Fixing to Change:

A Best Practices Assessment of
One-Stop Job Centers Working with Welfare Recipients

Chapter 6:
Empirical Evidence of Success

Contents of Chapter:

• Methodology
• Policy Changes and Limitations of the Data
• Kenosha County Job Center
• Tarrant County Employment Network
• Northwest Michigan JobNet
• Whatcom County NetWork Consortium
• What Clients Are Best Served by One-Stop Systems?

Most analysts would agree that real success in moving welfare recipients to self-sufficiency should be measurable in terms of employment and wage outcomes. The data we have been able to gather on the one-stop models presented in this study do indicate that for the most part, these models have been at least partially successful.  However, these data also highlight some of the gaps and limitations in the One-Stop concept as it has been implemented so far. To the degree that most states' welfare reform programs and all of the models examined are still evolving, this by no means is a final commentary on the potential for one-stop centers to meet the needs of families moving from welfare to self-sufficiency.

Methodology

The original design of this research called for the collection of administrative data from welfare, employment services, and unemployment insurance records to assess the characteristics of welfare participants at different stages in their contact One-Stop services. Specifically, participating One-Stop models were asked to provide three types of data:

• General Characteristics: General characteristics of the welfare population within the service area of the One-Stop center.
• Administrative Records: Administrative records for a sample of One-Stop welfare clients referred to the One-Stop during a 12-month period, generally calendar 1996. Records were expected to contain characteristics data (age, number of dependents, education, etc.) along with some measures of the length, type, and/or intensity of services provided, along with data on termination reasons.
• Unemployment Insurance Records: Unemployment insurance records for the sample of clients for the most recent eight quarters, including industry of employer, wage and salary payments, and hours.

The first and last of these requested data proved to be difficult to gather and/or acquire. In the case of the general characteristics of the welfare population, the sites examined had little or no historical data tracking these measures. Acquisition of unemployment insurance data also proved to be very difficult, due to the lack of data sharing agreements between agencies and their authority to share this information with outside researchers.

Fortunately, the administrative records for some of the sites proved to offer sufficient characteristic, service, termination, and follow-up data to permit some limited, empirical assessment of three of the One-Stop models examined -- Tarrant County, Traverse City, and Bellingham. Records from Iowa simply did not contain enough observations to include in this analysis. Administrative records from Kenosha were quite extensive with regard to the services provided, termination, and follow-up outcomes, but we were not able to match these records with the characteristics of the clients served. As a result, our analysis of Kenosha is limited to that which might be gleaned from the aggregate data contained in quarterly management reports.

Policy Changes and Limitations of the Data

The data presented in this chapter were gathered during one of the most tumultuous years for welfare policies in recent history. Many states implemented their own welfare reforms in the wake of the federal welfare policy changes. As a result, the empirical measures presented typically reflect characteristics, services and outcomes over the course of multiple policy regimes, or during times when a new policy regime was anticipated and preparations were being made for transition. Typically, the new policies meant increased work and work search requirements, which may have the effect of motivating higher levels of service use and completion, and result in higher levels of accepting initial employment. This study does not attempt to differentiate any of the measures reported according to changes in policy, recognizing that to do so would introduce too many variables and uncertainties into the sampling frame and would imply a level of analytic rigor that would be unsupported by the methodology. Rather, the basis for this approach is the assumption that for the current time, state and local welfare programs are in a state of flux, and any assessment of One-Stop employment services models should simply take this changing policy environment into account.

As a result, the conclusions that can be reached based on the data presented here may be limited. Note that the time frames, measures, and definitions vary from site to site, making each data set unique and not strictly comparable to data from other sites. Nonetheless, these data do help us understand which subgroups of welfare recipients seem to be getting better access to services or are having trouble with the One-Stop systems.

Kenosha County Job Center

Despite the very impressive shared data system for providing client services in the Kenosha County Job Center, we were unable to obtain a sample of client service records that contained even limited client characteristics. Instead, our analysis is based on a series of fourth quarter reports (cumulative) for calendar year 1996. These internal reports are produced quarterly for purposes of Center management, tracking some very basic client characteristics and outcomes for clients with dependents under age 2 years and clients with children between 2-18 years old. All cases are included, including repeat clients. These data are summarized in Table 1. In addition, these reports track the use of different services offered by the One-Stop (summarized in Chart 1).

Table 1
Kenosha County Job Center

Overall, more than half of the welfare clients served have graduated from high school or have a GED, and about 15 percent have at least one year of higher education or training. More than 40 percent of the Kenosha clients are nonwhite. Key observations about the data from the Kenosha County Job Center include:

• Age of Youngest Dependent: Fewer than 30 percent of the JOBS participants served by the Center have children younger than 2 years old, indicating that a large majority of clients are not entering the welfare system as the consequence of a pregnancy.⁽³²⁾ There are no significant demographic or educational differences between clients with children under age 2 and those with children over 2 years of age.
• Vocational Skills Training: Only about 10 percent of clients are referred to specific vocational skills training and about 6 percent are actively participating in the training. Clients with children under age 2 are much more likely to enroll once referred to training than are clients with older children.
• Work Experience Leading to Employment: Most work experience placements lead to employment, although this is not the most common route to employment. Roughly 5 percent of all clients are placed in community work experience positions leading to employment. These employment results from work experience placements tend to be slightly higher for clients with children over the age of 2 years.
• Employment: About 43 percent of all clients are employed, and almost three-quarters are employed full time (more than 29 hours per week). Employment rates, both full-time and part-time, are slightly higher for clients with children under 2 years of age.
• Wages: Wages tend to be about $.50 per hour higher for full-time verses part-time employment, with parents of younger children tending to work for slightly lower wages.
• Job Retention: Full-time unemployed clients appear to have little difficulty with short-term (30 days) job retention, but may have greater difficulty retaining employment on a long-term (180 days) basis. However, this measure may not accurately reflect actual employment patterns, since the question reflects "retention of full-time placement" employment, and some workers may have changed employers during the interim.

Chart 1
Services:  Kenosha County Job Center

* Includes multiple referrals, note that the number of participants is roughly half of the participation rate.

These data tend to confirm earlier characterizations of the Kenosha County Job Center based on focus group interviews, indicating a strong emphasis on motivation, assessment, and job search with a relatively good track record for short-term employment. Although it is difficult to tell from this data, case managers felt that the difficulty of cases was increasing -- an observation that would be supported by the fact that over 45 percent of JOBS clients had not completed their high school degree or GED.

• Job Search: Job search accounted for roughly 30 percent of the individual service elements provided to clients; nearly two-thirds of those referred to job search actually participated as shown below.
• High Use Services: Almost 50 percent of the other service elements provided were divided pretty equally among the motivational workshop, group assessment, and vocational exploration. All of these services are provided in group settings for entering groups of clients. The adult learning lab and community work experience accounted for the next highest service volumes.
• Individual Assessment and Job Search Skills Workshop: Both of these activities combined accounted for fewer than 5 percent of service volume. Individual assessment was the only service for which there was a significant differential in participation (relative to referrals) between clients with older and younger children; clients with children under age 2 had a much higher participation rate.

Tarrant County Employment Network

The Tarrant County Employment Network provided us with a sample of 72 JOBS cases drawn from entering participants during calendar year 1996. This sample included information about client characteristics, case manager contacts, long term welfare recipiency, and status at termination from the JOBS program. No consistent data were available regarding wages and hours for those employed, nor follow-up data regarding employment retention.

Overall, JOBS participants at the Employment Network tend to be older, non-white, with a high number of dependents and low levels of education. Over 62 percent have no high school diploma or GED, more than half are over age 25, and 44 percent have three or more dependents. Fewer than one-third are white, and more than half have participated in the welfare system for 36 or more of the 60 months prior to entry to the JOBS program. It is useful to note that about 36 percent of all clients included in this data analysis were currently active in the JOBS program at the end of calendar 1996.

Note that the "termination status" reported in the table below lists a series of mutually exclusive reasons/status at termination. Some were current participants, some were employed, others received assessment only, while others completed a GED or were referred to training.

These data tend to confirm earlier characterizations of the Employment Network's client population. Key observations about the Tarrant County Employment Network data include:

• Case Manager Contacts: Age and length of time in the welfare system seem to have little relationship to the number of case manager contacts.⁽³³⁾ However, African American and Native American clients appear to have significantly fewer case manager contacts on average. Clients with no dependents (a very small group) also tend to have fewer contacts, while those with three or more dependents tend to have significantly more case manager contacts. The same is true for the large majority of clients with fewer than 12 years of education or equivalent (GED). These variations in case manager contact may not be well understood - they may mean that more clients are out of compliance, are reluctant to seek help, or that fewer services are being made available to these groups.
• Long Term Welfare Participation: There were no significant racial or ethnic differences in long term welfare participation.⁽³⁴⁾ Clients with only one child tended to be less likely to have extensive prior experience with the welfare system, although those with two dependents tend to have more prior welfare experience than those clients with three or more children. Age does seem to be correlated with long term welfare participation, particularly for teens and those age 40 and older; and participants with a high school diploma or GED tend to have higher long term participation rates as well.
• Length of JOBS Participation:⁽³⁵⁾ Overall, the average length of participation was 162 days for the sample analyzed. African American and Native American clients tend to have a disproportionate share of short periods of participation in the JOBS program, a corollary observation their lower average number of case manager contacts. The number of dependents appears to be correlated with the length of JOBS participation, while clients age 40 and older tend to be over-represented in both short and long-term program participation. Education seems to have a slight positive impact on the distribution of short participation cases.
• Assessment Only: About 7 percent of all JOBS clients at the Employment Network participate in assessment only prior to termination. This proportion is double for African American clients, and about half for clients of other races/ethnicity. The data suggest higher rates of termination after assessment for clients with only one dependent and those between the ages of 25 and 39, and for those with a high school diploma, GED, or some college.
• Employment at Termination: Overall, about 17 percent of JOBS clients were employed at the time their participation in the JOBS program was terminated. Employment rates are considerably higher for African American and Native American participants, reaching nearly one third of all participants in these racial/ethnic groups. The employment rate averages 40 percent for clients with two dependents, but is much lower for clients with one, or three or more dependents. Employment rates also tend to be higher for clients in their late teens and early twenties, and for those age 40 and older. There appears to be a strong correlation between years of education and employment - clients without a high school diploma or GED (over 62 percent of all JOBS clients) have an employment rate of only 7 percent.
• Education and Training Outcomes: Over 30 percent of JOBS participants at the Employment Network terminate their JOBS participation following an increased grade level, referral to training, or completion of a GED. Younger clients, and those with two dependents, tend to have a higher education and training outcomes at termination.
• Negative Termination: Overall, roughly 17 percent of JOBS participants are terminated for negative reasons, such as the failure to comply with work participation or eligibility rules. This rate is considerably higher for clients with two dependents and those age 40 and over.

Table 2
Tarrant County Employment Network

*  Insufficient cell size for accurate reporting.

Northwest Michigan JobNet

The Northwest Michigan JobNet provided us with a sample of 150 cases entering their system between mid-1996 and mid-1997. These data included basic characteristics (excluding years of education and number of dependents), number of services and amount of supplemental cash assistance provided, termination data, and employment, wage and hours data.

In general, the welfare population served by Northwest Michigan JobNet appears to be rural, predominantly white, older than 25 years, with a slightly higher proportion of males than perhaps some other sites. Less than one-third of JobNet clients fell into the target group for potentially difficult and complex cases, including clients with a preceding welfare cases, older children, or less than 24 years of age with no high school degree or work experience.⁽³⁶⁾  Key observations based on the data provided by the Northwest Michigan JobNet include:

• Number of Services: Most JOBS participants, nearly half, received multiple services via JobNet, while roughly 20 percent received no services and the remaining one-third had used one service. Rural clients and older clients tended to receive slightly more services, as well as those clients with preceding welfare cases. The small number of African American clients in the sample received significantly fewer services than the rest -- likely a reflection of their significantly shorter length of participation
• Orientation Only: In total, about 9 percent of JOBS participants terminated their enrollment after receiving only the orientation to the program. This proportion was at least twice as high for African Americans and for both younger and older clients, especially those younger clients with no high school degree, GED, or work experience. To some degree, this measure reflects those JOBS participants who voluntarily choose not to participate for one reason or another.
• Termination for Non-Compliance: Just over 10 percent of JOBS participants were terminated for non-compliance. This measure was lower for younger and rural clients, and higher for those with preceding welfare cases, African American clients, and Hispanic clients.
• Length of Participation: Overall, the average length of participation was 81 days. The relatively small group of African Americans in the sample had much shorter periods of participation, as did a large share of clients under age 19. The group with the lowest share of short JOBS program attachments was that with ages 25 to 39 years. Hispanic clients tended to have disproportionately short and long program stays.
• Supplemental Assistance: Just over half of all JOBS participants received some form of supplemental grant assistance for auto repairs, transportation, and other incidental assistance. Clients with preceding cases and those clients age 40 or more tended to be more likely to receive this assistance. Total supplemental assistance per client generally averaged between $150 and $315 per recipient, with Hispanic clients receiving the lowest amounts ($38 on average) and older clients receiving the highest ($360 on average).

Table 3
Northwest Michigan JobNet

*  Insufficient cell size for accurate reporting.

• Employment: Roughly 45 percent of JOBS participants were employed upon termination. Clients in their early 20s and those with preceding welfare cases had slightly lower employment rates, while African American clients had employment rates that were significantly lower. Over half of the clients ages 25 to 39 were employed upon termination.
• Wages: The average wage for JOBS placements was about $5.90 per hour, with no significant variation in wage rates for different demographic groups other than the average of $5.08 per hour for clients under age 19. Note that this latter average wage is below the federal minimum wage. Younger workers have a slightly lower average hourly wage.
• Hours: Hours worked by JOBS participants average almost 32 hours per week, with little variation by demographic group, other than African American clients, whose hours average only 20 hours per week.

In interpreting these data, it is useful to remember that both African American and Hispanic categories have very few observations (4 percent of the total for each), and so it is difficult to draw strong conclusions about these observations. Nonetheless, these data do tend to confirm earlier characterizations of the site in terms of the strong local emphasis on a full time work ethic. The low percentage of clients with preceding welfare cases provides a strong contrast with Tarrant County, Texas, where over half of all clients were long term welfare recipients.

Whatcom County NetWork Consortium

The Whatcom County NetWork Consortium provided us with a sample of 97 welfare clients who had used one or more of the services of the Consortium during calendar 1996. In addition to basic client characteristics, these data included information about the number of services provided, length of participation, and employment and wages upon follow-up.

Generally, the client base for Whatcom County is better educated and has fewer dependents than some of the other sites. Only 21 percent of all clients lack a high school diploma or GED, and 55 percent have only one dependent. Slightly more than one-third had worked during the preceding 26 weeks, with the bulk of the racial/ethnic minority group comprised of Hispanic clients. Key observations based on the data provided by the Whatcom County NetWork Consortium include:

• Services: About 46 percent of NetWork Consortium clients receive one service, while the rest of the clients receive an average of between three and four services. A significantly higher proportion of clients who have worked during the past 26 weeks receive multiple services, as do clients who are age 40 and older. On the other hand, nearly three-quarters of Hispanic clients receive only one service, and over 60 percent of clients with more than a high school education receive only one service.

Table 4
Whatcom County NetWork Consortium

*  Insufficient cell size for accurate reporting.

• Length of Participation: On average, clients participate in NetWork Consortium services for a total of 212 days. While this would appear considerably longer than in other sites examined, it is important to note that the sampling frame here may not correspond to those in other sites.⁽³⁷⁾ Clients with recent work experience are disproportionately represented in the longest quartile of participation, as are clients age 40 and above, and clients with less than a high school diploma or GED. Clients with more than a high school education are over represented in the shortest quartile of participation, while clients age 40 and older, and those with recent work experience are under represented in the shortest quartile.
• Follow-Up Employment: One-third of all NetWork Consortium clients reported being employed at the 90-day follow-up contact. For those clients with recent work experience and clients age 19 and below, the employment rate was over 50 percent, while for Hispanic clients and clients with three or more dependents, the employment rate was only 20 percent. Clients with higher levels of education also report lower rates of employment.
• Follow-Up Wages: Wages for NetWork Consortium clients employed at follow-up (90 days after termination) average $8.61 per hour, considerably higher than other sites reviewed. However, this site also has substantially more hourly wage variation among different groups. Clients with post high school education and those with three or more dependents earn the highest hourly wages on average. Clients between ages 25 and 40, those with recent work experience, and white clients are all groups that earn wages that are higher than the average for all clients. Hispanic clients and clients age 18 and younger earn less than $6.00 per hour on average. Clients age 40 and over, and clients with less than a high school education average less than $8.00 per hour.

Which Clients Are Best Served by One-Stop Systems?

The preceding unique data sets seem to confirm the answers identified through focus group interviews, i.e., that welfare clients without a high school education, clients with large numbers of dependents, and clients who are older tend to have greater needs, use more services, take longer to progress through the system, and have less encouraging employment outcomes. While this information is not particularly new, the fact that One-Stop systems are tending to provide these clients with more services and assistance is encouraging, even if the employment outcomes are not entirely comparable with those of other clients. Of course, it should not go unnoted that the corollary of this pattern is also valuable information, that welfare clients with more education, fewer dependents, and young enough to be resilient in the job market are consuming relatively fewer services and are progressing faster through the system.

These data also confirm some of the difficulties facing One-Stop systems. Non-compliance seems to range from 10 percent to 17 percent. High proportions of welfare clients continue to drop out after the orientation session, and length of participation and employment rates seem to suggest that some groups of welfare recipients are either slipping through the cracks or are not responding well to the services provided. For example, Hispanic clients in the Washington State site have a high proportion of single service cases, a low average length of participation, and a low employment rate and wage rate upon follow-up. A similar pattern emerges for African American clients in Traverse City. Both of these cases involve only small samples of these subgroups and as such, limit our ability to generalize, but the presence of this type of pattern should be cause for caution.

The variation in the characteristics of the populations served across sites illustrates the need for flexibility in defining services and designing service systems that respond to unique local needs. Differences in local economies are very important in understanding the types of work skills necessary to be successful, yet differences in education levels, age distribution, number of dependents, and racial/ethnic diversity also persist, requiring careful attention to unique local design issues.

Footnotes

32.  These figures represent only a cross-section of client characteristics in time, and therefore would not capture the longitudinal effects of entrances of pregnant women who remain in the system for long periods of time. However, this observation was confirmed by Kenosha Job Center managers.

33.  One exception appears to be clients age 40 years and over, who have a lower number of case manager contacts. However, the small number of clients in this group makes it difficult to draw conclusions about this group.

34.  "Long term" is defined as 36 or more months of welfare participation during the preceeding 60 months.

35.  Note that the length of participation in the JOBS program is measured from entry until the end of calendar 1996.

36.  Note that "complex cases" were defined and targeted by the One-Stop.

37.  Note that in the case of the NetWork Consortium, the sample included clients served during calendar 1996, and was not limited to clients entering the system during that year.

Where to now?

• Executive Summary
• Introduction
• Five One-Stop Models
• Successful Models and Program Elements
• Challenges in Reaching the Welfare Population
• Occupations and Employers
• Links to Other Policy Initiatives
• Empirical Evidence of Success
• Appendix:  One-Stop Model Summaries

Home Pages:

• Human Services Policy (HSP)
• Assistant Secretary for Planning and Evaluation (ASPE)
• Department of Health and Human Services (HHS)