Implementation, Participation Patterns, Costs, and Two-Year Impacts of the Detroit Welfare-to-Work Program: Executive Summary

08/01/2000

Prepared for:
U.S. Department of Health and Human Services
Administration for Children and Families
Office of the Assistant Secretary for Planning and Evaluation

U.S. Department of Education
Office of the Under Secretary
Office of Vocational and Adult Education

Prepared by:
Mary Farrell with
Gayle Hamilton, Christine Schwartz, Laura Storto

Manpower Demonstration Research Corporation

 

This document was prepared as part of the National Evaluation of Welfare-to-Work Strategies (NEWWS).  The Manpower Demonstration Research Corporation (MDRC) is conducting the NEWWS Evaluation under a contract with the U.S. Department of Health and Human Services (HHS), funded by HHS under a competitive award, Contract No. HHS-100-89-0030.  HHS is also receiving funding for the evaluation from the U.S. Department of Education.

The findings and conclusions presented herein do not necessarily represent the official positions or policies of the funders.

Contents

  1. An Overview in Brief
  2. Detroit's Evaluation Context
  3. The Implementation of MOST
  4. The Implementation of Work First
  5. Participation in Employment-Related Activities by Program and Control Group Members
  6. The Cost of the Detroit Program
  7. Impacts on Employment, Earnings, and Welfare Receipt
  8. Comparisons With Other Programs in the NEWWS Evaluation
  9. Conclusion

Michigan's current welfare-to-work program evolved over the past decade from one that emphasized participation in education and training activities to one that focused on quick job entry as the route to financial independence for welfare recipients. In addition, it shifted many of the responsibilities previously performed by the welfare department to private and public organizations outside the welfare department and exempted fewer welfare recipients from participating in the program. The program that emerged became one of the keystones of Michigan's overall welfare reform program, which was approved for implementation under the 1996 law.

This report examines the welfare-to-work programs operated in two of Detroit's welfare districts: Fullerton-Jeffries and Hamtramck. It describes Michigan Opportunity and Skills Training (MOST), an education-focused program that was in place in these two offices in 1992 at the start of the evaluation, and the transition to Work First, an employment-focused program emphasizing job search services that was implemented in October 1994 and is one component of Michigan's current welfare reform program. It follows for two years the welfare recipients who were assigned to MOST, almost one-quarter of whom were referred to the Work First program within the two-year period, and examines the types of services and messages that they received, the cost of both strategies, and the effects of the treatment received on welfare receipt, employment, and earnings. It follows an early group of individuals for three years.

The Detroit welfare-to-work program is being evaluated as part of the National Evaluation of Welfare-to-Work Strategies (NEWWS Evaluation; formerly called the JOBS Evaluation), conducted by the Manpower Demonstration Research Corporation (MDRC) under contract to the U.S. Department of Health and Human Services, with support from the U.S. Department of Education and the W. K. Kellogg Foundation. NEWWS is a comprehensive study of 11 welfare-to-work programs in seven sites. Throughout this report, comparisons are made between the Detroit program and the other NEWWS programs. Two recently released reports provide a more comprehensive comparison among all programs, including results on children's well-being, child care use while employed, supports provided to individuals who leave welfare for employment, and additional measures of self-sufficiency.(1)  A future report will examine five-year results for all programs and will compare program benefits with program costs.

This report examines the implementation of the MOST and Work First programs and analyzes data on participation, AFDC and Food Stamp receipt, employment, and earnings. Its results are based on a random assignment research design, in which welfare recipients were randomly assigned to one of two groups: the program group, whose members were assigned initially to MOST and were subject to MOST participation requirements; and the control group, whose members were not subject to any participation requirements for three years, but who could seek out, on their own, education and training programs available in the community and who were eligible for child care and transportation assistance provided by the MOST office. Program group members who were still on AFDC in the fall of 1994, and were mandatory, were referred to the new Work First program.

For this report, the combination of MOST and Work First services that program group members received during the follow-up period is referred to as the "Detroit program." Program group members who were randomly assigned early in the evaluation received more MOST program services and those randomly assigned later in the evaluation received more Work First services (but some MOST services). Thus, the early enrollees experienced a program that was very different from the program experienced by the later enrollees.

The report sample consists of 4,459 single-parent sample members (2,226 program group members and 2,233 control group members) who were randomly assigned to the MOST program between May 1992 and June 1994. The two groups are tracked over time, and the differences between the groups' particular outcomes (for example, average two-year earnings) constitute the effects, or "impacts," of the Detroit welfare-to-work program. The impacts discussed below are statistically significant unless otherwise noted.(2)

[ Go to Contents ]

An Overview in Brief

To assess the magnitude of Detroit's results, this report makes a number of comparisons, primarily between program and control group members. In addition, it compares the program's effects in the two district offices. The report also discusses impacts for subgroups defined by education level and cohorts defined by assignment date. Finally, it compares Detroit's impacts with those of the other programs in the NEWWS Evaluation to show the relative effectiveness of Detroit's approach. (See the accompanying text box for a brief description of the programs in the evaluation.)

 

The National Evaluation of Welfare-to-Work Strategies

The National Evaluation of Welfare-to-Work Strategies is assessing the effectiveness of 11 welfare-to-work programs in seven sites, including Detroit. Four sites in the evaluation operated two programs simultaneously in order to test the strengths and limitations of two different program approaches. Three of these four sites  Atlanta, Georgia; Grand Rapids, Michigan; and Riverside, California  ran two programs that used different employment preparation strategies:  one, called the Labor Force Attachment (LFA) approach, is based on the view that the workplace is where welfare recipients can best learn work habits and skills, and thus emphasizes placing people into jobs quickly, even at low wages. The second, called the Human Capital Development (HCD) approach, emphasizes education and training as a precursor to employment, reflecting the belief that the required skills levels for many jobs are rising and that an investment in the "human capital" of welfare recipients will allow them to obtain better and more secure jobs. The goal of the LFA programs was rapid employment, and job search was the prescribed first activity for virtually the entire caseload. In contrast, most people in the HCD programs were first assigned to education or training; basic education was the most common activity because of the generally low educational attainment of the enrollees at program entry.

In the fourth site  Columbus, Ohio  different case management approaches were compared side by side. "Traditional" case management required clients to interact with two staff members: one worker who processed welfare benefits and another worker who enrolled people in employment activities. "Integrated" case management required clients to interact with one worker for both welfare eligibility and employment services.

The study in the other two sites  Oklahoma City, Oklahoma, and Portland, Oregon  tested the net effects of the sites' welfare-to-work programs (similar to the study in Detroit). The Columbus and Oklahoma City programs primarily utilized an HCD approach. The Portland program can be considered to be a blend of strong LFA elements and moderate HCD elements.

In total, the 11 evaluation programs ranged from strongly LFA-focused to strongly HCD-focused and from somewhat voluntary to highly mandatory. The program sites offered diverse geographic locations, caseload demographics, labor markets, and AFDC grant levels. However, because of NEWWS Evaluation selection criteria, the programs were all "mature" programs, relatively free of the transitional problems associated with the start-up of a complex, multi-component welfare-to-work program. These programs, while not representing all programs in the nation, represent a wide range of welfare-to-work options.

The key findings include the following:

  • While the Detroit MOST program communicated to enrollees the importance of participating in MOST and the penalties for not doing so, it did not strongly enforce the mandate to participate. The MOST staff recommended that clients enroll in education and training activities; however, they did not monitor participation closely and rarely requested financial sanctions (reductions in AFDC grants) for noncompliant clients.
  • Control group members' levels of participation in employment and training activities were high for a NEWWS program. Relative to the control group's participation, the Detroit program produced only a small increase in the use of employment-related activities. More than 40 percent of controls participated in an employment-related activity within two years of follow-up. The Detroit program increased participation in any employment-related activity by only 9 percentage points above the control group rate. Much of this increase was accounted for by the greater use of vocational training and job search by Hamtramck program group members. Work First accounted for less than half of the job search participation.
  • During the first two years of the follow-up period, the Detroit program only modestly increased employment and earnings and reduced AFDC receipt; impacts grew larger in the third year. During the first two years of follow-up, Detroit increased employment for program group members by 4 percentage points and earnings by $367 (or 9 percent over that of the control group). The program did not generate AFDC savings until the second year, when it reduced average AFDC payments by $139 (or 3.5 percent). While statistically significant, these impacts are among the smallest found for a NEWWS program and reflect the small increases in rates of participation.

    Another year of follow-up data are available for people who entered the study by December 1993 (74 percent of the full research sample). In year 3, average earnings for this group increased by $585 (or 16 percent) and AFDC payments decreased by $183 (or 5 percent).

  • The impact of the MOST program alone is reflected in the experiences of sample members enrolled in the program early on. The Hamtramck office generated increases in employment and earnings for an early cohort who did not receive Work First services within the first two years, while the Fullerton-Jeffries office did not. An early cohort of sample members who were assigned to the program between May and December 1992 received only the MOST "treatment" within the first two years of follow-up. For this cohort, the Hamtramck program increased employment by 12 percentage points and earnings by $1,291 (or 54 percent). The Fullerton-Jeffries program did not generate similar impacts within two years, although impacts emerged in the third year. Hamtramck's impacts may be attributable in part to its success in increasing the number of program group members who obtained a trade license or certificate, which may have increased their ability to find employment or find higher-paying employment.
  • The effects of the Work First program are reflected in the impacts of a later cohort who received fewer months of MOST services and were more likely to be referred to and have participated in Work First. Impacts for the later Fullerton-Jeffries cohort emerged in the second year of follow-up, which suggests that the Work First program was having a positive effect on Fullerton-Jeffries program group members' employment and welfare outcomes. The Fullerton-Jeffries program referred 58 percent of program group members randomly assigned between January and June 1994 to Work First within two years. For this cohort, earnings increased in year 2 by a large and statistically significant $1,032 (or 35 percent) and AFDC payments declined by a statistically significant $441 (or 11 percent). Hamtramck, which referred 37 percent of its program group members to Work First, did not produce impacts on earnings for this later cohort.
  • The impacts by office and random assignment cohort suggest that both MOST and Work First programs can produce positive results. However, the implementation practices of each office likely influenced the results. The Hamtramck MOST program's emphasis on vocational training may explain its larger employment and earnings impacts relative to Fullerton-Jeffries' impacts for the early cohort. The Fullerton-Jeffries office referred a greater portion of its sample to Work First relative to Hamtramck  this may explain its larger impacts for the later cohort.

Following a brief discussion of the context of the Detroit program, this summary describes the implementation of the MOST and Work First programs, the levels of participation by both the program and control groups, cost findings, and the effects on employment, earnings, and welfare receipt for the full sample and defined cohorts.

[ Go to Contents ]

Detroit's Evaluation Context

The results in this report should be considered in the context of Detroit's research sample and program environment. Detroit is the largest urban area in the NEWWS Evaluation and provides the opportunity to evaluate the effectiveness of a welfare-to-work program under conditions found in many inner-city welfare offices and employment programs. Fullerton-Jeffries, an inner-city Detroit office, serves a predominately black caseload. Hamtramck, which is located in the northeast section of Wayne County and includes parts of Detroit, serves a more racially and ethnically diverse caseload that has a majority of blacks, but also Eastern European, Middle Eastern, and other immigrants.

Detroit had a more disadvantaged sample than the other six sites in the evaluation. It had the lowest percentage employed in the year prior to random assignment (approximately 20 percent), and the highest percentage who had received public assistance for two years or more on their own or spouse's case (almost 75 percent). In addition, almost half (42 percent) did not have a high school diploma or GED certificate at the time of random assignment.

Throughout the follow-up period, Detroit's labor market was improving, with increasing employment and decreasing unemployment rates. Still, it was less robust than in the other six sites.

[ Go to Contents ]

The Implementation of MOST

  • The Detroit MOST program emphasized building clients' human capital through education and training rather than moving clients into the labor market quickly.

The Detroit MOST staff urged enrollees to obtain education and training before seeking employment. Program group members without a high school diploma or GED were encouraged to obtain a GED; those with this credential were encouraged to enroll in vocational training or college.

  • The income maintenance (IM) staff knew little about the MOST program and had little discussion with clients regarding their participation in MOST.

MOST staff relied on IM staff to refer clients to the MOST program, impose and lift financial sanctions, and alert them to circumstances that could affect clients' participation. Staff surveys established that fewer IM workers knew about MOST and fewer reported that they received helpful training on MOST than IM staff from any other program in the NEWWS Evaluation. MOST staff mentioned that they preferred to control the flow of information regarding the program, ensuring that they communicated consistent messages to clients.

  • Assessment was not a strong component of the MOST program.

Fewer MOST staff members in Detroit tried to learn in depth about their clients, tried to identify and remove barriers, and encouraged and provided positive reinforcement to clients than in any other program in the NEWWS Evaluation. This is supported by results from the Two-Year Client Survey: few Detroit program group members felt that their case manager knew a lot about them and their family or would help them resolve their problems.

  • The Detroit MOST program communicated to enrollees the importance of participating in MOST and the penalties for not doing so. However, the MOST program was considerably less mandatory than other programs in the NEWWS Evaluation.

As evidenced by both the client and staff surveys, the majority of program group members were informed about penalties for noncompliance. Nonetheless, field research findings suggest that Detroit staff did not closely monitor individuals in program activities and follow up quickly when attendance problems arose, as staff did in other programs. In addition, the client and staff surveys indicate that MOST staff rarely imposed financial sanctions in response to noncompliance. Only 3 percent of Detroit program group members reported that they had been sanctioned, the smallest sanctioning rate of any program in the NEWWS Evaluation. Almost all MOST staff reported that they delayed requesting sanctions for noncompliant clients. There was no substantial difference in the degree of mandatoriness between the two district offices.

There are several explanations for the MOST program's weak monitoring of participation and low sanctioning. It was not fully staffed during the evaluation,(3) and case managers reported spending most of their limited time working with clients who were participating on their own initiative, and who requested child care and other support services. Also, unlike some programs in the evaluation, the education and training service providers supplied staff with very little information on clients' attendance problems and progress. Therefore, program staff often did not know the full extent to which their caseload was noncompliant. Finally, staff believed that clients should be given numerous opportunities to come into compliance; they used sanctioning only when all else failed.

[ Go to Contents ]

The Implementation of Work First

  • The Work First program introduced a new system for delivering services that relied on contractual arrangements with private and public organizations to provide case management and service responsibilities.

Prior to the implementation of Work First, the state welfare department's MOST program performed case management duties, which included assigning individuals to activities, monitoring the participation, referring noncompliant clients for sanctioning, and arranging child care and other support services. Most of the employment and training services were provided by outside organizations. After Work First was implemented, many of the case management and service responsibilities shifted to the Work First program, overseen by the Jobs Commission, a cabinet-level agency. Specifically, Work First had contractual arrangements with public and private organizations in the area that were responsible for assigning clients to activities and for monitoring their participation (the MOST program continued to handle child care administration and sanctioning responsibilities).

  • Work First emphasized immediate employment rather than participation in longer-term education and training activities.

Work First sought to move clients immediately into the labor market. As a result, a large majority of Work First participants were initially assigned to job search. Those who found employment, including part-time employment, met their Work First requirements and were not expected to continue in the program.

  • Almost half of all program group members were referred to Work First within three years after enrolling in MOST.

Less than one-quarter of program group members were referred to Work First within two years; two-fifths were referred within three years. Program group members were referred to Work First after October 1994 if they met three criteria: (1) they were still on AFDC, (2) they were not exempt from participating in the program, and (3) they were not enrolled in and making satisfactory progress in an education or training activity that was to be completed soon. Because more advantaged program group members were likely to leave AFDC quickly, those who were still on AFDC later and were referred to Work First tended to be more disadvantaged than the full sample. In addition, for the same reason, those referred to Work First tended to be those randomly assigned to a research group later in the study.

  • Contrary to the research design, some control group members were referred to the Work First program.

The research design used for the evaluation specified that control group members were not to be referred to MOST, and later Work First, for at least three years following random assignment. Nevertheless, almost one-quarter of all controls were referred to the program between years 2 and 3. While only 8 percent of them actually participated in Work First within three years, the referral itself may have influenced some control group members' behavior. This departure from the research design mitigates the influence of Work First on the Detroit program's impacts. Still, program group members were more likely than control group members to be referred to Work First, especially program group members who were randomly assigned later in the evaluation.

[ Go to Contents ]

Participation in Employment-Related Activities by Program and Control Group Members

  • Control group members' levels of participation in employment and training activities were high for a NEWWS program.

A sizable proportion of control group members (42 percent) participated in employment-related activities on their own within two years. This level of participation was higher than that in any other program in the evaluation except Grand Rapids. Basic education had the highest participation (almost half of all sample members did not have a high school diploma or GED at the point of random assignment), followed by college and then vocational training. Few controls participated in job search.

Several factors contributed to the high level of participation by controls. First, as was true for the other programs in the NEWWS Evaluation, Detroit control group members who participated on their own were entitled to child care services. Detroit controls were also eligible for bus passes. While control group members were not required to participate, the child care and transportation entitlement likely allowed some individuals to participate who would not have participated otherwise. Second, as discussed earlier, the governor and state legislature implemented a series of welfare reforms throughout the evaluation period that generated considerable publicity and perhaps encouraged all AFDC clients to find and enroll in programs on their own initiative throughout the state. Finally, because of resource constraints, which limited the number of clients who could be served by MOST, the program maximized the use of these scarce slots and first selected those individuals for random assignment who volunteered for the program. Consequently, a relatively high percentage of Detroit control group members were participating in an education and training activity at random assignment. Moreover, education and training services were extensively available in the Detroit community.

  • The Detroit program modestly increased the use of vocational training and job search among program group members.

Figure 1 shows that the Detroit program increased the likelihood of participating in vocational training by 7 percentage points over the control group level and increased job search participation by 7 percentage points (half through MOST and half through Work First). The lack of increases in the other activities reflects the very high participation rate by the control group and the low enforcement of the participation mandate for the program group, discussed above. Most of the increase in training and job search was among program group members in the Hamtramck program.

Data from the Work First management information system show that participation by program group members in Work First employment-related activities within two years after random assignment increased by 4 percentage points over the control group level, virtually all due to an increase in job search participation. Within three years, the program increased participation rates by less than 6 percentage points.

  • The Hamtramck district office increased the number of individuals who obtained a vocational training certificate.

More than 12 percent of program group members assigned to the Hamtramck MOST program received a training credential during the two-year follow-up period, a gain of 10 percentage points over the control group level (the highest reported in the NEWWS Evaluation). The increase was even more pronounced for those Hamtramck program group members who entered the program without a high school diploma or GED. For this subgroup, almost 18 percent of program group members received a trade license or certificate within the two years of follow-up, resulting in an impact of about 14 percentage points. Fullerton-Jeffries did not produce an increase in the receipt of licenses or certificates.

[ Go to Contents ]

The Cost of the Detroit Program

  • The two-year cost of the MOST and Work First programs was estimated to be $2,955, most of which was spent on MOST services, making the Detroit program a relatively low-cost education-focused program.

The Detroit program cost is the sum of the expenses incurred by the welfare department, the Jobs Commission, and area education and training providers that provide MOST and Work First services to program group members. Approximately 85 percent of the Detroit cost was attributable to the MOST program, excluding support services, and 4 percent was attributable to the Work First program. Child care and other support services accounted for the remaining 11 percent of the costs (see Table 1).

The Detroit cost is lower than the average program-related cost for the three education-focused programs studied as part of the NEWWS Evaluation (the average cost of the Atlanta, Grand Rapids, and Riverside education-focused programs was $3,883). While the MOST program had relatively high monthly costs per program participant, they were mitigated by relatively low levels of participation among program group members. Also, few program group members participated in Work First within the two-year follow-up period, resulting in a very low Work First cost. In addition, the welfare department spent less on child care and other support services than did almost all other welfare departments in the evaluation.

Table 1
Two-Year Program-Related, Gross, and Net Costs (in 1993 Dollars)
Program Program-Related Cost per Program Group Member ($) Gross Cost per Program Group Member ($) Gross Cost per Control Group Member ($) Net Cost per Program Group Member ($)
Detroit
  Operating costs 2,643 3,491 1,750 1,762
  Support services 312 421 318 104
  Total 2,955 3,913 2,067 1,845
Average of 3 employment-focused programs
  Operating costs 1,958 2,768 1,546 1,222
  Support services 434 499 171 328
  Total 2,391 3,267 1,717 1,550
Average of 3 education-focused programs
  Operating costs 3,327 4,088 1,481 2,607
  Support services 556 637 166 471
  Total 3,883 4,724 1,647 3,077
SOURCES:  MDRC calculations of Detroit costs are based on fiscal and participation data from the Michigan Family Independence Agency, Department of Education, and Jobs Commission; information collected on tuition charged at proprietary schools attended by sample members; and the MDRC Two-Year Client Survey.  MDRC calculations of the three employment-focused and education-focused programs are from Evaluating Two Welfare-to-Work Program Approaches:  Two-Year Findings on the Labor Force Attachment and Human Capital Development Programs in Three Sites.  Prepared by Gayle Hamilton, Thomas Brock, Mary Farrell, Daniel Friedlander, and Kristen Harknett, MDRC, 1997.  Washington, D.C.:  U.S. Department of Health and Human Services, Administration for Children and Families and Office of the Assistant Secretary for Planning and Evaluation; and U.S. Department of Education.

NOTE:  Rounding may cause slight discrepancies in calculating sums and differences.

Figure 1
Rates of Participation, By Research Group Status

Rates of Participation, by Research Groups Status

 

  • Excluding spending that would have occurred in the absence of a mandatory welfare-to-work program, the two-year net per person cost of Detroit's program was $1,845. This net cost is lower than the average net cost for the three education-focused programs.

Overall, it cost the government an estimated $3,913 per program group member, over a two-year period, to provide education and employment-related services through either the MOST or Work First program or through self-initiated activity outside both programs (referred to as the gross cost in Table 1). Control group members made use of similar services outside the program at a cost of $2,067 per person.

The difference between the program and control group costs results in a net cost to the government of $1,845. This net cost was the additional investment that taxpayers made to provide services for the program group above and beyond the cost of services obtained by control group members. This net cost is much lower than the average net cost per person for the three education-focused programs ($3,077).

The cost analysis by district office shows that Fullerton-Jeffries had higher gross costs for both the program group and control group compared with Hamtramck, due to higher levels of participation in employment-related activities for both groups. Fullerton-Jeffries' net cost per program group member is equivalent to Hamtramck's.

[ Go to Contents ]

Impacts on Employment, Earnings, and Welfare Receipt

  • During the first two years of the follow-up period, the Detroit program modestly increased employment rates and earnings and reduced the AFDC receipt rate. These small impacts were concentrated in the second year of follow-up.

Table 2 shows that the Detroit program produced a small increase of $367 in the average earnings of program group members (a 9 percent increase over the control group average). Earnings increased from a statistically insignificant $57 in the first year to a statistically significant $311 in the second year. Earnings impacts occurred primarily because it helped some individuals find jobs who would not have found employment on their own and, secondarily, because it helped some individuals who would have been employed to increase their earnings.

In year 2, the Detroit program reduced AFDC payments by $139 per program group member (or 3.5 percent) over control group payments.

Although the Detroit program reduced the proportion of program group members receiving AFDC in the last quarter of year 2, 70 percent were still receiving it at the end of the second year; 48 percent were on AFDC and not working at this time.

Table 2
Two-Year Impacts on Employment, Earnings, and AFDC
Outcome Program
Group
Control
Group
Difference
(Impact)
  Percentage
Difference (%)
Years 1-2
Ever employed (%) 62.3 58.2 4.1 *** 7.0
Average total earnings ($) 4,369 4,001 367 * 9.2
Average total AFDC payments received ($) 8,457 8,615 -158   -1.8
Last quarter of year 2
Employed (%) 38.6 35.5 3.1 ** 8.7
Ever received any AFDC payments (%) 70.1 73.7 -3.6 *** -4.8
Sample Size (total = 4,459) 2,226 2,233      
NOTES:  A two-tailed t-test was applied to differences between outcomes for the program and control groups.  Statistical significance levels are indicated as:  * = 10 percent; ** = 5 percent; and *** = 1 percent.

"Percentage difference" equals 100 times "difference" divided by "control group."

 

  • Impacts on employment, earnings, and AFDC receipt increased in the third year.

Three-year earnings and AFDC payment data are available for individuals who entered the program through December 1993 (74 percent of the full sample). Table 3 shows that the Detroit program increased earnings by $937 (or 13 percent) over the three years of follow-up.

Table 3
Three-Year Impacts on Employment, Earnings, and AFDC
for Sample Members Randomly Assigned Through December 1993
Outcome Program
Group
Control
Group
Difference
(Impact)
  Percentage
Difference (%)
Years 1-3
Ever employed (%) 72.4 68.8 3.6 ** 5.3
Average total earnings ($) 8,307 7,370 937 ** 12.7
Average total AFDC payments received ($) 11,710 12,024 -314 -2.6
Last quarter of year 3
Employed (%) 39.8 37.4 2.3   6.2
Ever received any AFDC payments (%) 58.4 62.9 -4.5 *** -7.1
Sample Size (total = 3,293) 1,649 1,644      
NOTES:  A two-tailed t-test was applied to differences between outcomes for the program and control groups.  Statistical significance levels are indicated as:  * = 10 percent; ** = 5 percent; and *** = 1 percent.

"Percentage difference" equals 100 times "difference" divided by "control group."

Earnings gains in the third year were larger than in the previous two years. Specifically, the program group earned an average of $274 more than the control group in the second year of follow-up (not statistically significant) and $585 more in the third year.

The Detroit program reduced AFDC payments by $94 in year 2 (not statistically significant) and $183 in year 3. Over the three-year follow-up period, the Detroit program produced $314 (or 3 percent) in savings.

  • Impacts for an early cohort of individuals who received only MOST services within the first two years of follow-up reveal that the Hamtramck MOST program produced large earnings gains and welfare savings, while the Fullerton-Jeffries MOST program did not.

Table 4 shows that the Hamtramck program produced two-year earnings impacts of $1,291 for an early cohort of individuals who entered the research sample by December 1992 and received only MOST services within the first two years. The earnings impact was $346 and not statistically significant in the first year, but increased to $946 in the second year (a statistically significant increase). The Hamtramck program also reduced two-year AFDC expenditures by $570 (or 6 percent) per program group member over the control group expenditures. In contrast, the Fullerton-Jeffries MOST program did not increase earnings or reduce AFDC expenditures during the two-year follow-up period.

 

Table 4
Two-Year Impacts on Employment, Earnings, and AFDC,
by District Office and Random Assignment Cohort
Outcome and Subgroup Sample
Size
Program
Group
Control
Group
Difference
(Impact)
  Percentage
Difference
(%)
Cumulative earnings over 2 years ($)
   Fullerton-Jefferies
  • Early Cohort
513 3,171 3,537 -365   -10.3
  • Middle Cohort
994 4,348 3,975 373   9.4
  • Late Cohort
452 5,560 4,767 793   16.6
   Hamtramck
  • Early Cohort
531 3,687 2,396 1,291 *** 53.9
  • Middle Cohort
1,254 4,184 4,062 122   3.0
  • Late Cohort
712 5,350 4,962 388   7.8
Cumulative AFDC payments over 2 years ($)
   Fullerton-Jefferies
  • Early Cohort
513 9,054 8,596 458 * 5.3
  • Middle Cohort
994 8,501 8,614 -113   -1.3
  • Late Cohort
452 7,826 8,298 -473   -5.7
   Hamtramck
  • Early Cohort
531 8,512 9,082 -570 ** -6.3
  • Middle Cohort
1,254 8,375 8,562 -186   -2.2
  • Late Cohort
712 8,405 8,641 -236   -2.7
NOTES:  A two-tailed t-test was applied to differences between outcomes for the program and control gorups.  Statistical significance levels are indicated as:  * = 10 percent; ** = 5 percent; and *** = 1 percent.

"Percentage difference" equals 100 times "difference" divided by "control group."

The early cohort consists of those sample members randoly assigned between May and December 1992, the middle cohort of those randoly assigned between January and December 1993, and the late cohort of those randomly assigned between January and June 1994.

As discussed earlier, Hamtramck generated increases in the proportion of program group members who participated in vocational training and the proportion who received a trade certificate. The trade licenses may have given Hamtramck participants access to certain types of jobs for which they might not otherwise qualify. In other evaluations conducted by MDRC, participation in skills training and receipt of a training certificate were associated with substantial earnings gains.(4)

  • An analysis of members of a late cohort, who were more likely than earlier cohorts to receive Work First services, shows that impacts on earnings and AFDC payments emerged in the second year for the Fullerton-Jeffries sample. Conversely, the Hamtramck program group members in this late cohort did not earn more than their control group counterparts.

Earnings and AFDC impacts emerged in the second year of follow-up for a cohort of program group members who enrolled in the Fullerton-Jeffries program between January and June 1994. Specifically, the program increased earnings in year 2 by $1,032 and reduced AFDC expenditures by $441. This cohort received fewer months of MOST services and was more likely to have been referred to Work First than the earlier cohorts. Hamtramck program group members who enrolled in the program during this period did not achieve earnings gains or AFDC reductions in year 2 (although AFDC impacts emerged in the last two quarters of follow-up). This could be due to the fact that Hamtramck group members were less likely to have been referred to Work First than Fullerton-Jeffries group members (Hamtramck referred 22 percentage points more program group members than control group members to Work First within two years, while Fullerton-Jeffries referred 48 percentage points more of its program group members to Work First).

[ Go to Contents ]

Comparisons With Other Programs in the NEWWS Evaluation

  • Detroit's two-year impacts on employment, earnings, and AFDC for the full sample are smaller than impacts in the other programs in the NEWWS Evaluation.

To provide a context for gauging the magnitude of the two-year impacts presented above, it is useful to compare them with those generated by other programs in the NEWWS Evaluation. It is important, however, to keep in mind that in addition to differences in how the programs were implemented, differences in the demographics of the sample and labor market conditions may have influenced the magnitude of these impacts. Detroit had a more disadvantaged sample than the other programs in the evaluation; and while labor market conditions in Detroit improved in the course of the study, the employment growth rate was still lower and the unemployment rate was higher than in the other sites.

Table 5 presents the earnings and AFDC impacts for all 11 programs in the evaluation. The table shows that the Detroit program produced smaller reductions in welfare payments over the two years of follow-up than the other programs. In addition, two-year employment and earnings impacts were smaller than those in nearly all the other programs in the evaluation.

  • Because large impacts are emerging in the third year, two years is not enough time in which to fully assess the effectiveness of the Detroit program relative to the other programs.

 

Table 5
Two-Year Impacts on Total Earnings and Total AFDC Payments
for All 11 Programs in the NEWWS Evaluation
Site and Program Program
Group
Control
Group
Difference
(Impact)
  Percentage
Difference
(%)
Average total earnings in years 1 and 2 ($)
Detroit 4,369 4,001 367 * 9.2
 
Atlanta Labor Force Attachment 5,850 5,006 813 *** 16.2
Atlanta Human Capital Development 5,502 5,006 496 ** 9.9
 
Grand Rapids Labor Force Attachment 5,674 4,639 1035 *** 22.3
Grand Rapids Human Capital Development 5,219 4,639 580 ** 12.5
 
Riverside Labor Force Attachment 5,488 4,213 1,276 *** 30.3
  • Lacked high school diploma or basic skills
4,124 3,133 992 *** 31.7
Riverside Human Capital Development 3,450 3,133 317   10.1
 
Columbus Integrated 7,565 6,892 673 ** 9.8
Columbus Traditional 7,569 6,892 677 *** 9.8
 
Oklahoma 3,518 3,514 5   0.1
 
Portland 7,133 5,291 1,842 *** 34.8
Average total AFDC payments received in years 1 and 2 ($)
Detroit 8,457 8,615 -158   -1.8
 
Atlanta Labor Force Attachment 4,553 4,922 -369 *** -7.5
Atlanta Human Capital Development 4,634 4,922 -288 *** -5.8
 
Grand Rapids Labor Force Attachment 5,944 7,347 1,404 *** -19.1
Grand Rapids Human Capital Development 6,512 7,347 -835 *** -11.1
 
Riverside Labor Force Attachment 8,292 9,600 1,308 *** -13.6
  • Lacked high school diploma or basic skills
8,894 10,302 1,408 *** -13.7
Riverside Human Capital Development 9,253 10,302 1,049 *** -10.2
 
Columbus Integrated 4,775 5,469 -694 *** -12.7
Columbus Traditional 4,939 5,469 -530 *** -9.7
 
Oklahoma 3,391 3,624 -233 *** -6.4
 
Portland 5,818 7,014 1,196 *** -17.1
SOURCE:  StephenFreedman, Daniel Friedlander, Gayle Hamilton, JoAnn Rock, Marisa Mitchell, Jodi Nudelman, Amanda Schweder, and Laura Storto.  Evaluating Alternative Welfare-to-Work Approaches:  Two-Year Impacts for Eleven Programs.  Washington, DC:  U.S. Department of Health and Human Services, Administration for Children and Families, and Office of the Assistant Secretary for Planning and Evaluation; and U.S. Department of Education, Office of the Under Secretary and Office of Vocational and Adult Education.

NOTES:  A two-talied t-test was applied to differences between outcomes for the program and control groups.  Statistical significance levels are indicated as:  * = 10 percent; ** = 5 percent; and *** = 1 percent.

"Percentage Difference" equals 100 times "difference" divided by "control group".

 

As discussed earlier, impacts grew progressively in the three-year follow-up period; earnings impacts more than doubled between the second and third years of follow-up for the group randomly assigned through December 1993. Therefore, additional follow-up for all programs in the evaluation is required to determine the full effects of the programs and the relative success of the Detroit program.

Future reports, as part of the full seven-site evaluation, will provide up to five years of follow-up on the sample members, analyze the programs' impacts on a wider array of outcomes, and compare the programs' five-year costs with their five-year benefits.

[ Go to Contents ]

Conclusion

  • Detroit's low enforcement of the participation mandate limited the program's effectiveness in increasing employment and reducing welfare receipt. A more strongly enforced program would presumably have resulted in larger impacts.

While impacts on employment and AFDC receipt emerged in the second year, they were small compared with impacts of other programs in the NEWWS Evaluation. The low participation impacts, generated in part by the low enforcement of the program requirements, resulted in the program being only marginally successful in producing earnings gains and welfare reductions.

  • The findings for Hamtramck's early cohort reveal the potential of utilizing a program that stresses vocational training.

The Hamtramck program increased participation in training and increased the receipt of training certificates, which may have led to an increase in earnings and a slight reduction in welfare. This finding is consistent with a growing body of research on the effects of vocational training. Several programs studied in the past that emphasized occupational training instruction increased participants' earnings (although they did not always reduce welfare receipt). This suggests that welfare-to-work programs that can increase welfare recipients' participation and completion of training programs may be able to substantially increase their earnings.

  • The findings from the Fullerton-Jeffries late cohort suggest that the current Work First program can be effective, in some settings, in increasing earnings and reducing welfare receipt. As this evaluation continues to study the late cohort, more will be learned about the effects of the Work First program.

The Fullerton-Jeffries cohort that was most likely to be referred to Work First had larger earnings and lower welfare expenditures toward the end of follow-up, which suggests that the Work First program was having a positive effect on program group members' employment and welfare outcomes. This sample will be tracked for five years, and more will be learned about the effect of the Work First treatment on welfare recipients' employment, earnings, and welfare receipt.

[ Go to Contents ]

 


Footnotes

1.  See Evaluating Alternative Welfare-to-Work Approaches: Two-Year Impacts for Eleven Programs, prepared by Stephen Freedman, Daniel Friedlander, Gayle Hamilton, JoAnn Rock, Marisa Mitchell, Jodi Nudelman, Amanda Schweder, and Laura Storto, MDRC; and Impacts on Children and Families Two Years After Enrollment: Findings from the Child Outcomes Study, prepared by Sharon M. McGroder, Martha J. Zaslow, Kristin A. Moore, and Suzanne M. LeMenestrel, Child Trends. (Child Trends, as a subcontractor, is working with MDRC on detailed child outcomes.) Both of these 1999 reports were published by the U.S. Department of Health and Human Services, Administration for Children and Families and Office of the Assistant Secretary for Planning and Evaluation, and the U.S. Department of Education, Office of the Under Secretary and Office of Vocational and Adult Education.

2.  Statistical significance indicates the probability that the program actually produced the observed difference.

3.  One manager noted that the offices were operating under a hiring freeze during part of the evaluation and the MOST program was hampered by staffing shortages of up to 40 percent at a given time.

4.  In particular, the Alameda County GAIN program, the Portland NEWWS Evaluation program, Florida's Family Transition Program, the New Chance program, and the JOBSTART program.