories of household income, and to controls for detailed health characteristics and are consistent with work by Sullivan et al. (2008) using the same data set. In Columns 2, 5, and 7 of Table 4, we control for health conditions to test whether differences in well-being across application/receipt status can be explained by observable differences in health status between the three groups. In all cases, the inclusion of observable health status reduces the magnitude of the application and receipt variables relative to the baseline. In most cases, the coefficients are no longer statistically significant, suggesting that most of the differences across the three groups in terms of material hardships and subjective well-being can be explained by group differences in health status. It is worth noting that mental health problems are positively and significantly associated with all dependent variables. The individual fixed-effects analyses (Columns 3, 6, and 9) yield similar results. In most cases, the coefficients on received benefits and applied unsuccessfully are no longer statistically different from zero or from each other. The one exception in both the cross-sectional and fixed effects regressions is the subjective measure of expected hardships in the next two months. Inclusion of health conditions (Column 8) reduces the magnitude of the coefficients on both application and receipt status. However, unsuccessful applicants are still 13.1 percentage points significantly more likely to expect hardships than either recipients or those who never applied. Results from the fixed-effects regressions in Column 9 also show significantly higher expectations of future hardship among unsuccessful applicants. In Table 5, we examine three individual hardships separately – food insufficiency, the absence of maternal health insurance, and an indicator for whether the mother reported not receiving needed medical or dental care. We have also estimated regressions using the other eight material hardships as dependent variables. For most, there are no statistically significant effects – disability applicants (whether successful or not) do not seem to differ from those who never apply. 30 Column 1 shows that both recipients and unsuccessful applicants are more likely than nonapplicants to report food insufficiency. However, the coefficient is much larger for recipients (a 25.5 percentage point increase relative to those who never applied) than for unsuccessful applicants (a 10.3 percentage point increase). This result is robust to the inclusion of observable health conditions (Column 2) and individual fixed-effects (Column 3), and is surprising, as both successful and unsuccessful applicants were similar along health, demographic, and socioeconomic characteristics, but recipients have an additional source of income.31 One possibility is that these results are driven by unmeasured differences in disability severity between these two groups that remain after we control for observable differences in health status or time-invariant, unobservable differences. Columns 4 and 7 show that disability benefit receipt significantly reduces the probability of maternal medical hardships (having no health insurance and not receiving needed medical care) relative to unsuccessful applicants and those who never applied. This is not surprising, given that recipients are entitled to government-provided health insurance and that many nonapplicants work for firms that do not subsidize health insurance. These results hold when we control for observable health conditions (Columns 5 and 8), but become insignificant in the fixed effects models (Columns 6 and 9). In Table 6, we examine the extent to which mothers’ use of alternative sources of support to make ends meet mediates differences in well-being by application status. We focus on subjective measures of well-being and individual hardship measures. For each measure, the first column provides results from cross-sectional regressions that control for several alternative sources of support – the number of adults in the household; receipt of other income sources (including SNAP, TANF, child support, unemployment insurance and/or workers’ compensation, respondent earnings, earnings of other individuals in the household, transfers from friends or family, and all other sources of income)32; whether any child in the household received SSI; and whether the respondent reported receiving any help from charity. These regressions do not include measures of health status. Even though a great deal of the differences in well-being can be explained by differences in health conditions, from a disability policy perspective it is the absolute differences in well-being that should matter, not the differences conditional on differences in health status. The second column for each outcome presents results that include individual fixed effects. These regressions exclude the child SSI variable, as it is only available in Wave 5. Table 6. How do alternative sources of support mediate differences in well-being? Difficulty living on current income Expect hardships Food insufficiency No health insurance (Mother) Mother did not receive needed medical care Received disability benefits 0.102 (0.093) −0.048 (0.091) 0.030 (0.070) 0.092 (0.071) 0.225⁎⁎⁎ (0.080) 0.240⁎⁎⁎ (0.077) −0.127⁎ (0.076) −0.093 (0.081) −0.212⁎⁎ (0.097) −0.108 (0.088) Applied for disability unsuccessfully 0.077 (0.057) 0.073 (0.047) 0.113⁎⁎⁎ (0.043) 0.078⁎⁎ (0.037) 0.038 (0.049) −0.008 (0.040) −0.012 (0.047) −0.023 (0.042) −0.071 (0.059) −0.041 (0.045) Other adults in household 0.029 (0.053) 0.044 (0.031) −0.004 (0.040) 0.033 (0.024) −0.021 (0.045) 0.015 (0.026) −0.006 (0.043) −0.002 (0.027) −0.032 (0.055) 0.005 (0.030) Received food stamps 0.073 (0.048) 0.012 (0.026) 0.022 (0.037) −0.019 (0.021) 0.008 (0.041) −0.017 (0.023) −0.126⁎⁎⁎ (0.040) −0.136⁎⁎⁎ (0.024) −0.034 (0.051) −0.090⁎⁎⁎ (0.026) Received TANF 0.094 (0.058) 0.047⁎ (0.027) 0.100⁎⁎ (0.044) −0.032 (0.021) 0.006 (0.050) −0.028 (0.023) −0.146⁎⁎⁎ (0.048) −0.140⁎⁎⁎ (0.024) −0.080 (0.061) −0.072⁎⁎⁎ (0.026) Received child support 0.018 (0.043) −0.043 (0.028) −0.029 (0.033) −0.032 (0.022) −0.031 (0.037) 0.003 (0.024) −0.040 (0.036) −0.030 (0.025) −0.040 (0.045) −0.042 (0.027) Received UI/workers comp 0.169⁎ (0.098) −0.029 (0.058) 0.090 (0.075) −0.027 (0.045) 0.021 (0.084) 0.014 (0.049) 0.062 (0.081) 0.001 (0.052) 0.284⁎⁎⁎ (0.103) 0.013 (0.056) Respondent had earnings −0.029 (0.052) −0.070⁎⁎⁎ (0.027) 0.071⁎ (0.039) −0.046⁎⁎ (0.021) 0.081⁎ (0.044) 0.004 (0.023) −0.023 (0.043) 0.003 (0.024) −0.039 (0.054) −0.027 (0.026) Others in hh had earnings 0.065 (0.053) −0.027 (0.028) 0.042 (0.041) −0.009 (0.022) 0.001 (0.046) 0.013 (0.024) 0.024 (0.044) −0.016 (0.025) 0.052 (0.056) −0.019 (0.027) Received transfers from friends and family 0.252⁎⁎⁎ (0.053) 0.101⁎⁎⁎ (0.029) 0.127⁎⁎⁎ (0.041) 0.052⁎⁎ (0.022) 0.051 (0.046) 0.085⁎⁎⁎ (0.024) 0.033 (0.044) 0.016 (0.026) 0.149⁎⁎⁎ (0.056) 0.077⁎⁎⁎ (0.028) Received other income −0.090 (0.124) −0.070 (0.057) −0.033 (0.095) −0.063 (0.045) −0.058 (0.107) 0.021 (0.049) −0.095 (0.102) −0.052 (0.051) −0.100 (0.130) −0.057 (0.055) Child receives SSI −0.142⁎⁎ (0.060) – −0.057 (0.045) – −0.026 (0.051) – 0.062 (0.049) – 0.035 (0.062) – Charity receipt 0.091⁎ (0.048) 0.091⁎⁎⁎ (0.026) 0.125⁎⁎⁎ (0.036) 0.070⁎⁎⁎ (0.020) 0.285⁎⁎⁎ (0.041) 0.075⁎⁎⁎ (0.022) 0.077⁎ (0.039) 0.009 (0.023) 0.160⁎⁎⁎ (0.050) 0.053⁎⁎ (0.026) Fixed effects N Y N Y N Y N Y N Y χ2 test of Receive = Apply 0.07 2.14 1.41 0.05 5.61⁎⁎ 12.37⁎⁎⁎ 2.28 0.89 2.11 0.69 Notes: Regressions include all demographic and socioeconomic variables from Tables 4 and 5, but do not control for physical or health conditions or hard drug use. ⁎ Significant at 10%. ⁎⁎ Significant at 5%. ⁎⁎⁎ Significant at 1%. Table options A number of these potential alternative sources of support have no significant effects on well-being, including the number of adults in the household, child support income, earnings of others in the household, and receipt of other income. Household receipt of child SSI reduces the likelihood of difficulty living on current income, and is consistent with Duggan and Kearney (2007). Receipt of unemployment insurance income by a household member is associated with an increased probability of reporting difficulty living on current income and not receiving medical care, but those results go away in the fixed-effects versions of the regressions, suggesting that they reflect unobservable differences across respondents. Respondent earnings have a negative and significant effect on subjective measures of well-being in the fixed effects regressions. SNAP and TANF income are negatively associated with the probability of medical hardships, but this could be due to interactions between Medicaid and other transfer programs. The largest and most consistent effects are found for charity receipt and receipt of transfers from friends and family. In all regressions except for the outcome of no health insurance, these variables have positive and significant coefficients. These positive coefficients are likely to reflect the fact that individuals who are in the most difficult situations tend to be the ones that turn to those sources. As such, these coefficients are likely to measure the effect of experiencing severe hardships, and not the possibility that these factors relieve hardship. This is similar to findings by Parish et al. (2009), who include a measure of charity, whether a respondent received emergency food, as one of nine hardship outcomes. When all of these variables are included, our main results from Table 4 and Table 5 hold. Neither successful nor unsuccessful applicants differ from nonapplicants in terms of subjective difficulty living on current income. Recipients are still more likely to report food insufficiency and less likely to report medical hardships. However, unsuccessful applicants are still statistically more likely to expect future hardships than either of the other two groups.