ABSTRACT

MDRC is evaluating Los Angeles County’s Jobs-First GAIN program, a mandatory Work First welfare-to-work program for single parents (AFDC-FGs) and members of two-parent households (AFDC-Us). First-year results indicate that the vast majority of welfare recipients who entered Jobs-First GAIN left the program within a year—many for employment. Most of those who left the program remained on cash assistance at the end of year 1. Relatively few enrollees participated in employment-related activities; most who did attended job clubs. A large proportion of enrollees received a grant reduction for noncompliance (a sanction), indicating that Jobs-First GAIN strongly enforces its participation mandates. The program produced substantial early increases in employment and earnings for AFDC-FGs and AFDC-Us, a necessary condition for future success. It also reduced first-year AFDC/TANF receipt and expenditures by moderate to large amounts. Additional follow-up for an early cohort suggests that welfare savings will continue through year 2. Jobs-First GAIN also decreased Food Stamp receipt; savings should persist in the second year of follow-up.

I. Introduction

This paper presents preliminary participation and impact findings for the Los Angeles Jobs-First GAIN program, the largest county welfare-to-work program in the nation. Consistent with the philosophy and goals of the 1996 federal welfare reform legislation that created the Temporary Assistance to Needy Families (TANF) program, Los Angeles Jobs-First GAIN emphasizes job search assistance and imparts a strong pro-work message in attempting to move thousands of AFDC/TANF recipients quickly into jobs and off the welfare rolls. These characteristics make Jobs-First GAIN a "Work First" program.

For several reasons, the study of Jobs-First GAIN is essential to understanding how welfare reform under TANF will fare:

First, as will be discussed below, Jobs-First GAIN is a relatively "mature" and well-implemented Work First program, with many components that are encouraged by TANF’s work participation requirements. Results for Los Angeles County’s program, along with those of other recently evaluated Work First programs,¹ will provide important benchmarks for administrators in California and other states as they implement and reform their programs under TANF.

Second, Los Angeles County is the most populous in the nation, with a welfare caseload larger than any state’s, except New York. To date, welfare-to-work programs have generally not performed well in the nation’s largest cities. The Jobs-First GAIN Evaluation provides an opportunity to study how well a Work First approach succeeds within an urban context and on a very large scale. If Los Angeles succeeds in moving significant numbers from welfare to work, the program will serve as a model for many other large urban areas.

Third, California has, by far, the largest welfare population and receives the largest federal outlays for welfare and Food Stamps in the nation. Thus, California is an important place for studying welfare reform.

II. Overview of the Jobs-First GAIN Evaluation

The paper is part of an ongoing evaluation of Jobs-First GAIN conducted by the Manpower Demonstration Research Corporation (MDRC) under contract with the Los Angeles County Department of Public Social Services (LA DPSS).² A previous MDRC report describes how welfare administrators transformed the program from provision of high-cost education services, shown to have helped relatively few recipients reach self-sufficiency, to adoption of a Work First strategy.³ The rest of the Evaluation studies the results of this transformation. It describes Jobs-First GAIN’s participation patterns, program costs, and program effects on employment, earnings, and receipt of AFDC/TANF and Food Stamps. The follow-up for the Evaluation begins in 1996, when the key features of Jobs-First GAIN were firmly in place, and extends through 1998, when Jobs-First GAIN was incorporated into California’s TANF program, CalWORKs. The analysis presented below covers the first year after welfare recipients entered Jobs-First GAIN (1996-97). Future reports and papers will include additional outcome measures and longer follow-up.

A key task of the Evaluation is to compare the effects of Jobs-First GAIN to those attained by three previously-evaluated welfare-to-work programs:⁴

• Los Angeles GAIN, the county’s education-focused program, operated during the late 1980s and early 90s.

Los Angeles GAIN was primarily a basic education program. Most enrollees who participated in a GAIN employment-related activity attended classes in Adult Basic Education, English as a Second Language, or, less often, GED preparation. Relatively few participated in job search, work experience, or occupational skills training. The program’s emphasis on basic education conformed to statewide requirements to provide these services to welfare recipients who had not attained a high school diploma or GED certificate, or who scored below minimum levels in reading or math tests administered at program entry, or who were not proficient in English. Nearly everyone brought into Los Angeles GAIN during the late 1980s and early 90s—8 in 10 AFDC-FGs and more than 90 percent of AFDC-Us—met at least one of these three criteria for determining need for basic education.

Among AFDC-FGs, Los Angeles GAIN reduced welfare expenditures to some extent, but did not raise earnings. The program had more positive effects for AFDC-Us, although earnings gains were still small, averaging less than $300 per year per enrollee. Other evaluations of welfare-to-work programs that emphasized basic education for single-parent enrollees with low educational attainment have shown inconsistent results. Some programs produced similar effects as Los Angeles GAIN; others achieved larger earnings increases, but often with little welfare savings. Several education-focused programs that provided occupational skills training for high school graduates and GED recipients also produced more positive results.⁵

• Riverside County GAIN, an employment-focused, mixed services program, operated in a neighboring county during the same years as Los Angeles GAIN.

Similar to Work First, the employment-focused, mixed-services approach emphasizes provision of job search services and encourages rapid entry into paid work. It also, however, makes greater use of short-term education and training as an initial activity for enrollees deemed likely to experience trouble finding work right away. Riverside GAIN is a good example of this type of program. It offered job search services to a large segment of the caseload, employed job developers to help move enrollees quickly into jobs, issued job placement goals for program staff, and encouraged enrollees to find work as soon as possible. In keeping with statewide directives, however, Riverside GAIN also offered basic education instruction to enrollees determined to need these services. As discussed below, Riverside’s GAIN program achieved unprecedented employment and earnings increases and welfare savings.

• Riverside Labor Force Attachment (LFA), the county’s Work First program, operated in the early-to-mid 1990s.

Riverside County welfare administrators, convinced that a Work First approach would increase the already strong effects of their GAIN program, created the Labor Force Attachment (LFA) program. Started in 1991, Riverside LFA superseded Riverside GAIN but retained many of its key features. In addition, Riverside LFA assigned nearly all enrollees who were eligible for participation to job club as their first activity (including those who would have been assigned to basic education by Riverside GAIN) and communicated a stronger message promoting quick entry into the job market. Recent evaluations of Riverside LFA have found that the program produced larger earnings gains and welfare savings than many education-focused programs, including Los Angeles GAIN. Its effects, however, were not as large as those attained by the previous employment-focused, mixed-services Riverside GAIN program.

As discussed in Weissman, 1997, LA DPSS administrators consulted with their counterparts in Riverside County when creating Jobs-First GAIN during the mid-1990s. Sharing Riverside’s growing commitment to the Work First approach, LA DPSS administrators adapted several features of the Riverside LFA program (some, like use of job developers and encouragement of quick entry into the job market were also present in Riverside GAIN.) Other features, like Riverside’s strong emphasis on placement goals for program staff were not incorporated into Jobs-First GAIN.

The similarities between the two county programs in welfare-to-work approach, their operation under the same state-wide welfare regulations, and their sharing of a regional labor market make comparison between Jobs-First GAIN and Riverside LFA particularly meaningful.

III. Research Design and Analysis Samples

The Evaluation utilizes a rigorous experimental design based on random assignment to experimental and control groups to estimate program effects. Between April 1 and September 11, 1996, nearly 21,000 single parents (AFDC-FGs) and members of two-parent households (AFDC-Us), required to participate in Jobs-First GAIN, were randomly assigned when they attended an orientation meeting. These welfare recipients make up the full sample.

Experimental group members have access to Jobs-First GAIN’s program services and its Work First message. They are subject to the program’s mandatory participation requirements and may incur a sanction (a reduction in their welfare grant) for non-compliance. Control group members receive AFDC/TANF payments but are precluded from receiving Jobs-First GAIN services. They may, however, seek other services in the community. Results for control group members represent the outcomes that welfare recipients would be expected to achieve in the absence of Jobs-First GAIN. Experimental-control differences in outcome measures represent the impacts of Jobs-First GAIN—i.e., the extra value associated with access to Jobs-First GAIN services and exposure to its Work First message and mandatory participation requirements.

The paper discusses first-year impacts on AFDC/TANF and Food Stamp receipt and expenditures for the full sample, as well as slightly longer-term trends for an early cohort. The early cohort includes persons who entered the sample during the first three months of random assignment and makes up a little over half of the full sample. (See Table 1.) At present, only a limited amount of earnings data are available for analysis. The paper therefore presents only two-quarter (half-year) impacts on employment and earnings for the early cohort.⁶ Participation and program status findings also cover a full year of follow-up for the full sample, but only for experimental group members.

The paper presents participation and impact results separately for AFDC-FGs and AFDC-Us. It also includes findings for two subgroups of the AFDC-FG sample: regular enrollees and early enrollees. The difference between these subgroups concerns how their members were first referred to Jobs-First GAIN prior to random assignment. During the months of sample intake, Los Angeles County lacked the funding to serve every welfare recipient who met the criteria to be considered mandatory for Jobs-First GAIN. In response, DPSS administrators created a waiting list and a system for assigning each recipient a place on the list. The "Hierarchy of Appointment Types," as this system was called, used several criteria to determine a person’s position in the queue, including her age,⁷  how long she had been receiving welfare payments, as well as her previous experience with welfare-to-work programs operated by LA DPSS.⁸

Regular enrollees are welfare recipients (including some newly approved for assistance) whose name reached the top of the waiting list when the Evaluation began. Regular enrollees received a letter from DPSS informing them that they were required to participate in Jobs-First GAIN and scheduling them for a program orientation session. Those who did not show up for orientation and failed to report good cause for their absence could incur a financial sanction, a reduction in their welfare grant. Regular enrollees who attended a program orientation during sample intake, April 1 through September 11, 1996, were randomly assigned on the day of orientation to the experimental and control groups.

LA DPSS also reserved places in Jobs-First GAIN for persons willing to enter the program before their names had reached the top of the waiting list. These recipients are called early enrollees. Most had received a mailer from DPSS describing the program and requesting (but not yet requiring) that they contact the agency to schedule an orientation. Early enrollees, like their counterparts in the regular enrollee subgroup, were randomly assigned to an experimental or control group during April through September 1996, on the day they attended their program orientation session.

The AFDC-U sample also contains both regular and early enrollees; however, the early enrollee subgroup is too small for a meaningful analysis.

A. Sample Sizes

The following sample size table breaks down the full sample and early cohort by assistance category (AFDC-FG or AFDC-U), enrollee subgroup⁹ (regular or early), and research group ("E" for experimental group or "C" for control group):

IV. Key Findings

• During the first year of follow-up, relatively few experimental group members participated in an employment-related activity: 38 percent of AFDC-FGs and 30 percent of AFDC-Us. The vast majority of those who participated in an activity, as well as of those who did not, left the program during year 1. Nearly all participants in program activities attended job club, demonstrating the Work First character of the program.
• A relatively large proportion of experimental group members (34 percent of AFDC-FGs and 29 percent of AFDC-Us) received a grant reduction for non-compliance (a sanction) at some point during year 1. This finding indicates that Job-First GAIN strongly enforces its participation mandates.
• Jobs-First GAIN produced large employment and earnings gains in the first half-year of follow-up. AFDC-FG experimental group members in the early cohort earned $407 more, on average, than their control group counterparts. For AFDC-Us in the early cohort, the earnings impact totaled $495. Employment and earnings increases were larger for single-parent early enrollees than for single-parent regular enrollees.
• In year 1, Jobs-First GAIN reduced AFDC/TANF expenditures and receipt by moderate to large amounts. Experimental group members in both assistance groups were on cash assistance about half a month less, on average, than control group members. Average welfare payments decreased by $433 (8 percent) for AFDC-FGs and $666 (10 percent) for AFDC-Us. Welfare impacts for an early cohort continued through the first half of year 2, suggesting that the program will generate savings during the rest of year 2 and possibly beyond.
• Jobs-First GAIN also produced unusually large first-year reductions in Food Stamp receipt and expenditures for both assistance groups. Savings will likely continue in year 2.

V. Background

A. Program Overview

During the mid-1990s, Los Angeles County welfare administrators transformed their original education-focused GAIN program into a Work First program, which they named Jobs-First GAIN. They made this change after concluding that the first GAIN program did not move enough recipients from welfare to work. Launched in 1988, the original Los Angeles GAIN program, in keeping with state-wide directives, placed a strong emphasis on upfront basic education. Working only with long-term welfare recipients, the program assigned most to adult basic education, GED preparation, or English as a Second Language classes, and relatively few to job search activities.

An evaluation of the GAIN program in Los Angeles and five other counties found that Los Angeles GAIN incurred large per capita costs but showed little earning gains and only modest savings in welfare expenditures.¹⁰ Los Angeles GAIN staff also voiced frustration over the program’s shortcomings: enrollees were neither completing their education activities nor finding jobs.¹¹ In contrast, Riverside County’s program, a mixed services approach with a strong employment focus, achieved unprecedented earning gains, large reductions in welfare payments, and substantial savings to government budgets.

In response, LA DPSS began in mid-1993 to restructure its GAIN program to a Work First model, adapting practices and policies from successful employment-focused programs, especially Riverside’s. DPSS completed these changes by December 1994, well before the start of the evaluation. DPSS also expanded the scale of the program, bringing several thousand more recipients per month into Jobs-First GAIN than had entered the program in previous years. By minimizing the provision of basic education and expanding job search services, which cost less, DPSS could serve more recipients at the same cost. Fieldwork by MDRC researchers indicated that DPSS was operating a mature Work First program by the start of the evaluation.¹²

Key features of Jobs-First GAIN include:

Communicating a strong Work First message. Los Angeles DPSS administrators have strongly articulated their commitment to the Work First philosophy of Jobs-First GAIN. Over the past few years, they have worked with program supervisors and staff to promote a common mission of helping enrollees quickly find jobs. GAIN staff, in turn, regularly communicate to program enrollees that finding employment is the goal of all program activities. In keeping with the Work First philosophy, GAIN case mangers refer almost all enrollees to job search activities as their first program activity, irrespective of previous educational attainment. Education and training activities, offered to some enrollees, were restructured to reinforce the goal of employment. Education providers shortened the length of their activities and focused them on workplace skills and knowledge. Further, job club leaders and other service provider staff reinforce the program’s Work First philosophy.

Providing high quality job search assistance. Jobs-First GAIN managers contracted with the Los Angeles County Office of Education (COE) to provide job search services for program enrollees. COE staff crafted an upbeat curriculum for job search activities that stresses the Work First message. COE offers job search services at its 15 Job Centers, located throughout the county. Job Club, the component in which most enrollees participate, consists of a five-day group workshop, followed by two weeks of supervised job search, that is monitored by both COE and Jobs-First GAIN staff. Facilitated by energetic and enthusiastic COE staff, workshops help enrollees identify their skills and strengths and present themselves with confidence to prospective employers. COE staff emphasize that the job club workshop is like a job itself; attending consistently and punctually in professional dress will help prepare them for the world of work. Each COE job center makes available to participants telephones, computers, printers, fax machines, and job listings. Well-trained staff help enrollees use these tools in their job search activities. Further, GAIN job developers, working in all Jobs-First GAIN offices, aggressively develop linkages to local employers, match enrollees to specific job openings, and maintain periodic contact with both enrollees and their employers to monitor their work performance.

Demonstrating that Work Pays. The Work First message is supported in a number of ways. During program orientations and in later meetings with enrollees, Jobs-First GAIN case managers emphasize that California’s rules for calculating welfare grants, known as Work Pays,¹³ make welfare recipients financially better off by working, often by hundreds of dollars per month.

Strongly enforcing the participation mandate. LA DPSS administrators and staff have demonstrated a strong commitment to using formal enforcement procedures, including grant reductions (sanctions), when enrollees fail to participate in program activities without good cause.

B. Recent Program Developments

DPSS continued to strengthen the Work First approach of Jobs-First GAIN after the start of the Evaluation. Staff began urging GAIN enrollees in agency handouts and individual and group meetings to find work immediately, because California was certain to impose time limits on welfare eligibility. They communicated this message even before passage of the federal legislation that created the TANF program. In addition, DPSS began a job search program for TANF applicants, entitled the GAIN Applicant Program (GAP). DPSS introduced GAP in 11 of the 24 DPSS cash assistance offices in April 1997.¹⁴

CalWORKs

[CalWORKs started after the follow-up period covered by this paper.]

DPSS started to phase in its TANF plan, CalWORKs, in April 1998. When fully implemented in October 1998, the plan will feature an array of components including:¹⁵

• A welfare diversion program that provides a lump sum payment to applicants—equivalent to three months of welfare for the applicant’s family size—in lieu of enrolling in TANF
• Expansion of the Jobs-First GAIN program to serve all mandatory TANF recipients, including parents of children as young as one year¹⁶
• Offering vocational training, plus ongoing case management and mentoring services for Jobs-First GAIN enrollees who find employment
• Providing substance abuse, mental health, and domestic violence services for program enrollees who need them, and continuing these services for enrollees who find employment
• Extending Transitional Child Care benefits to two years for persons who leave welfare for employment

Time-Limited Welfare

CalWORKs places time limits on eligibility for TANF for the adult recipient, but not for her dependent children. Starting on April 1, 1998, ongoing recipients may receive up to 24 months of continuous TANF benefits—18 months for new applicants for TANF. Recipients who exhaust their eligibility without finding employment must work at community service jobs to remain on their family’s TANF grant. DPSS is still planning its community service jobs component and will provide details as an addendum to their CalWORKs plan in 1999.¹⁷ Similarly, CalWORKs imposes a five-year lifetime limit on welfare eligibility for the adult recipient only. DPSS is presently considering whether to provide program services to recipients who reach this limit.¹⁸

The changeover from Jobs-First GAIN to CalWORKs will not directly affect the results of this Evaluation. The start-up of CalWORKs occurred toward the end of the Evaluation’s second year of follow-up. Further, DPSS postponed the start-up of the 18/24 month time limit "clock" for all sample members until the end of September 1998, at least two years after random assignment. Sample members may have responded, however, to the message that time-limited welfare would soon come into being.¹⁹

C. Program Environment

Los Angeles County is the most populous in the nation. It contains 9.6 million people spread over 4,000 square miles, 3.7 million of whom live in the city of Los Angeles. Complementing the county’s vast size is its diversity. Approximately 40 percent of the population is of Hispanic descent, with African-Americans and Asian-Americans each composing about 10 percent.²⁰ Though many of these minority residents live in specific communities, such as South-Central Los Angeles and East Los Angeles, many others are spread out throughout the county.

According to the most recent estimate, almost one-fourth of all county residents had incomes below the poverty level,²¹ and over one-third of the children were in poverty.²² To a greater extent than in most U.S. urban areas, the poor are spread across the county. There are pockets of poverty not only in the city of Los Angeles, but also in many of the outlying suburban communities. Further, although the overall county economy has significantly improved over the last several years, local community unemployment rates vary considerably.

The unemployment rate in Los Angeles County rose from 5.4 percent in April 1990 to a high of 10.8 percent in July 1992 during the last recession, and declined to 8.3 percent in April 1996, when the evaluation began. Since then, the rate continued to drop, falling to 6.2 percent as of April 1998 (but remaining above the national average). Employment numbers follow this trend, growing from about 4 million working residents in April 1996 to almost 4.3 million two years later, a 7 percent increase.²³

The AFDC/TANF caseload mirrored fluctuations in employment. In April 1996, the AFDC caseload included slightly more than 876,000 recipients in single-parent and two-parent households. Since then, the caseload has steadily decreased to its present level of approximately 725,000 recipients—which still accounts for about one-third of the entire California caseload and is larger than any state caseload, except for New York’s.²⁴

AFDC/TANF grant levels decreased during the course of the evaluation. For instance, the maximum aid payment (MAP) for a family of three in April 1996 amounted to $607. California lowered the MAP level to $594 in July 1996 and then to $565 in July 1997.²⁵ (A drop in welfare benefits triggers an increase in Food Stamps that partially covers the loss of income.) In general, lowering grant levels makes staying on welfare less financially attractive compared to working. It also increases the chances that recipients who begin working will earn enough to lose welfare eligibility.

California’s statewide welfare eligibility and payment regulations, known as the "Work Pays" program, however, increased the amount of money that a recipient could earn and still remain on assistance. Los Angeles County implemented Work Pays prior to the start of the Evaluation and operated it throughout the follow-up period covered by this paper.²⁶ The most important feature of Work Pays was the extension of the "$30 and 1/3" disregard of earned income to all months with earnings. (Until TANF, most states limited this disregard to the first four months.) Under this rule, the welfare department, when calculating monthly grant amounts, discounts or "disregards" the first $90 of a recipient’s monthly earnings to compensate for work-related expenses. Then the next $30 of earnings are disregarded plus one-third of the remaining total.²⁷ California also maintains the federal standard of disregarding the first $50 in child support payments and up to an additional $175 per child in monthly child care expenses after other disregards are applied.

In most states’ grant calculations before TANF, any remaining earnings after disregards were subtracted from the maximum aid payment, and the welfare recipient got the difference as her monthly welfare check. Work Pays, however, included a grant calculation procedure called "fill-the-gap budgeting" that allowed recipients to keep an extra amount of earnings without reducing their welfare grants. Under fill-the-gap budgeting, the welfare agency set a minimum monthly income level for each household size, called the "standard of need," that exceeded the maximum aid payment. The recipient’s earnings were subtracted from this higher standard of need after all other disregards were applied. This procedure worked like an earnings disregard, because it did not count additional earnings before reducing grant levels. It also increased the maximum amount a person could earn and still remain on Medicaid.

To illustrate, in April 1996, at the start of the Evaluation, the standard of need for a family of three was $730, whereas the maximum aid payment was $607, yielding a "gap" of $123. Under fill-the-gap budgeting, a mother of two who worked for pay would have a portion of her earnings disregarded, as described above, and have the remaining amount subtracted from $730, the standard of need. The first $123 of earnings (after disregards) had no effect on the recipient’s welfare grant; she still received the maximum aid payment of $607.²⁸ Earnings (after disregards) above $123 began to reduce grant levels, but it took another $607 (or $730 total) to terminate a recipient from welfare eligibility.

Jobs-First GAIN staff made a concerted effort to inform enrollees that Work Pays provisions would have made them financially better off by working than by remaining jobless and on welfare. One handout prepared by Jobs-First GAIN staff in March 1996 illustrates this point by showing the AFDC grant calculation for a mother of two earning $850 per month (full-time work at about $5 per hour) and paying $350 per month in child care expenses. The handout demonstrates that, because of Work Pays, the recipient gets $593 in AFDC—only $14 below the maximum aid payment—and $1443 in combined income from welfare and earnings.²⁹ Further, this same person could earn up to $1,740 per month and still remain eligible for a reduced welfare grant as well as Medi-Cal (Medicaid) benefits.

In addition, many recipients who leave welfare altogether qualify for Transitional Child Care and Medi-Cal benefits—although not all who qualify actually receive them. Taken together, these services ensured that virtually all clients who started working would be financially better off than if they remained unemployed and on welfare.

D. Demographic Characteristics of the Jobs-First GAIN Samples

The AFDC-FG and AFDC-U samples are drawn from throughout the metropolitan area: from central city neighborhoods like Florence and Watts to suburban neighborhoods in the San Fernando and San Gabriel Valleys. All sample members were at least 19 years old at random assignment. The AFDC-FG sample includes parents of children aged three and over, whereas AFDC-Us could include parents of younger children.³⁰

The next sections present baseline demographic characteristics of the AFDC-FG and AFDC-U samples, measured at the time sample members entered the Evaluation.

Characteristics of the AFDC-FG Sample

As shown in Table 2, many Jobs-First GAIN AFDC-FGs faced serious barriers to employment. Over half lacked a high school diploma or GED at random assignment, credentials that employers often desire and that are necessary for entry into many job training programs. Recent work experience was uncommon: 73 percent did not work for pay in the year prior to random assignment, and 62 percent remained jobless during the three years before random assignment. About 14 percent of the AFDC-FG sample received less than an eighth-grade education, and 20 percent were not proficient in English. Almost two-thirds (62 percent) received AFDC, on their own or spouse’s case, for at least two years cumulatively during their adult life. These people are considered "long-term recipients." Around 28 percent of sample members fell into the "more disadvantaged" category, which contains long-term recipients who have neither worked in the year prior to random assignment nor obtained a high school diploma or GED.

Hispanics form the largest ethnic group (45 percent) among AFDC-FG sample members. About 31 percent of the sample are Black, and 17 percent are non-Hispanic Whites. Asians make up a much smaller proportion of the sample (6 percent). Just over half of all AFDC-FGs have at least one preschool-age child (under the age of six) for whom child care would be needed.

Differences in AFDC-FG Early Enrollee Characteristics

AFDC-FG early enrollees are slightly more disadvantaged than regular enrollees in terms of educational attainment and prior employment experience, but less disadvantaged in terms of prior AFDC receipt. (See Table 2.) A smaller proportion of early enrollees than regular enrollees earned a high school diploma or GED. On average, early enrollees are slightly younger and are more likely to be parents of preschool-age children. In addition, Hispanics and Blacks make up a larger percentage of the early enrollees, whereas fewer Whites and Asians are included.

Characteristics of the AFDC-U Sample

As shown in Table 2, females make up 47 percent of the AFDC-U sample, whereas over 90 percent of AFDC-FGs are female. The percent of long-term recipients is similar among both groups. AFDC-Us are less disadvantaged than AFDC-FGs in terms of employment experience, yet slightly more disadvantaged in terms of educational attainment. Notably, nearly 20 percent of AFDC-U sample members are Asians (primarily Indochinese), and about half the sample has limited English proficiency. The AFDC-U sample also contains a larger percentage of Whites and a much smaller percentage of Blacks compared to AFDC-FGs. Further, the AFDC-U sample members have, on average, a greater number of children on their cases than the AFDC-FG sample members (2.4 versus 2.0, respectively).

VI. Participation and Program Status Findings³¹

This section presents findings on participation and program status for AFDC-FG and AFDC-U experimental group members in the full sample and early cohort. All rates presented in this section were calculated using data from LA DPSS’s automated tracking system, GEARS, and apply to the first 12 months after random assignment.³² A longer amount of follow-up is available for the early cohort: from 16 to 18 months following random assignment.³³

For each analysis sample, the following measures are discussed:

• the proportion who were ever assigned to a job search, education, or training activity
• the proportion who ever participated in an activity for at least one day³⁴
• the proportion who were deregistered (i.e., permanently excused) from the program because they had obtained a full-time job, were sanctioned, left welfare, or met other specific criteria (many persons deregistered from Jobs-First GAIN may still receive AFDC/TANF)
• the proportion who were deferred (i.e., temporarily excused) from the program because of illness, a family crisis, or other reasons
• the proportion who ever encountered the program’s formal enforcement procedures, called "conciliation," for noncompliance with participation mandates; and the proportion who ever received a sanction (a reduction in their welfare grant) for noncompliance

A few of the rates presented in this section apply only to experimental group members who ever participated in a Jobs-First GAIN activity. These individuals will be referred to as participants. A few other rates apply to experimental group members who never participated in a Jobs-First GAIN activity, or non-participants. Rates for participants and non-participants are not shown in tables.

A. Results for AFDC-FGs

Rates of Assignment and Participation

Immediately following random assignment, experimental group members met individually with Jobs-First GAIN case managers. During this "appraisal" meeting, case managers collected additional background information on the person and determined whether she could participate in program activities or whether she required a temporary or permanent exemption from participation. Those who were determined to be ready to participate received their initial assignment to a Jobs-First GAIN activity. Case managers also arranged for transportation, child care, or other assistance to support participation. Additional appraisal meetings took place when experimental group members, after leaving Jobs-First GAIN, reentered the program. It should be noted that appraisal meetings, or other meetings with Jobs-First GAIN staff, do not count as activities in the analysis below.

In the first year of follow-up, 55 percent of AFDC-FGs were assigned to a job search, education, or training activity, and 38 percent participated in an activity for at least one day. (See Table 3.) Participation levels in Jobs-First GAIN are low compared to the rates of the earlier, education-focused Los Angeles GAIN program (51 percent), as well as of the employment-focused Riverside GAIN program (60 percent). The Riverside Labor Force Attachment (LFA) program, however, engaged a similar proportion of recipients (44 percent) in work-related activities.

The low Jobs-First GAIN assignment and participation rates resulted in part from the large number of exemptions from program participation that were granted to experimental group members. Exemptions are granted to recipients who do not meet the criteria to be considered mandatory for the program, and they ordinarily result in deregistration. Immediately following random assignment, Jobs-First GAIN case managers recommended exemptions for 13 percent of experimental group members (not shown in tables). They later deregistered virtually all of these individuals from the program. ³⁵

Of the AFDC-FGs who never participated in an activity, over 90 percent were either deregistered or deferred at some point in the follow-up period (not shown in tables). In other words, almost all of the non-participants were officially excused, either temporarily or permanently, from participation requirements. Only a very small portion were expected to participate but never did, and roughly four-fifths of these recipients entered the conciliation process. These results suggest that Jobs-First GAIN staff rarely "lost track" of enrollees.

As expected for a Work First program, job search was the most common activity, and participation in education and training activities was minimal. About a third of AFDC-FGs attended job club or other job search activities, whereas only 4 percent participated in basic education.³⁶ (Assignment rates to job search and basic education activities were 50 percent and 4 percent, respectively. These results are not shown in tables.) The vast majority of participants went to job search first, reflecting Los Angeles GAIN’s Work First approach.

Rates of Deregistration

Within the first year after random assignment, 83 percent of AFDC-FGs were deregistered from Jobs-First GAIN. (See Table 3.) This rate is substantially higher than for the previous Los Angeles GAIN program, 46 percent, and similar to the Riverside GAIN rate, 79 percent. The most common reason for deregistration was employment: 40 percent of AFDC-FGs left the program because of a job. Jobs-First GAIN case managers also deregistered recipients after imposing grant sanctions. Of all AFDC-FGs, 27 percent were deregistered because of a sanction. As demonstrated in Section VII.C below, most enrollees who left the program continued to receive AFDC/TANF benefits.

Enrollment in activities tended to be short-term: 79 percent of sample members who participated in an activity were deregistered within the first year of follow-up. Over half of participants left the program due to paid work,³⁷ and 23 percent because of a sanction. (Rates for participants only are not shown in tables.)

Rates of Deferral

Under California regulations in effect prior to April 1, 1998, welfare recipients with certain barriers to participation were temporarily excused, or deferred, from Jobs-First GAIN’s participation requirements.³⁸ Common reasons for granting deferrals included medically verified illness and "severe family crisis." Originally, recipients who were employed part time were also deferred. A state law passed in late 1995, however, required these recipients to participate in Jobs-First GAIN until they found full-time employment.³⁹

Mainly as a result of this change in policy, the Jobs-First GAIN deferral rate is lower than the deferral rates of previously evaluated GAIN programs. As shown in Table 3, about one-fourth of AFDC-FGs were deferred from participation in Jobs-First GAIN during the first year of follow-up. The earlier, six-county GAIN evaluation, however, found that almost half of all single parents in Los Angeles and Riverside received deferrals (enrollees in these programs could be deferred for as little as 15 hours per week of employment).⁴⁰

Jobs-First GAIN case managers commonly deferred recipients who were attending an education or training program on their own initiative at the time of random assignment, provided that this activity could not fulfill participation requirements.⁴¹ Such activities are known in the Jobs-First GAIN office as "unapproved SITs" (self-initiated training). Case managers granted these deferrals to allow recipients to complete their unapproved SIT before having to participate in an approved activity. For single parents, an unapproved SIT was the most common reason for deferral: 7 percent of them were deferred for this reason.

Rates of Conciliation and Sanctioning

A recipient who failed to attend an assigned activity was sent a notice outlining the sanctions that may be applied if the problem continues. If she did not comply at that point, a conciliation process was initiated, providing her with another notice and opportunity to resolve the problem and avoid a sanction—a reduction in her grant amount. If the enrollee continued to fail to comply, her welfare grant was reduced.⁴² The first instance of a sanction remained in effect until the enrollee met with program staff and resolved her noncompliance with participation requirements. A second sanction lasted a minimum of three months, and subsequent sanctions at least six months, even if the enrollee resumed participation sooner.

Of all AFDC-FGs, 72 percent were involved in the conciliation process during year 1. (See Table 3.) This proportion is higher than the proportion that was ever assigned to an activity, partially because it includes conciliation for failure to show up for deferral reviews or for scheduled appraisal meetings for enrollees reassigned to the program following a deregistration.

As shown in Table 3, about one of every three single parents in Jobs-First GAIN (34 percent) received a grant sanction in the first year of follow-up. This rate is relatively high and demonstrates the program’s strong enforcement orientation. In comparison, the earlier Los Angeles GAIN and Riverside GAIN programs sanctioned a lot fewer of their enrollees: less than 10 percent. Some welfare-to-work programs operating in the 1990s, however, sanctioned comparably large proportions.⁴³

Participation Patterns for Regular Enrollees and Early Enrollees⁴⁴

As early enrollees entered Jobs-First GAIN voluntarily, it is not surprising that their rates of assignment and participation were substantially higher than those of regular enrollees. In the first year of follow-up, 70 percent of early enrollees were assigned to an activity, compared to 51 percent of regular enrollees. (See Table 3.) The overall participation rates of early and regular enrollees were 54 and 34 percent, respectively. More early enrollees than regular enrollees participated in every specific type of activity: 45 percent versus 30 percent went to job search, and 8 percent versus 3 percent attended a basic education activity.

About the same proportions in each subgroup were deregistered for employment (41 percent of early enrollees and 39 percent of regular enrollees) as well as for sanction (29 percent and 26 percent). Deferral rates were slightly lower for early enrollees (22 percent versus 27 percent), probably because early enrollees would not be inclined to enter the program if they were ill or had other problems that made it difficult to participate. The same percent of early and regular enrollees entered the conciliation process (72 percent), and a similar proportion received a grant sanction (33 and 35 percent, respectively).⁴⁵

B. Results for AFDC-Us

Rates of Assignment and Participation

As shown in Table 3, Jobs-First GAIN case managers assigned 45 percent of AFDC-Us to a program—even fewer than the percent of AFDC-FGs they assigned (55 percent). Just under a third (30 percent) of AFDC-Us participated in an activity for at least one day. This rate is slightly lower than for the earlier, education-focused Los Angeles GAIN (36.0 percent), and less than half the Riverside GAIN rate (66.0 percent).⁴⁶

As for AFDC-FGs, the low rates of assignment and participation for AFDC-Us can be partially explained by their high rate of exemptions: 18 percent were recommended for an exemption right after random assignment, and most of these recipients were eventually deregistered from the program (not shown in tables). The vast majority (91 percent, not shown in tables) of non-participants were either deregistered or deferred.

Job search was the most common activity among AFDC-Us: 28 percent went to job search, and almost all participants (93 percent) attended job search first (rates for participants only are not shown in tables). No more than two percent of AFDC-Us attended any of the education or training activities, including basic education.⁴⁷

Rates of Deregistration

During the first year of follow-up, Jobs-First GAIN case managers deregistered 85 percent of the AFDC-U sample from the program. (See Table 3.) A lot fewer AFDC-Us (just 34 percent) were deregistered from the education-focused Los Angeles GAIN program, whereas a similar proportion (80 percent) left Riverside GAIN. Almost half of all AFDC-Us were deregistered because of employment, and about one in five left the program following a grant sanction. Among those who participated in an activity, case managers deregistered 86 percent—mostly because they found full-time jobs (not shown in tables).

Rates of Deferral

At some point in the follow-up period, case managers temporarily excused 38 percent of AFDC-Us from participation in Jobs-First GAIN. (See Table 3.) In comparison, 70 percent of the Los Angeles GAIN sample and 42 percent of the Riverside GAIN sample were deferred. Of all AFDC-Us, 7 percent received deferrals for part-time employment, and 6 percent for an unapproved SIT.

Rates of Conciliation and Sanctioning

About two-thirds (68 percent) of AFDC-Us encountered the conciliation process. A similarly large proportion of AFDC-Us as AFDC-FGs received a sanction for noncompliance: 29 percent.

C. Participation Findings for the Early Cohort

For AFDC-FGs and AFDC-Us in the early cohort, the extra months of follow-up did not increase participation rates by much in comparison to the full sample. (See Appendix Table B.) Rates of deregistration, however, increased slightly.

VII. Impacts of Jobs-First GAIN

This section describes the impacts of Jobs-First GAIN on employment, earnings, and receipt of two forms of public assistance: AFDC/TANF and Food Stamps. It discusses findings for AFDC-FGs first, followed by findings for AFDC-Us. Employment and earnings results are presented for the early cohort only, and they apply to the first two quarters, or half-year, following random assignment. Cumulative, first-year public assistance results are presented for the full sample only, whereas quarterly results appear for both the full sample (which has five quarters of follow-up public assistance data) and the early cohort (which has six). Employment and earnings impacts are based on unemployment insurance (UI) earnings records from the California Employment Development Department. The data used to calculate impacts on public assistance came from the Los Angeles Department of Public Social Services automated Integrated Benefit Payment System (IBPS).

All impact estimates in this paper are regression-adjusted for differences in sample members’ baseline characteristics, prior earnings and employment, and prior AFDC and Food Stamp receipt. Regression-adjustment improves the precision of the estimates and reduces their sensitivity to pre-random assignment differences between research groups that occur by chance. Impacts for the entire AFDC-FG sample are weighted averages of the corresponding impacts for regular enrollees and early enrollees.⁴⁸ The weights compensate for differences in sampling ratios between the two subgroups and recreate the proportions of regular and early enrollees in the total AFDC-FG sample. Differences between the experimental and control groups are considered statistically significant if there is less than a ten percent probability that they could have occurred by chance. All impact estimates discussed in the text are statistically significant unless otherwise indicated.

A. Analysis Issues

Employment-focused programs (both Work First and mixed services), such as Los Angeles Jobs-First GAIN, are expected to produce large gains in employment and earnings early in the follow-up period. Their heavy reliance on job search activities should lead to quick employment for a portion of recipients who would have remained jobless without the program. In addition, employment-focused programs typically help find jobs sooner in the follow-up for recipients who would have eventually gotten work on their own.

Programs like Jobs-First GAIN may also positively affect even those recipients who do not participate in program activities. For instance, enrollees who only experience a Work First message may increase their job seeking effort. The threat of a grant sanction may also encourage non-participants to find a job.

Employment and earnings gains usually coincide with AFDC/TANF reductions; however, in states that offer generous earnings disregards, like California, employment and earnings may increase without a corresponding decrease in welfare receipt. Effects of employment and earnings gains on Food Stamp receipt are also difficult to predict. The value of a recipient’s earnings and welfare benefits helps determine how much she receives in Food Stamps, so the combination of earnings gains and welfare reductions may "cancel out" and result in little or no change to Food Stamp grants. On the other hand, a former welfare recipient may experience a decrease in (or complete loss of) Food Stamps if earnings gains are relatively large.

For this analysis, a large impact on employment is defined as a statistically significant program-control group difference in employment levels of 10 percentage points or more; whereas moderate impacts fall within the 5 to less than 10 percentage point range, and small impacts below five percentage points. Large earnings gains are considered to be in excess of $225 per quarter, equivalent to $900 per year.

Similarly, reductions in months of public assistance receipt or in total expenditures of 10 percent or more are considered large. Moderate reductions range from 5 to less than 10 percent; small reductions fall below 5 percent. A similar standard is applied to percentage point differences in levels of AFDC/TANF and Food Stamp receipt: impacts of 10 percentage points or more are considered large; 5 to less than 10 percentage point differences are considered moderate; and reductions of less than 5 percentage points are described as small.⁴⁹

The benchmarks described above are based on ranges of impact findings from previous experimental evaluations of welfare-to-work programs.

B. Impacts on Employment and Earnings for AFDC-FGs

During the first half-year of follow-up, about one-third (32 percent) of control group members in the early cohort worked for pay. (See Table 4.) Jobs-First GAIN increased the percent of recipients who were employed by a large amount: 11 percentage points. The typical experimental group member earned $1,286 over two quarters, $407 more than her counterpart in the control group.⁵⁰ In quarter 3, experimental group members earned $243 above control group levels, a substantial gain that exceeded the impact recorded in quarter 2. Los Angeles County’s falling unemployment rate (see Section V.C, "Program Environment") may have contributed to these large employment and earnings effects by increasing the chances that job search activities would lead to employment.

By boosting employment and earnings immediately, Jobs-First GAIN met the expectations for Work First programs. Additional follow-up is needed, however, to determine whether these impacts were sustained over the long term. For some previously evaluated employment-focused programs, such as Riverside GAIN, large initial gains persisted into the second year of follow-up. For others, including the Riverside LFA program, they grew much smaller as control group members found jobs on their own.⁵¹ Future MDRC reports will rely on longer-term follow-up data to demonstrate into which pattern Jobs-First GAIN falls.

C. Impacts on Public Assistance for AFDC-FGs

AFDC/TANF

During the first year of follow-up, control group members in the full sample received cash assistance for an average of 10 and a half months. (See Table 5.) At the end of year 1 (quarter 5), 83 percent of control group members were still on welfare. Total first-year AFDC/TANF payments averaged $5,796 per control group member.

Jobs-First GAIN lowered the average length of time on welfare by half a month, a moderate reduction of 5 percent relative to the control group. In quarter 5, 78 percent of program group members received a welfare check,⁵² 4 percentage points lower than the control group level. For each enrollee, the program saved $433 (8 percent) in welfare payments. Although substantial, Jobs-First GAIN’s welfare impacts are not as large as the initial employment and earnings gains, most likely because California’s generous earnings disregards make it easier for recipients to combine work and welfare. In comparison, the earlier Los Angeles GAIN program reduced AFDC expenditures by a slightly smaller amount (5 percent), whereas the Riverside GAIN and LFA programs produced somewhat larger impacts (12 percent and 11 percent, respectively).⁵³

As shown in Figure 1, quarterly impacts on AFDC/TANF payments were very similar for single parents in the full sample and the early cohort. They grew larger over time and were large and statistically significant at the end of follow-up. These findings suggest that Jobs-First GAIN will continue to produce welfare savings at least through the end of year 2.

A year and a half after random assignment (in quarter 7), 76 percent of control group members in the early cohort were still on welfare. (See Table 4.) Jobs-First GAIN reduced this proportion by 6 percentage points, a moderate amount. These findings, while positive, suggest that DPSS will face a significant challenge in moving large numbers of recipients off assistance after they complete their second year of welfare receipt.

Food Stamps

In the year following random assignment, control group members in the full sample received Food Stamps for approximately the same amount of time that they were on welfare: a little over ten months. (See Table 5.) Jobs-First GAIN reduced the length of Food Stamps receipt by as much as it reduced the length of AFDC/TANF receipt: about two weeks.

Total Food Stamp expenditures for control group members averaged $2,179 in year 1. In comparison, the typical Jobs-First GAIN enrollee received $2,005 in Food Stamps, a decrease of $174 or 8 percent (around the same size as percent reductions in AFDC payments).

For the early cohort and full sample, quarterly impacts on Food Stamp payments grew throughout most of the follow-up and remained strong and statistically significant at the end (see Tables 4 and 5). Therefore, the Food Stamp reductions should persist beyond the current follow-up period.

In quarter 7, almost three-fourths of early cohort control group members received Food Stamps—about the same proportion who received AFDC/TANF. Jobs-First GAIN lowered Food Stamp receipt by a small amount: 4 percentage points.

D. Impacts for Regular Enrollees and Early Enrollees in the AFDC-FG Sample

During the first half-year of follow-up, a similar proportion of regular and early enrollee control group members in the early cohort were employed. (See Tables 6 and 7.) Regular enrollee control group members, however, earned slightly more, on average, than their early enrollee counterparts: $922 versus $734. This difference could have resulted from early enrollees possessing less work experience than regular enrollees at random assignment.

Employment and earnings gains were large for both subgroups and largest for early enrollees.⁵⁴ Over two quarters, Jobs-First GAIN raised employment by 10 percentage points for regular enrollees and by 14 percentage points for early enrollees. Earnings increases totaled $364 (regular) nd $554 (early). In quarter 3, employment and earnings impacts for both subgroups were large and statistically signficant, so they should persist at least through the end of year 1.

In the first year of follow-up, control group levels of public assistance receipt and expenditures look nearly identical across the two subgroups, and impacts on public assistance are quite similar. (See Tables 8 and 9.) For regular and early enrollees in the full sample, the program reduced time on welfare by half a month and saved $417 (7 percent) and $491 (8 percent), respectively, in AFDC/TANF.

As shown in Figure 1, reductions in cash assistance became larger over time for both subgroups in both cohorts. As of a year and a half after random assignment, regular and early enrollees in the early cohort were 7 and 4 percentage points less likely to be on welfare than their counterparts in the control group. (See Tables 6 and 7.) The early enrollee reduction is not statistically significant, so it is unclear whether the program will reduce welfare receipt for this subgroup at the end of year 2 or beyond. Decreases in payments, however, should continue for both subgroups.

In year 1, Jobs-First GAIN reduced Food Stamp expenditures slightly less for regular enrollees, by $158 or 7 percent, than for early enrollees, who experienced a $233 or 11 percent decrease. (See Tables 8 and 9.) At the end of follow-up, experimental-control differences in Food Stamp receipt and payments were statistically significant for both subgroups in the early cohort and full sample (as shown in Tables 6 through 9).

E. Impacts on Employment and Earnings for AFDC-Us

In the first two quarters following random assignment, 35 percent of AFDC-U control group members worked for pay, a similar proportion as their AFDC-FG counterparts. (See Table 10.) They earned $1,013 on average, a little more than AFDC-FG control group members (who averaged $879).⁵⁵ Jobs-First GAIN raised employment for AFDC-Us by 10 percentage points and increased earnings by $495 above control group levels. These impacts are similar to the corresponding AFDC-FG impacts.

F. Impacts on Public Assistance for AFDC-Us

AFDC/TANF

During the first year of follow-up, AFDC-U control group members in the full sample spent an average of 10 ½ months on cash assistance, the same amount of time as their single-parent counterparts. (See Table 11.) Jobs-First GAIN lowered the length of welfare receipt by about half a month, a moderate reduction of 6 percent relative to the control group mean. At the end of year 1 (quarter 5), 77 percent of program group members versus 83 percent of control group members were on welfare.⁵⁶

AFDC/TANF expenditures totaled $6,846 per control group member in year 1, about $1,000 more than what was spent on single parents. This difference can partially be attributed to the larger average family size for AFDC-Us.⁵⁷ The program saved $666 (10 percent) in welfare payments per experimental group member, a large amount that exceeds the single-parent impact, as well as the impact for AFDC-Us in the earlier LA GAIN program (4 percent). Jobs-First GAIN did not, however, save as much as Riverside GAIN, which boasted a 17 percent reduction.⁵⁸

Figure 2 shows that dollar reductions became larger for most of the follow-up but started to grow smaller towards the end. Nonetheless, for both the full sample and early cohort, impacts on AFDC/TANF payments were still large and statistically significant in the last quarter of follow-up.

In quarter 7, 72 percent of control group members in the early cohort received a welfare check, about the same proportion as their single-parent counterparts. (See Table 10.) Jobs-First GAIN lowered this proportion by a moderate amount: 5 percentage points.

Food Stamps

During year 1, experimental group members in the full sample were on Food Stamps for an average of 10 months, about 2 ½ weeks less than control group members (a 6 percent reduction). (See Table 11.) Jobs-First GAIN lowered Food Stamp expenditures from $2,758 to $2,449, a decrease of $309 (11 percent). Percent reductions in Food Stamps were about as large as percent reductions in AFDC.

For AFDC-Us in the early cohort, Food Stamp dollar and percent reductions peaked towards the end of year 1 but remained substantial in quarter 7. (See Table 10.) During this last quarter of follow-up, 73 percent of control group members received Food Stamps (about the same amount that received AFDC). Jobs-First GAIN reduced this proportion by a moderate amount: 5 percentage points.

Notes:

1 See Hamilton, Gayle; Brock, Thomas; Farrell, Mary; Friedlander, Daniel; and Harknett, Kristen. 1997. Evaluating Two Welfare-to-Work Program Approaches: Two-Year Findings on the Labor Force Attachment and Human Capital Development Programs in Three Sites. 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. This report includes findings for Work First programs in Atlanta, Georgia; Grand Rapids, Michigan; and Riverside, California, as well as for education-focused programs in these three sites.

2 The Evaluation receives additional support from the U.S. Department of Health and Human Services and from the Ford Foundation.

3 Weissman, Evan. 1997. Changing to a Work First Strategy: Lessons from Los Angeles County’s GAIN Program for Welfare Recipients. ReWORKing Welfare: Technical Assistance for States and Localities. New York: Manpower Demonstration Research Corporation.

4 This paper simply reports results of these three programs. An upcoming report will feature more formal comparisons, controlling for differences in the background characteristics of the research samples. For an evaluation of Los Angeles and Riverside GAIN, see Riccio, James; Friedlander, Daniel; and Freedman, Stephen. 1994. GAIN: Benefits, Costs, and Three Year Impacts of a Welfare-to-Work Program. New York: Manpower Demonstration Research Corporation. Especially note Tables 4.1 and 6.1. For an evaluation of Riverside LFA, see Hamilton et al., 1997, especially Table 9.4.

5 Basic education programs for welfare recipients with low educational attainment that, like Los Angeles GAIN, did not increase earnings include the GAIN program in Alameda County and the Human Capital Development (HCD) programs in Atlanta, Georgia; Grand Rapids, Michigan; and Riverside California. Positive effects on earnings were found for the GAIN programs in Butte and Tulare Counties. Unlike Los Angeles GAIN, the Grand Rapids and Riverside HCD programs and Butte County GAIN achieved large welfare savings. Occupational skills-oriented programs for high school graduates and GED recipients that increased earnings include Alameda County GAIN and Atlanta and Grand Rapids HCD. Of these three programs, only Atlanta HCD reduced AFDC payments by a statistically significant amount. See Tables 2.8, 4.4, and 4.5 of Riccio et al., 1994, and Tables 5.1 and 11.1 of Hamilton et al., 1997.

6 Only one quarter of follow-up employment and earnings data was available for the full sample, which is not enough for a meaningful analysis.

7 Feminine pronouns will be used when referring to welfare recipients, because most adult recipients are female.

8 LA GAIN only served welfare recipients who had received AFDC for at least three years in the five years before random assignment, and it gave priority to the longest-term recipients. LA DPSS changed the Hierarchy at the start of the Jobs-First GAIN Evaluation to test program effects on a wider segment of the caseload. For instance, the agency increased the number of places in the program reserved for persons whose application for AFDC/TANF had been recently approved. DPSS also reserved more places than before for persons in a "middle level" of welfare dependency—i.e., receiving assistance for most but not all of the five years preceding random assignment. The upcoming first-year participation and impact report will discuss in greater detail the process by which regular enrollees and early enrollees entered the program.

9 Regular enrollee AFDC-FGs were randomly assigned at a ratio of 69 percent experimental group members to 31 percent control group members, whereas the early enrollee ratio was 89 percent Es to 11 percent Cs.

10 See Chapters 4, 6, and 7 of Riccio et al., 1994.

11 Weissman, 1997, p. 17.

12 Weissman, 1997, Chapters 3 and 4.

13 The Work Pays income disregards were superseded by the disregards under CalWORKs which took effect on January 1, 1998. (Other features of CalWORKs were not implemented in Los Angeles until April 1, 1998.) Nonetheless, the disregards under CalWORKs still render both part-time and full-time employment—even at minimum wage—financially beneficial to TANF recipients. See the discussion of grant calculations in section V.C, "Program Environment," for more details.

14 GAP ended on April 1, 1998, when California’s TANF plan, CalWORKs, took effect in Los Angeles. Consequently, DPSS never implemented it in the remaining 13 offices. To preserve the evaluation research design, MDRC and DPSS devised and implemented screening procedures at the program offices to identify and exclude control group members among the TANF applicants from receiving GAP services.

15 See pp. 2-6, County of Los Angeles Department of Public Social Services. 1997. "Los Angeles County DPSS CalWORKs Implementation Plan: Executive Summary" (draft).

16 See p. 21, County of Los Angeles Department of Public Social Services. 1997. "Los Angeles County DPSS CalWORKs Implementation Plan" (draft).

17 For a description of the community service plan, see Los Angeles County Department of Public Social Services. "Los Angeles County CalWORKs Plan." (h) Community Service Plan. (Los Angeles County, CA. County of Los Angeles Department of Public Social Services Website.)

18 DPSS will initiate community discussions regarding this issue in 2002. See Los Angeles County Department of Public Social Services. "Los Angeles County CalWORKs Plan." (m) Assisting Families Transitioning off Aid. (Los Angeles County, CA. County of Los Angeles Department of Public Social Services Website.)

19 Control group members will not be eligible for CalWORKs services until after September 1998, the end of follow-up for the Evaluation. The final report will examine survey data on the extent to which sample members knew about time limited welfare.

20 Los Angeles County. "County of Los Angeles Statistical Data." (Los Angeles County, CA. County of Los Angeles Website.)

21 U.S. Census Bureau. "County Estimates for People of All Ages in Poverty for California: 1993." Table A93-06, Estimated Number and Percent People of all Ages in Poverty by County: California 1993. (Washington, D.C. U.S. Census Bureau Website.)

22 U.S. Census Bureau. "County Estimates for People Under Age 18 in Poverty for California: 1993." Table D93-06, Estimated Number and Percent People Under Age 18 in Poverty by County: California 1993. (Washington, D.C. U.S. Census Bureau Website.)

23 California Employment Development Department Labor Market Information Division. "Civilian Labor Force, Employment, and Unemployment." (Sacramento, CA. California Employment Development Department Website.)

24 California Department of Social Services, Statistical Services Branch. Undated. "AFDC Family Group Unemployed Statewide Cash Grant Caseload Movement and Expenditures Report."

25 California Department of Social Services, Information Services Bureau. "Public Assistance Facts and Figures: January 1998." (Sacramento, California. California Department of Social Services Website.)

26 CalWORKs changed the earnings disregard to $225 plus one-half of the remainder and eliminated fill-the-gap budgeting. This new policy, which DPSS implemented on January 1, 1998, enables recipients who work full time to keep more of their welfare grant than they would have under Work Pays, but makes part-time workers slightly worse off. For more information, see Legislative Analyst’s Office. 1998. "CalWORKs Welfare Reform: Major Provisions and Issues." (Legislative Analyst’s Office Website.)

27 Weissman, 1997, p. 47.

28 A person earning less than $123 did not receive more than the maximum aid payment of $607.

29 LA DPSS handout, cited in Weissman, 1997, p. 48. The recipient has $1093 in net income if child care costs are subtracted. Following the July 1997 decrease in the maximum grant amount (and both before and after CalWORKs was implemented), the same person would receive $1415 in combined income.

30 During sample intake, a single parent with a child under three was not Jobs-First GAIN mandatory. According to Table 2, however, the GAIN Employment Activity and Reporting System (GEARS) indicated that 9 percent of the AFDC-FG sample had a child under three at the time of random assignment. The cause of this inconsistency is unclear.

31 This section presents comparisons of results for Los Angeles Jobs-First GAIN to those for Los Angeles GAIN, as operated from 1989 to 1991; Riverside GAIN, as operated from 1988 to early 1991; and the Riverside LFA program, as operated from late 1991 to 1994. These comparisons do not control for differences in the background characteristics of the research samples, nor do they standardize the time periods to which the results apply: rates for Los Angeles Jobs-First GAIN apply to the first 12 months following random assignment, as opposed to 11 months for Los Angeles and Riverside GAIN rates and two years for Riverside LFA rates. No formal tests of statistical significance were performed on differences in findings across programs. For GAIN rates of participation, deregistration, and deferral, see Tables 2.1 and 2.3 of Riccio, James; and Friedlander, Daniel. 1992. GAIN: Program Strategies, Participation Patterns, and First-Year Impacts in Six Counties. New York: Manpower Demonstration Research Corporation. For GAIN sanction rates, see Table 2.12 of Riccio et al., 1994. For the Riverside LFA participation rate, see Table 5.1 of Hamilton et al., 1997.

32 More specifically, the rates apply to twelve 30-day intervals (360 days), beginning with the day of random assignment.

33 18 months for those assigned in April 1996, 17 months for those assigned in May, and 16 months for those assigned in June.

34 The participation rates presented in this section include program-referred, as well as approved self-initiated, activities.

35 While Income Maintenance (IM) workers attempted to screen out non-mandatory recipients before making referrals to Jobs-First GAIN, a recipient’s status could change from mandatory to non-mandatory between her meeting with an IM worker and her appraisal with a Jobs-First GAIN case manager. To avoid costly modifications to GEARS, Jobs-First GAIN administrators decided that exemptions would be granted after random assignment, rather than before. Jobs-First GAIN case managers could only recommend that an IM worker exempt a recipient; they could not grant exemptions themselves. Once the IM worker granted the exemption, the Jobs-First GAIN worker could then deregister the recipient. It is certain that the control group contains a similar proportion that no longer met Jobs-First GAIN mandatory criteria at the time of random assignment. These persons could not be identified, however, because control group members did not attend appraisal meetings with Jobs-First GAIN staff.

36 Basic education activities include English as a Second Language (ESL), Adult Basic Education (ABE), GED preparation, and high school.

37 These participants could have obtained employment after attending a program activity; some could have reported employment they already had.

38 Under CalWORKs, case managers technically can no longer grant deferrals. Short-term barriers to participation are now recorded as good causes for non-compliance. Long-term barriers, lasting 30 days or more, lead to an exemption from participation requirements and, subsequently, to deregistration from Jobs-First GAIN.

39 Weissman, 1997, p. 67.

40 Table 3 shows that Jobs-First GAIN case managers granted deferrals to a small portion of recipients because of employment. It is unclear why exceptions to the rule of no deferrals for part-time employment were made.

41 The participation rates presented above do not include unapproved SITs, so the actual rate of overall participation was slightly higher than 38 percent. Case managers did not track participation in unapproved SITs, except during scheduled deferral review meetings with enrollees.

42 Weissman, 1997, p. 66.

43 The Grand Rapids LFA, Columbus Integrated, and Columbus Traditional programs imposed grant sanctions on about 40 percent of their enrollees. See Table 5.3 of Hamilton et al., 1997, and Table 4 of Brock, Thomas; and Harknett, Kristen. 1998. Welfare-to-Work Case Management: A Comparison of Two Models. New York: Manpower Demonstration Research Corporation.

44 Differences between regular enrollees and early enrollees were not tested for statistical significance. Significance tests will be performed for future reports.

45 One of the reasons early enrollees received as many sanctions for noncompliance as regular enrollees despite having volunteered to participate may have been because they were not aware of what they were volunteering for. Welfare recipients often expected that Jobs-First GAIN’s primary purpose was to provide them with education and training (see Weissman, 1997, p. 42). Also, some early enrollees may have signed up for the program with the hope of obtaining child care assistance while continuing self-initiated education or training activities.

46 Participation findings for AFDC-Us in the Riverside LFA program are not available at this time.

47 Case managers assigned 43 percent of AFDC-Us to job search activities and just 2 percent to basic education. These results are not shown in tables.

48 The AFDC-FG impact equals the regular enrollee impact times the proportion of regular enrollees in the AFDC-FG sample plus the early enrollee impact times the proportion of early enrollees in the AFDC-FG sample.

49 To make comparisons to results of other programs more meaningful, reductions in public assistance dollars or month of receipt should be converted to a uniform measure that is less sensitive to site variations in maximum grant levels or in sample member characteristics. One such measure, the percentage change in public assistance dollars or months of receipt (a program’s impact divided by the control group mean), will be presented throughout this section.

50 Average earnings include zero dollar values for jobless sample members.

51 The Grand Rapids Labor Force Attachment (LFA) program is another example of this phenomenon. See Chapter 9 of Hamilton et al., 1997.

52 A comparison of this proportion to the proportion ever deregistered from Jobs-First GAIN in year 1, 83 percent (see Table 3), indicates that most single parents who left the program remained on welfare.

53 See Table 4.1 of Riccio et al., 1994; and Table 9.4 of Hamilton et al., 1997.

54 Differences between regular enrollees and early enrollees were not tested for statistical significance. Significance tests will be performed for future reports.

55 Average earnings include zero dollar values for jobless sample members.

56 As shown in Table 3, 85 percent of AFDC-U program group members were deregistered from Jobs-First GAIN in year 1. Therefore, most who exited the program remained on welfare.

57 Not only do AFDC-Us have a second parent on assistance, but their average number of children is greater (see Table 2).

58 See Table 6.1 of Riccio et al., 1994. Impacts for AFDC-Us in the Riverside LFA program are not available at this time.

MDRC is conducting the Los Angeles Jobs-First GAIN Evaluation under a contract with the Los Angeles Department of Public Social Services. The U.S. Department of Health and Human Services and the Ford Foundation are providing additional funding for the Evaluation.

Dissemination of MDRC's work is also supported by MDRC's Public Policy Outreach Funders: the Ford Foundation, the Ambrose Monell Foundation, the Alcoa Foundation, and the James Irvine Foundation.