Table of Contents | Previous | Next |
Effects on Employment and Public Assistance Receipt
This section presents the effects of New York City’s Personal Roads to Individual Development and Employment (PRIDE) program on employment, employment retention, cash assistance receipt, and food stamp receipt in the first two years after random assignment. Administrative records data are used to compare outcomes for the PRIDE and control groups among single parents and for Safety Net recipients without dependent children. The tables and figures present effects on summary measures. Effects on the full set of outcomes are shown in Appendix B.
Effects for Single Parents
-
On average, the PRIDE program increased employment and employment stability among single parents. The effects persisted throughout the two-year follow-up period.
The upper panel of Table 6 summarizes the impacts of PRIDE on employment that was covered by unemployment insurance (UI) and on public assistance receipt for the two-year follow-up period, and the middle and lower panels of the table show each of the two years separately. Note that off-the-books jobs and unpaid work, such as work experience placements, are not captured by the UI system.1 As shown, a larger percentage of the PRIDE group was “ever employed” in UI-covered jobs at some point during the follow-up period. Table 6 shows that 33.7 percent of the PRIDE group versus 26.5 percent of the control group worked at some point during the two-year period, for an increase of 7.2 percentage points.2 However, the low employment levels for both groups are noteworthy. About two-thirds of the PRIDE group and almost three-fourths of the control group did not work in a UI-covered job during the two-year follow-up. Furthermore, only 15.7 percent of the PRIDE group and 12.8 percent of the control group worked during a typical quarter during this period.3 These patterns reflect the limited employability of the PRIDE target group: Individuals who have physical and mental health barriers have lower rates of employment than the general population. (See Box 3.)
The middle and bottom panels of Table 6 and the top panel of Figure 4 show employment rates and effects over time. These data show that the employment impacts persisted throughout the follow-up period. The pattern suggests that PRIDE may continue to have an employment effect beyond Year 2.
| Outcome | PRIDE Group | Control Group | Difference (Impact) | P-Value | ||
|---|---|---|---|---|---|---|
| Years 1 and 2 | Employment (%) | Ever employed | 33.7 | 26.5 | 7.2 *** | 0.000 |
| Average quarterly employment | 15.7 | 12.8 | 2.9 *** | 0.004 | ||
| Employed 8 consecutive quarters | 3.2 | 2.9 | 0.2 | 0.725 | ||
| Income ($) | Earningsa | 3,536 | 2,982 | 554 a | NA | |
| Amount of cash assistance received | 10,732 | 11,550 | -818 *** | 0.000 | ||
| Amount of food stamps received | 6,256 | 6,386 | -130 | 0.123 | ||
| Total measured incomea b | 20,455 | 21,016 | -562 a | NA | ||
| Year 1 | Employment (%) | Ever employed | 23.0 | 18.7 | 4.3 *** | 0.004 |
| Average quarterly employment | 13.1 | 11.3 | 1.8 * | 0.065 | ||
| Employed 4 consecutive quarters | 5.5 | 4.7 | 0.7 | 0.384 | ||
| Income ($) | Earningsa | 1,330 | 1,167 | 163 a | NA | |
| Amount of cash assistance received | 5,806 | 6,100 | -293 *** | 0.001 | ||
| Amount of food stamps received | 3,301 | 3,334 | -34 | 0.395 | ||
| Total measured incomea b | 10,396 | 10,658 | -262 a | NA | ||
| Year 2 | Employment (%) | Ever employed | 27.1 | 22.0 | 5.1 *** | 0.002 |
| Average quarterly employment | 18.3 | 14.3 | 4.0 *** | 0.001 | ||
| Employed 4 consecutive quarters | 9.8 | 7.9 | 1.9 * | 0.092 | ||
| Income ($) | Earningsa | 2,206 | 1,815 | 391 a | NA | |
| Amount of cash assistance received | 4,925 | 5,450 | -525 *** | 0.000 | ||
| Amount of food stamps received | 2,956 | 3,052 | -96 * | 0.072 | ||
| Total measured incomea b | 10,058 | 10,358 | -300 a | NA | ||
| Sample size (total = 2,648) | 1,553 | 1,095 | ||||
| SOURCES: MDRC calculations from unemployment insurance (UI) wage records from the State of New York and public assistance records from New York City. NOTES: See Appendix E. a This difference is not tested for statistical significance because the UI earnings data were provided as group averages and the number of groups was too small to provide for a fair test. b This measure represents the sum of UI-covered earnings, cash assistance, and food stamps. |
Employment stability can be examined based on the percentage working for several quarters consecutively. As shown in Table 6, employment stability was low for both research groups. About 3 percent of the sample members in each group worked for eight consecutive quarters in a UI-covered job during the entire follow-up period. Nevertheless, PRIDE increased employment stability during Year 2, as seen by the increase in the percentage of sample members employed for four consecutive quarters. Over the two-year period, PRIDE group members earned $3,536, on average, while control group members earned $2,982. Note that this difference was not tested for statistical significance.4
| Outcome | PRIDE Group | Control Group | Difference (Impact) | P-Value | ||
|---|---|---|---|---|---|---|
| Employment status (%) | Ever employed since random assignment | 33.4 | 24.4 | 9.0 *** | 0.005 | |
| Currently employed | 15.6 | 12.9 | 2.7 | 0.276 | ||
| No longer employed | 17.5 | 11.5 | 6.0 ** | 0.019 | ||
| Characteristic of current/most recent job (%) | Working status | Full time | 22.8 | 16.4 | 6.4 ** | 0.022 |
| Part time | 10.3 | 7.9 | 2.3 | 0.270 | ||
| Employed at a "good job"a | 7.9 | 3.7 | 4.2 ** | 0.012 | ||
| Hours | Average hours per week | 10.6 | 7.8 | 2.8 ** | 0.012 | |
| Total hours per week (%) | Less than 30 | 10.3 | 7.9 | 2.3 | 0.270 | |
| 30-34 | 2.9 | 2.7 | 0.2 | 0.850 | ||
| 35-44 | 15.5 | 11.4 | 4.1 * | 0.098 | ||
| 45 or more | 4.5 | 2.4 | 2.1 | 0.111 | ||
| Average hourly wage (%) | Less than $5.00 | 9.3 | 6.0 | 3.4 * | 0.080 | |
| $5.00 - $6.99 | 7.3 | 5.6 | 1.7 | 0.336 | ||
| $7.00 - $8.99 | 7.8 | 5.9 | 1.9 | 0.308 | ||
| $9.00 or more | 8.7 | 6.9 | 1.8 | 0.368 | ||
| Average hourly wage among those employed ($) | 6.82 | 6.98 | -0.17 | NA | ||
| Earnings | Average earnings per week ($) | 73 | 55 | 19 ** | 0.044 | |
| Total earnings per week (%) | Less than $200 | 18.1 | 11.1 | 7.0 *** | 0.006 | |
| $201-$300 | 6.4 | 8.4 | -2.0 | 0.286 | ||
| $301-$500 | 6.3 | 3.5 | 2.8 * | 0.077 | ||
| $500 or more | 2.4 | 1.3 | 1.0 | 0.303 | ||
| Average weekly earnings among those employed ($) | 215 | 202 | 13 | NA | ||
| Benefits (%) | Employer-provided benefits | Sick days with full pay | 6.1 | 6.0 | 0.0 | 0.979 |
| Paid vacation | 7.6 | 6.4 | 1.2 | 0.505 | ||
| Paid holidays other than Christmas and New Year | 6.6 | 6.8 | -0.2 | 0.903 | ||
| Dental benefits | 4.8 | 4.9 | -0.1 | 0.967 | ||
| A retirement plan | 3.7 | 3.4 | 0.3 | 0.823 | ||
| A health plan or medical insurance | 6.8 | 5.6 | 1.2 | 0.480 | ||
| Scheduleb(%) | Regular | 18.2 | 11.6 | 6.7 *** | 0.010 | |
| Split | 0.6 | 0.2 | 0.4 | 0.448 | ||
| Irregular | 1.8 | 1.1 | 0.6 | 0.463 | ||
| Evening shift | 2.1 | 3.2 | -1.1 | 0.344 | ||
| Night shift | 1.5 | 0.9 | 0.6 | 0.443 | ||
| Rotating shift | 1.7 | 2.2 | -0.5 | 0.641 | ||
| Other schedule | 0.0 | 1.3 | -1.3 ** | 0.030 | ||
| Odd job | 7.0 | 3.6 | 3.4 ** | 0.036 | ||
| Job skills index (%) | Percentage reporting that the job requires each at least monthly: | Reading and writing skills | 19.9 | 14.6 | 5.4 ** | 0.047 |
| Work with computers | 7.4 | 7.1 | 0.3 | 0.865 | ||
| Arithmetic | 11.7 | 9.4 | 2.2 | 0.309 | ||
| Customer contact | 25.7 | 20.9 | 4.8 | 0.107 | ||
| Sample size (total = 759) | 380 | 379 | ||||
| SOURCE: MDRC calculations from responses to the ERA 12-Month Survey. NOTES: See Appendix F. a This definition of a good job is adapted from Johnson and Corcoran (2003). A "good job" is one that offers 35 or more hours per week and either (1) pays $7.00 or more per hour and provides health insurance or (2) pays $8.50 or more per hour and does not provide health insurance. b A split shift is defined as one consisting of two distinct periods each day. An irregular schedule is defined as one that changes from day to day. A rotating shift is one that changes regularly from days to evenings to nights. |
Box 3Cross-Site Comparison of Control Groups in Year 1The PRIDE program targeted a population with health problems thought to limit their employability. This box compares the PRIDE control group with other welfare recipients in the ERA evaluation to shed light on the extent to which the PRIDE group can be considered “hard-to-employ.” Note that the other ERA programs targeted different populations. The Minnesota program targeted long-term TANF recipients who were unable to find jobs through the standard welfare-to-work services. The Portland program targeted individuals who cycled back onto TANF and who were unemployed. Finally, the Houston, Corpus Christi, and Fort Worth programs served both welfare recipients and applicants. (See Appendix Table D.1 for further information on the ERA sites.) The comparisons suggest that the PRIDE group was more disadvantaged relative to these other groups. The PRIDE control group fared worse in terms of employment and earnings outcomes. Also, compared with the other control group samples, cash assistance receipt fell at a much lower rate for the PRIDE control group. This reflects the limited employability of the PRIDE target group.
|
Data from the ERA 12-Month Survey provide additional information on employment and job characteristics for single parents.5 The first panel of Table 7 shows the percentage of sample members in each research group who were employed since random assignment and their employment status at the time of the survey interview. Similar to the administrative record results, the survey shows that the PRIDE program increased the percentage of single parents who worked during Year 1, although there was no effect on employment at the time of the survey.6 The remaining rows of the table display the characteristics of respondents’ current or most recent job at the time of the survey. Note that when the percentages for the categories are summed, they add up to the percentage of sample members who worked since random assignment.
The PRIDE program increased average hours worked per week and weekly earnings. The program also increased the percentage of sample members in “good jobs”7 and the percentage of sample members in low-paying jobs.8 Note that most of the employment increase appears to have been in jobs that paid less than $200 per week.9
-
The PRIDE program reduced cash assistance receipt and payments among single parents.
PRIDE produced a significant reduction in cash assistance receipt and payments. Table 6 shows that the program reduced welfare payments by $293 during Year 1 and by $525 during Year 2. There was also a small reduction in food stamp receipt during Year 2.10
[D] |
During the two-year follow-up period, the PRIDE group received an average of 20 months of cash assistance, compared with 21 months for the control group, for a significant difference of -1 month (Appendix Table B.1). The lower panel of Figure 4 shows that the difference in cash assistance receipt grew over time. At the end of the two-year follow-up period, 78.1 percent of the PRIDE group, compared with 82.1 percent of the control group, received cash assistance, for a 4.0 percentage point difference (Appendix Table B.6).
The decreases in cash assistance payments were driven by a combination of lower receipt rates and lower amounts per recipient –– the results of an increase in employment and an increase in sanctioning. Going to work can make an individual eligible for a reduced grant amount or no longer eligible for any assistance.11 The program increased the number of sample members who were employed and not receiving cash assistance (Appendix Table B.5), which shows that some PRIDE group members exited welfare as they found work. However, there is also evidence that some sample members who did not find employment had their grants reduced.
A reduction in cash assistance receipt and grant amounts without an increase in earnings may be the result of an increase in sanctioning. As noted earlier, a sanction would result in reduction of an individual’s welfare grant. There is some evidence that the program decreased cash assistance through increased sanctioning among the PRIDE group. First, the program decreased cash assistance payments among subgroups that did not have an increase in employment (see Table 8). Second, an examination of participation data shows that about 32 percent of the PRIDE group were sanctioned during the two-year follow-up period, compared with only about 8 percent of the control group (see the preceding section).
Finally, Table 6 shows that the PRIDE program did not increase measured income, defined as the sum of earnings, cash assistance, and food stamps. Any increases in earnings appear to have been more than offset by reductions in cash assistance.
- The employment impacts generated by PRIDE were concentrated among sample members who had received assistance for fewer than 60 months.
Table 8 present impacts for single parents who were still receiving TANF at study entry, compared with those who had already been moved into the Safety Net program.12 All the employment increase for single parents was concentrated among TANF recipients. The PRIDE program increased employment for TANF recipients by 10.9 percentage points, and this effect persisted through the end of the follow-up period. For example, within this group, 23.6 percent of the PRIDE group, compared with 18.5 percent of the control group, were employed in a UI-covered job during Quarter 9, for a statistically significant difference of 5.2 percentage points (not shown). Although the earnings effects were not tested for statistical significance, the data suggest that the program may have increase earnings for this group, showing a fairly sizable difference of $873 over the two-year period.
In contrast, the program did not significantly increase employment among single parents receiving Safety Net assistance. Recall that this group consists of long-term recipients who had not found jobs and who had reached their 60-month TANF limit prior to entering the study. This group may have been at a greater disadvantage in finding employment. The differences in the employment impacts between the two subgroups are statistically significant.
Although the program increased employment only for the subgroups of TANF recipients, PRIDE significantly decreased cash assistance payments and receipt (not shown) for all single parents, both TANF and Safety Net recipients. This decrease may reflect the fact that the sanctioning rate was higher for the PRIDE group than for the control group.13
-
The PRIDE program had no effect on disability income receipt or on health outcomes.
PRIDE did not affect the percentage of sample members who reported receiving disability income. According to the ERA 12-Month Survey, 21.8 percent of PRIDE households received Supplemental Security Income (SSI) or disability income at the time of the interview, compared with 19.1 percent of the control group (Appendix Table C.1). The survey results also show that the program had no effect on the rate of application for disability benefits (not shown). As noted in the implementation section of this report, individuals who were initially assessed to be likely candidates for SSI were not included in the research sample.
Given that the program took into account the sample members’ health limitations for their work placements, PRIDE may have had positive effects on their health status. As shown in Appendix Table C.3, however, the health outcomes for both research groups were similar, showing no effects of the program. For example, the majority of sample members in both groups had experienced bodily pain at the time of the survey interview.14
| Outcome | PRIDE Group | Control Group | Difference (Impact) | P-Value | P-Value for Difference | ||
|---|---|---|---|---|---|---|---|
| TANF recipients | Employment (%) | Ever employed | 39.6 | 28.7 | 10.9 *** | 0.00 | 0.028 |
| Average quarterly employment | 18.8 | 14.7 | 4.1 *** | 0.00 | 0.267 | ||
| Employed 4 consecutive quarters | 4.2 | 3.5 | 0.7 | 0.50 | 0.725 | ||
| Income ($) | Earnings | 4,514 | 3,641 | 873 a | NA | NA | |
| Amount of cash assistance received | 9,889 | 10,721 | -832 *** | 0.00 | 0.042 | ||
| Amount of food stamps received | 5,908 | 6,074 | -165 | 0.15 | 0.527 | ||
| Total measured incomeb | 20,311 | 20,435 | -124 a | NA | NA | ||
| Sample size (total = 1,615) | 945 | 670 | |||||
| Safety Net recipients | Employment (%) | Ever employed | 24.4 | 22.6 | 1.8 | 0.48 | |
| Average quarterly employment | 10.9 | 9.7 | 1.2 | 0.39 | |||
| Employed 4 consecutive quarters | 1.6 | 1.9 | -0.3 | 0.70 | |||
| Income ($) | Earnings | 2,145 | 1,703 | 442 a | NA | ||
| Amount of cash assistance received | 12,087 | 12,854 | -767 *** | 0.00 | |||
| Amount of food stamps received | 6,797 | 6,890 | -93 | 0.44 | |||
| Total measured incomeb | 21,029 | 21,448 | -419 a | NA | |||
| Sample size (total = 1,033) | 608 | 425 | |||||
| SOURCES: MDRC calculations from unemployment insurance (UI) wage records from the State of New York and public assistance records from New York City. NOTES: See Appendix E. a This difference is not tested for statistical significance because the UI earnings data were provided as group averages and the number of groups was too small to provide for a fair test. b This measure represents the sum of UI-covered earnings, cash assistance, and food stamps. |
Effects for Safety Net Recipients Without Dependent Children
-
The PRIDE program increased employment among Safety Net recipients without dependent children.
Table 9 shows the effects of PRIDE on UI-covered employment and public assistance for Safety Net recipients without dependent children. The program increased the percentage of Safety Net recipients who worked in a UI-covered job at some point during the two-year follow-up period, and it also led to a small increase in employment stability at the end of the period. Like single parents, however, Safety Net recipients without dependent children also had very low employment rates, regardless of research group.
The program did not have an overall effect on public assistance receipt or payments for Safety Net recipients without dependent children. Participation data suggest that these results may be due to relatively low sanctioning rates among this group. As noted above, among single parents, 32 percent of the PRIDE group and 8 percent of the control group were sanctioned during the two-year follow-up period. Sanctioning rates were lower among the Safety Net recipients: 19 percent of the PRIDE group and 7 percent of the control group were sanctioned. By the end of the follow-up period, however, the percentage of Safety Net recipients in the PRIDE group who were receiving public assistance was lower than the percentage in the control group. In Quarter 9, 57.8 percent of the PRIDE group received benefits, compared with 64.6 percent of the control group (not shown). The difference of 6.8 percentage points just misses statistical significance (p = 0.11).
| Outcome | PRIDE Group | Control Group | Difference (Impact) | P-Value | ||
|---|---|---|---|---|---|---|
| Years 1 and 2 |
Employment (%) | Ever employed | 25.4 | 18.9 | 6.4 * | 0.074 |
| Average quarterly employment | 12.3 | 8.3 | 4.0 * | 0.052 | ||
| Employed 8 consecutive quarters | 2.2 | 1.8 | 0.4 | 0.744 | ||
| Income ($) | Earnings | 2,603 | 1,573 | 1,030 a | NA | |
| Amount of cash assistance received | 5,427 | 5,418 | 10 | 0.972 | ||
| Amount of food stamps received | 2,854 | 2,834 | 20 | 0.867 | ||
| Total measured incomeb | 10,884 | 9,825 | 1,060 a | NA | ||
| Year 1 | Employment (%) | Ever employed | 20.2 | 12.8 | 7.4 ** | 0.021 |
| Average quarterly employment | 11.1 | 7.0 | 4.0 ** | 0.037 | ||
| Employed 4 consecutive quarters | 3.1 | 1.8 | 1.3 | 0.356 | ||
| Income ($) | Earnings | 884 | 525 | 358 a | NA | |
| Amount of cash assistance received | 3,114 | 2,995 | 120 | 0.355 | ||
| Amount of food stamps received | 1,521 | 1,481 | 40 | 0.466 | ||
| Total measured incomeb | 5,519 | 5,001 | 518 a | NA | ||
| Year 2 | Employment (%) | Ever employed | 19.6 | 14.6 | 5.1 | 0.123 |
| Average quarterly employment | 13.5 | 9.6 | 4.0 | 0.113 | ||
| Employed 4 consecutive quarters | 8.8 | 4.9 | 3.9 * | 0.097 | ||
| Income ($) | Earnings | 1,719 | 1,047 | 672 a | NA | |
| Amount of cash assistance received | 2,313 | 2,423 | -110 | 0.513 | ||
| Amount of food stamps received | 1,333 | 1,353 | -20 | 0.791 | ||
| Total measured incomeb | 5,366 | 4,824 | 542 a | NA | ||
| Sample size (total = 540) | 356 | 184 | ||||
| SOURCES: MDRC calculations from unemployment insurance (UI) wage records from the State of New York and public assistance records from New York City. NOTES: See Appendix E. a This difference is not tested for statistical significance because the UI earnings data were provided as group averages and the number of groups was too small to provide for a fair test. b This measure represents the sum of UI-covered earnings, cash assistance, and food stamps. |
1 Other jobs not covered by the UI system include federal, out-of-state, and military jobs and self-employment. (back)
2 The employment, public assistance, and survey impacts are estimated in a regression framework, which also controls for a range of background characteristics, including race/ethnicity, number of children, age of children, service provider, intake period, prior food stamp receipt, prior employment, and prior TANF receipt. (back)
3 The average quarterly employment rate is calculated as total quarters employed divided by 8 (the number of quarters in the follow-up period), expressed as a percentage. (back)
4 Earnings impacts were not estimated because New York State did not provided MDRC with earnings data for each individual but, instead, provided average earnings for groups of individuals; the groups were defined by research status, assistance type, and quarter of random assignment. It was determined that the number of groups (16) was too small to provide for a fair test. (back)
5 Safety Net recipients without dependent children were not surveyed. (back)
6 No difference in employment was found when UI-covered employment was measured at a point in time (the survey interview date), which is indicative of fairly unstable employment. As shown in Appendix Table B.2, the UI-covered employment in Quarter 5 –– which is the time when many survey respondents were interviewed –– shows a sudden decrease when compared with other quarters. Furthermore, since the survey respondent sample was limited to only a few months of intake, a “cohort effect” may have been introduced. As shown in Appendix Table H.3, the effects on employment for the respondent sample were smaller than the effects for the full research sample. (See Appendix H for further information.) (back)
7 As defined by Johnson and Corcoran (2003), a “good job” is one that offers 35 or more hours per week and either (1) pays $7.00 or more per hour and offers health insurance or (2) pays $8.50 or more per hour and does not provide health insurance.(back)
8 Some fraction of the new employment also appears to have been in jobs paying $5.00 or less per hour. For some respondents, the reported wage may be only a rough measure of hourly earnings. In many cases, for example, wages on the survey are not reported per hour but are calculated using weekly or monthly earnings, divided by usual hours per week or month. In other cases, respondents were reporting earnings from informal jobs or odd jobs –– such as piecemeal work or baby-sitting, for example –– that may pay less than the minimum wage. (back)
9 Appendix Tables B.3 and B.4 provide additional information on job retention and advancement for survey respondents. (back)
10 The reduction in food stamp receipt is also evident in responses to the ERA 12-Month Survey. (See Appendix Table C.1.) (back)
11 In New York State, the monthly earnings limit for continued TANF eligibility for a single parent with two children in 2003 was $1,067 (U.S. House of Representatives, Committee on Ways and Means, 2003). (back)
12 Impacts were estimated for other subgroups, including subgroups defined by type of physical or mental disability, race/ethnicity, number of children, age of children, age of participant, and service provider (not shown). These results show no or few consistent statistically significant differences in impacts between these PRIDE member subgroups and their control group counterparts. One exception is for recipients younger than age 41. The PRIDE program led to a larger reduction in cash assistance payments. (back)
13 No differences were found between the sanctioning rates of the TANF and the Safety Net recipients (not shown). (back)
14 Impacts on noneconomic outcomes, such as child care and household composition, can be found in Appendix C. Except for child care use, the program did not have effects on these outcomes. (back)
| Table of Contents | Previous | Next |