This document
summarizes the two-year findings from a large-scale, rigorous
evaluation of Jobs-First GAIN, a strongly employment-focused
mandatory welfare-to-work program. The Los Angeles County
Department of Public Social Services (DPSS) operated Jobs-First
GAIN from January 1995 through March 1998. The evaluation,
conducted by the Manpower Demonstration Research Corporation
(MDRC), has been jointly funded by DPSS, the U.S. Department
of Health and Human Services, and the Ford Foundation. Los
Angeles operates the largest county welfare program in the
nation, serving more recipients than all states except New
York and California. The size and diversity of Los Angeles
County’s population mean that any success achieved by
Jobs-First GAIN will have broad significance.
In a report on the Jobs-First GAIN Evaluation after one
year of follow-up (year 1), it was concluded that Jobs-First
GAIN produced substantial increases in employment and earnings
and reductions in welfare expenditures relative to what welfare
recipients would have achieved had they not entered the program.
Furthermore, the program produced positive results for many
different types of welfare recipients. The report summarized
here examines whether Jobs-First GAIN sustained these effects
through a second year of follow-up (year 2). It then assesses
the program’s two-year effects on a wide range of additional
outcomes, including (1) employment stability and wage growth,
(2) income and self-sufficiency, (3) medical coverage, (4)
child care use, (5) household structure, (6) food insecurity,
and (7) children’s academic and behavioral adjustment
and safety. The report also examines Jobs-First GAIN’s
cost-effectiveness.
Jobs-First GAIN anticipated
the philosophy and goals of the federal Personal Responsibility
and Work Opportunity Reconciliation Act (PRWORA) of 1996,
which replaced Aid to Families with Dependent Children (AFDC),
the nation’s largest cash welfare program, with block
grants to the states called Temporary Assistance for Needy
Families (TANF). (This type of cash assistance is referred
to here as AFDC/TANF or welfare.) Jobs-First GAIN emphasized job search assistance and imparted
a strong pro-work message in an effort to move thousands of
welfare recipients quickly into jobs. Its message and emphasis
place Jobs-First GAIN in the category of Work First programs,
the approach strongly encouraged by PRWORA and followed by
most current state and local welfare-to-work programs. Most
features of Jobs-First GAIN continued under Los Angeles County’s
TANF program, California Work Opportunity and Responsibility
to Kids (CalWORKs), which replaced Jobs-First GAIN in April
1998.
The similarities between
Jobs-First GAIN and its successor, CalWORKs, make the findings
of this evaluation especially useful to practitioners and
researchers who need reliable information on the effects of
TANF-era welfare-to-work initiatives in large urban settings.
The differences between Jobs-First GAIN and CalWORKs are also
important. Like many other TANF-era programs, CalWORKs modified
the Jobs-First GAIN program model by adding time limits on
welfare eligibility (although only for adult recipients),
somewhat stronger financial incentives to work, extended transitional
benefits, post-employment services aimed at increasing job
retention and advancement, and special services for victims
of domestic violence and people with mental health or substance
abuse problems. The Jobs-First GAIN Evaluation therefore tests
the effects of CalWORKs’ primary pre-employment strategy
(and that of many other TANF-era programs), but without time
limits and post-employment and special services and with smaller
financial incentives to keep working. Jobs-First GAIN’s
effects on employment, earnings, welfare dependency, and income
will serve as a benchmark for gauging the effects of CalWORKs’
(and other TANF programs’) more comprehensive approach
to promoting self-sufficiency.
Jobs-First GAIN, which
encouraged welfare recipients to start working as
soon as possible, replaced Los Angeles GAIN, the county’s
previous GAIN (Greater Avenues for Independence) program,
which encouraged welfare recipients to return to school to
improve their basic skills. Launched in 1988, Los Angeles
GAIN assigned most
of its enrollees, all of whom were long-term welfare recipients,
to adult basic education (remedial English and math), General
Educational Development (GED) test preparation, or English
as a Second Language classes. Relatively few enrollees were
assigned to job search activities. Evidence from several sources,
including an evaluation of the program by MDRC, convinced
DPSS’s top administrators that Los Angeles GAIN’s
basic education approach, despite being costly, helped relatively
few people find employment. The administrators resolved that
adopting a Work First approach would benefit welfare recipients
financially and save taxpayer dollars. The Jobs-First GAIN
Evaluation tests this premise by comparing Jobs-First
GAIN’s effects with those of the county’s earlier
GAIN program. The findings from this comparison are important because
administrators in many other localities made similar
changes to their welfare-to-work programs during the 1990s.
Central
to the evaluation is an experimental design based on random
assignment. From April 1 through September 11, 1996, nearly
21,000 single parents (AFDC-FGs, or Family Group cases) and
members of two-parent households (AFDC-Us, or Unemployed Parent
cases) who showed up at a Jobs-First GAIN office for their
scheduled orientation were randomly assigned either to the
experimental group or to the control
group. Experimental group members had access to Jobs-First
GAIN’s program services and exposure to its Work First
message. They were also subject to the program’s mandatory
participation requirements and could incur a sanction —
a reduction in their welfare grant — for noncompliance.
Control group members were precluded from receiving Jobs-First
GAIN services until October 1998, the end of the follow-up
period for the evaluation, but remained eligible to receive
welfare and Food Stamps. Control group members could also
seek out other services in the community and receive child
care assistance from DPSS for employment-related programs
in which they enrolled on their own initiative. Finally, both
experimental and control group members were covered by California’s
Work Pays rules for calculating welfare grants (described
in Section II). Work Pays allowed most welfare recipients
who found a job to continue receiving welfare benefits and
to retain their eligibility for Medicaid.
I.
Overview of the Findings
·
Los Angeles County successfully transformed its previous,
basic-education-focused welfare-to-work program into a Work
First program. This change was accomplished without a major
reorganization of the county’s welfare agency and before
passage of federal welfare legislation. The
implementation of Jobs-First GAIN succeeded for several reasons.
Most importantly, top administrators within DPSS were committed
to refashioning the welfare-to-work program’s services
and message into a Work First model and had the administrative
authority to carry out these changes. In addition, many of
DPSS’s supervisors and staff supported and participated
in this process. Further, DPSS developed strong partnerships
with outside service providers, in particular, with the Los
Angeles County Office of Education, which designed and operated
Jobs-First GAIN’s motivational orientation sessions and
job clubs.
A. Impacts for Single Parents
(AFDC-FGs)
·
Jobs-First
GAIN led to substantial two-year increases in employment (that
is, in the proportion of people ever employed in the two years
of follow-up) and in earnings. During
the two-year follow-up period, Jobs-First GAIN increased employment
by 10 percentage points and earnings by an average of $1,627
(26 percent) relative to control group levels. These increases
compare favorably with those achieved by several previously
evaluated Work First programs and are particularly impressive
for a welfare-to-work program in a large urban area.
·
The program produced modest reductions in welfare and
Food Stamp receipt (that is, the proportion of people receiving
each of these benefits) and large reductions in welfare and
Food Stamp payments (that is, actual expenditures for each
type of assistance). At the end of year 2, welfare receipt in the experimental
group was 62 percent, 5 percentage points below the control
group level. Jobs-First GAIN reduced the average two-year
welfare outlay by $972 per experimental group member, or 10
percent, relative to the control group average. Jobs-First
GAIN produced similar reductions in Food Stamp receipt and
payments.
·
Jobs-First GAIN produced a small net increase in total
income in year 2; the results appear more positive for the
last month of the year. Although
experimental group members earned more and received more in
Earned Income Tax Credit (EITC) payments than control group
members, these gains in income were almost exactly counterbalanced
by reductions in income from lower welfare and Food Stamp
payments and by higher payroll taxes. As a result, Jobs-First
GAIN produced only a small (2 percent) increase in income
from these sources in year 2. An estimate of monthly income
at the end of year 2 based on survey data, however, showed
a larger increase of about 9 percent.
·
The program did not affect whether people had medical
coverage, but did produce a shift from public to private insurers.
About 90 percent of people in the control
group reported receiving medical coverage for themselves and
their children from Medi-Cal (California’s Medicaid program),
from their employer, or from another source at the end of
year 2. Experimental group members reported similar levels
of coverage, but a larger proportion of them relied on medical insurance from employers.
·
There were few statistically significant
impacts on indicators of health and well-being. About the same percentage of experimental
and control group members reported having a health, emotional,
or family problem that made it difficult for them to work.
Similarly, no statistically significant differences were found
in reported housing status or neighborhood quality and safety.
For reasons that are
unclear, 12 percent of experimental
group members, compared with only 6 percent of control group
members, reported that they had not eaten for a whole day
at some point during the follow-up period because they lacked
money to feed everyone in their family.
·
Jobs-First GAIN increased the use of child care and the
incidence of child care problems that affected employment.
Jobs-First GAIN’s large effects on full-time employment
resulted in an equally large (13 percentage point) increase
in the use of child care. Roughly half of the increase was
for structured child care arrangements that required payment
for services, and the other half was for less formal, unpaid
arrangements. Nearly all sample members who used paid care
covered the expenses out of pocket; very few reported submitting
records of expenses to DPSS for reimbursement or receiving
transitional child care or other government subsidies. Much
of the increased use of child care while parents were working
was of child care that was considered unreliable — that
is, that caused experimental group members to miss or be late
for work at least once in a typical month of employment. This
problem was most common among families with preschool-aged
children.
·
The program had no systematic effects on the child outcomes
examined. Jobs-First
GAIN produced little or no change in the selected aspects
of children’s academic achievement and schooling, behavioral
and emotional adjustment, and safety that were studied. Separate
analyses for boys and girls under 18 and for school-aged children
(boys and girls together) showed similar results. Jobs-First
GAIN increased the incidence of some academic and behavioral
problems among a small group of preschool-aged children. Only
tentative conclusions about child outcomes can be drawn from
the evaluation, however, because tests of school readiness
and cognitive development commonly used in research on young
children were not included.
·
Jobs-First GAIN achieved larger employment and earnings
gains than the county’s previous, basic-education-focused
program. A comparison of impacts for welfare recipients in Jobs-First
GAIN with those for recipients with similar background characteristics
in Los Angeles GAIN showed Jobs-First GAIN to have increased
two-year earnings by more than $1,700, compared with a two-year
impact of $200 for the earlier GAIN program. Jobs-First GAIN
and its predecessor reduced average welfare expenditures by
a similar amount, however: about $1,000 per experimental group
member.
·
Many different types of welfare recipients benefited
from Jobs-First GAIN. Such consistency in findings is unusual
and impressive. The program increased earnings and reduced welfare payments
for recipients in the central city and outer regions of Los
Angeles County, for different racial/ethnic groups (including
single parents with a limited command of English), and for
recipients with either many or few serious barriers to employment.
·
Jobs-First GAIN’s costs were more than offset by
savings in welfare payments and other types of assistance. Like other Work First programs previously evaluated
by MDRC, Jobs-First GAIN led to only modest increases in expenditures
for employment-related services. Estimated over the five-year
period starting at random assignment, and based on both observed
effects (in years 1-2) and projected effects (in years 3-5),
the cost of the program is expected to be more than compensated
for by savings in welfare, Food Stamps, and Medi-Cal and associated
administrative costs and by small increases in tax revenues.
B. Impacts
for Two-Parent Families (AFDC-Us)
·
The program’s two-year impacts on earnings and welfare
expenditures were somewhat larger for members of two-parent
families than for single parents.
The program boosted employment levels for members of two-parent
families by 10 percentage points and increased their two-year
earnings by an average of $2,050, or 31 percent, relative
to control group levels. Jobs-First GAIN reduced the average
two-year welfare outlay by $1,429, or 12 percent, and produced
similar reductions in Food Stamp payments.
·
The program positively affected many subgroups of two-parent
families, but not as consistently as it did single-parent
subgroups. Jobs-First
GAIN increased two-year earnings for both men and women in
two-parent families, although the average earnings gain for
men ($2,645) was nearly twice as large as that for women ($1,486).
Whereas non-Hispanic whites did not experience a two-year
earnings increase, Hispanics achieved an unusually large increase
of $3,824, and Asians experienced a moderate gain of $1,429.
The latter two findings are impressive because half of Hispanics
and almost three-fourths of Asians lacked English proficiency
at random assignment. Earnings increases were also larger
for sample members who entered the program without a high
school diploma or GED certificate than for graduates.
II.
Key Features of Jobs-First GAIN
In response to the passage of the federal welfare reform
legislation (PRWORA) in 1996, most states and localities are
implementing some kind of Work First approach, which entails offering job search assistance as a primary service (possibly
followed by work-focused education and training) and encouraging
welfare recipients to start working as soon as possible.
Los Angeles’s version — Jobs-First GAIN, which was
put in place prior to the federal law — had a number
of features that together represent a strong commitment to
a Work First philosophy. As noted above, most of these features
have continued under CalWORKs (see Table
1).
·
Communicating a strong Work First message. DPSS administrators stated clearly that the goal of Jobs-First
GAIN was to move people into employment as rapidly as possible.
This philosophy was communicated to program enrollees through
written handouts and group presentations and in individual
meetings with program staff.
·
Warning enrollees that time-limited welfare is coming
and urging them to get a job right away to preserve their
eligibility for assistance.
Even before the passage of PRWORA in August 1996, Jobs-First
GAIN staff were informing new enrollees that the federal and
state governments would limit welfare eligibility, possibly
to two years, and were encouraging them to find work in order
to avoid the expected cuts in welfare. As one agency flier
put it:
Everyone will be expected to work. These
changes could occur as early as 1996. It is critical that
you prepare now for these social changes. Work experience
is the best training. Remember: “WORK IS IN, WELFARE
IS OUT.”
This message
was repeated during program activities such as job club and
in individual meetings with program staff.
·
Operating an unusually intensive program orientation. All new enrollees attended a six-hour
group orientation session, followed by a one-on-one appraisal
meeting with a case manager during their first day in the
program. In contrast, most other welfare-to-work programs,
including some that share the Work First philosophy, run
much shorter orientations. Further, whereas in these other
programs staff use most of the orientation to collect background
information on new enrollees and assign them to their first
employment-related activity,
Jobs-First GAIN staff devoted most of the orientation to communicating
Jobs-First GAIN’s message to new enrollees and increasing
their self-esteem — particularly with regard to their
ability to find work.
·
Providing high-quality job search assistance. Well-trained staff from the Los Angeles County Office
of Education ran job search services at 15 Job Centers around
the county, and — along with program staff — monitored
participants’ progress. Jobs-First GAIN’s job clubs
provided instruction in many of the skills needed to obtain
employment, including finding job openings, writing a résumé
and filling out a job application, and being interviewed.
Job club participants then conducted up to two weeks of supervised
job search with the aid of agency phone banks, job listings,
and program staff. These characteristics are typical of job
clubs in many other welfare-to-work programs. Jobs-First GAIN’s
job clubs, however, also featured a strong motivational
component. Their message and specially developed
curriculum were upbeat, stressing how work
can lift self-esteem and how a low-paying first job can lead
to a better one in the future. In addition, Jobs-First GAIN
staff aggressively developed relationships with local employers
and matched enrollees to specific job openings. These job
development efforts went well beyond what is traditionally
offered in job search activities.
Jobs-First
GAIN offered short-term basic education and vocational training
classes, but assigned few enrollees to these activities. The
program also made limited use of unpaid work experience jobs.
·
Using job development activities to support enrollees’
job search efforts. Each
Jobs-First GAIN office had job developers who cultivated relationships
with local employers and created lists of job openings. Job
developers then tried to match enrollees to available job
openings, based on enrollees’ prior experience and interests.
Job developers began working with enrollees during orientation
and appraisal and continued assisting their job search efforts
during job club and other stages of program participation.
Job developers also arranged and hosted job fairs for enrollees
— small, weekly fairs with one or two employers and larger,
quarterly fairs with numerous employers. One program office
even experimented with having its job developers work on a
one-on-one basis with program enrollees who had received a
financial sanction for not complying with program requirements.
·
Demonstrating
that work pays. As noted above, California’s Work Pays rules for calculating
welfare grants allowed many recipients to combine work and
welfare. Using waivers granted by the U.S. Department of Health
and Human Services, Work Pays increased, above national standards,
the amount of earnings that the welfare department “disregarded”
(did not count) when calculating welfare grants. As a result,
most welfare recipients who combined work and welfare could
receive hundreds of dollars per month in income above what
they would have received in welfare alone. Work Pays became
part of Jobs-First GAIN’s strategy for convincing people to
find employment as quickly as possible even if available jobs
paid little. Jobs-First GAIN staff made a concerted effort
to explain the financial benefits of Work Pays to new enrollees
by walking them through several examples of grant calculations
during motivational sessions at program orientation and by
repeating this message during job club and other employment-related
activities.
Both
experimental and control group members were covered by California’s
Work Pays rules. Control group members may have been motivated
by these rules to look for work on their own initiative or
to increase their hours of work. However, it is likely that
fewer control than experimental group members knew about Work
Pays because they did not attend orientation or job club and
did not meet with Jobs-First GAIN case managers.
·
Running a relatively tough, enforcement-oriented program.
Jobs-First GAIN
case managers made frequent use of the program’s formal enforcement
procedures, including threats to reduce welfare grants, to
encourage enrollees to participate in program activities or
show good cause why they could not. As discussed in the full
report, the vast majority of program enrollees received at
least one warning that they were out of compliance with program
rules. About 30 percent of single parents and a quarter of
adults in two-parent families incurred a sanction for noncompliance;
a sanction entailed dropping the recipient (but not the recipient’s
children) from the grant. Program administrators intended
this high-enforcement case management approach and the strong
pro-employment message to complement the program’s high-quality,
motivational job clubs. Together, these components of Jobs-First
GAIN encouraged enrollees to find work quickly and discouraged
them from spending a long time in the program.
III.
The Research Sample and Program Environment
The research sample for the Jobs-First GAIN Evaluation
includes 15,683 single parents (AFDC-FGs) and 5,048 members
of two-parent families (AFDC-Us). During the evaluation, DPSS
followed the eligibility criteria written into the federal
Family Support Act (FSA) of 1988, which preceded PRWORA, in
determining which recipients had to enroll in Jobs-First GAIN.
According to the FSA, any single-parent welfare recipient
whose youngest child was 3 or over and who did not meet certain
exemption criteria was mandated to participate in a welfare-to-work
program. Grounds for exemption included having a disabling
illness, being employed full time (30 hours or more per week),
living in a remote area that made program activities inaccessible,
or being in at least the second trimester of pregnancy. These
criteria also pertained to AFDC-U welfare recipients, except
that AFDC-U parents of children under 3 were also required
to enroll in a welfare-to-work program. Exercising an option
given to states and localities under the FSA, DPSS also required
both parents on an AFDC-U case to enroll in Jobs-First GAIN.
Because
DPSS did not have the resources to serve all welfare recipients
mandated to participate, prior to the evaluation it reserved
nearly all places in Jobs-First GAIN for people identified
by the FSA as being at the greatest risk of remaining on welfare
for a long time. DPSS gave highest priority to serving those
who had received welfare continuously for at least three years.
To enable the evaluation to determine the effect of the Jobs-First
GAIN approach on a broad cross section of the welfare caseload
and on different types of welfare recipients, DPSS administrators
later implemented a complex procedure for selecting new enrollees.
The resulting sample, which included nearly everyone who came
into the program between April and early September 1996, appears
to reflect, in very broad terms, the diversity of the Jobs-First
GAIN-mandatory caseload. The sample differed from the full
Jobs-First GAIN-mandatory caseload principally in having a
substantially smaller percentage of people going through a
very long spell — of at least five years — on welfare and
in excluding teen parents and a few other groups.
The sample includes welfare recipients who inhabit the
inner-city neighborhoods of Los Angeles as well as recipients
in the outlying suburbs. The sample is also diverse with respect
to race and ethnicity, age, family size, and several indicators
of relative disadvantage in the labor market. Among AFDC-FG
sample members, Hispanics formed the largest ethnic group
(45 percent), followed by African-Americans (about 31 percent),
non-Hispanic whites (17 percent), and Asians (6 percent).
A little more than half of the AFDC-FGs had at least one preschool-aged
child (under the age of 6), for whom child care would be needed.
Nearly 20 percent of AFDC-U sample members were Asians (primarily
Vietnamese and Cambodian immigrants and refugees), and about
half had limited English proficiency. Relative to the AFDC-FG
group, the AFDC-U group included a larger percentage of non-Hispanic
whites (many of them recent immigrants from Armenia) and a
much smaller percentage of African-Americans. Further, the
AFDC-U sample members had, on average, more children than
did the AFDC-FG sample members (2.4 versus 2.0).
A large majority of AFDC-FG and AFDC-U sample members
faced one or more serious barriers to employment at the time
of random assignment: More than half of each group had not
graduated from high school or received a GED certificate;
about 60 percent had not worked for pay in the prior three
years; and about 70 percent had received welfare for at least
two years cumulatively. Other members of the research sample
faced fewer barriers to employment: About 30 percent of both
AFDC-FGs and AFDC-Us were newly approved applicants for assistance
or had received assistance for less than two years, and more
than a quarter of each group had worked for pay in the year
prior to random assignment.
A. Subgroups for Analysis
A key task of the Jobs-First GAIN Evaluation is to analyze
whether Los Angeles County’s Work First approach benefited
many types of recipients or primarily particular subgroups.
Key subgroups for analysis include:
1. Inhabitants of different geographic areas
of the county
2. Members of different racial/ethnic groups
3. People who
entered the program with a high school diploma or a GED certificate
and nongraduates
4. Short- and long-term welfare recipients
5. People with or without a recent work history
6. People with
multiple barriers to employment: no high school diploma or
GED certificate, no recent work history, and long-term welfare
receipt
7. Among AFDC-Us, men and
women (A large majority of AFDC-FGs were women.)
B.
Additional Background Information
Labor market conditions in Los Angeles County improved
during the evaluation period: Employment levels rose, and
unemployment declined. Still, the county’s unemployment rate
is higher than the national average and varies considerably
by region. For example, unemployment rates in South-Central
and East Los Angeles — communities where more than 90 percent
of the residents are either African-American or Hispanic —
still hover around more than 9 percent, 3 percentage points
above the county average.
The county’s AFDC/TANF caseload numbers followed the
trends in employment figures. In July 1996, Los Angeles County
had about 306,000 cases; two years later, the number declined
to 245,000. California reduced grant levels by nearly 7 percent
during these years (for instance, from $607 to $565 for a
family of three), although the state’s welfare grant levels
remained well above the national average.
IV.
Program Implementation and Participation
·
Jobs-First GAIN exposed all enrollees to a strong Work
First message and increased their use of job search services.
All experimental group members attended a six-hour informational
and motivational meeting at orientation during which program
staff strongly communicated the program’s Work First message.
Fewer than half of experimental group members (42 percent
of AFDC-FGs and 34 percent of AFDC-Us) subsequently participated
in a Jobs-First GAIN activity for at least one day. Nearly
all who participated in Jobs-First GAIN activities attended
job club only, and most took part in only one three-week job
club session. Jobs-First GAIN case managers assigned very
few experimental group members to education and training activities
and rarely approved requests to continue participation in
education and training activities that predated the program.
(Approval gave the enrollee credit for participating in a
Jobs-First GAIN activity and made her eligible for child care
and other assistance.) Nonetheless, about 40 percent of experimental
group members who answered the Two-Year Client Survey, which
was administered at the end of year 2 to 746 single parents
in the research sample, reported that they attended an education
or training activity on their own initiative after random
assignment. About half of these people, or about 20 percent
of experimental group members who responded to the survey,
also attended job club or another Jobs-First GAIN activity,
and the other half participated in activities only outside
Jobs-First GAIN.
In all, about 62 percent of experimental group members
participated in some type of employment-related activity in
the follow-up period other than the program orientation (see
Figure 1; AFDC-Us were not surveyed). About 44 percent of
control group members also participated in an employment-related
activity outside Jobs-First GAIN, typically vocational training
or post-secondary education. The difference in participation
rate between the experimental and control groups, 18 percentage
points, represents Jobs-First GAIN’s impact on participation
in activities, which is relatively modest. However, the program
also produced a large (31 percentage point) gain in use of
job search services, plus a small (5 percentage point) gain
in participation in unpaid work experience jobs.
·
Jobs-First GAIN case managers made extensive use of the
program’s enforcement procedures and imposed financial sanctions
for noncompliance relatively frequently.
Jobs-First GAIN staff initiated formal enforcement proceedings
for about 80 percent of both AFDC-FGs and AFDC-Us during the
two-year follow-up period. Grounds for commencing this conciliation
process (as DPSS termed it) included nonattendance at an assigned
activity or scheduled meeting with Jobs-First GAIN staff or
refusal to accept an assignment to job club.
If after repeated reminders
that participation was mandatory an enrollee did not provide
DPSS staff with an acceptable reason for nonattendance at
assigned activities, a sanction (a reduction in welfare benefits)
was imposed. About 30 percent of AFDC-FGs and 23 percent of
AFDC-Us incurred a sanction during follow-up. These rates
exceed those for several other employment-focused welfare-to-work
programs evaluated by MDRC and are much higher than those
for the Los Angeles GAIN program (5 percent).
V.
Program Costs
·
As is typical of Work First programs, the net cost of
Jobs-First GAIN was relatively low.
The full report presents a rough estimate of the gross
cost per experimental group member of providing employment-related
services, case management, and supportive service payments.
(All dollar amounts are expressed in 1998 dollars.) The gross
cost includes the cost to DPSS and outside providers of operating
Jobs-First GAIN activities, plus the cost of operating activities
that experimental group members attended outside Jobs-First
GAIN. The gross cost does not include the cost to DPSS of
authorizing and processing welfare checks. The two-year gross
cost per AFDC-FG experimental group member was about $4,300.
This average is relatively high, but reflects experimental
group members’ extensive use of education and training services
outside Jobs-First GAIN. DPSS paid about $1,800, or 40 percent,
of the gross cost, primarily to operate the program’s motivational
orientation session and job clubs and to perform case management.
A large percentage of AFDC-FG control group members attended
education and training activities on their own initiative.
The gross cost per control group member is therefore relatively
high as well, averaging nearly $2,900. The average experimental-control
difference, or the net cost of the program, was thus around
$1,400 (that is, $4,300 – $2,900) per experimental group member.
Jobs-First GAIN’s net cost is close to those of several other
Work First programs evaluated by MDRC, the net costs of which
ranged from about $1,200 to $2,500. Estimates of the gross
and net costs of Jobs-First GAIN for AFDC-Us are less precise
because MDRC did not collect information on experimental and
control group members’ participation in employment-related
activities outside Jobs-First GAIN. On the basis of a comparison
between AFDC-FG and AFDC-U experimental group members with
respect to participation patterns within Jobs-First GAIN,
it was assumed that AFDC-Us in both research groups participated
in these self-initiated activities less often than AFDC-FGs.
The full report estimates the gross cost of the program per
AFDC-U experimental group member to be about $2,500 and its
net cost to be around $1,200.
VI.Impacts for Single Mothers (AFDC-FGs)
Experimental designs based on random assignment typically
provide the most accurate and reliable findings on the effects
of welfare-to-work programs. Because people are assigned at
random to the experimental or control group, the two groups
do not differ systematically with respect to either measured
characteristics (such as length of time on welfare) or unmeasured
characteristics (such as strength of motivation to get a job).
Members of the two groups also face the same labor market
conditions. The outcomes for control group members represent
what would have happened to welfare recipients in the absence
of the program. Thus, any subsequent differences found between
the two groups can be attributed with confidence to the combination
of program services, messages, and participation mandates
that only experimental group members experienced. These differences
are known as program impacts. Unless otherwise noted, all are statistically significant,
that is, have a more than 90 percent chance of arising from
the program rather than by chance.
·
Jobs-First GAIN increased employment and earnings during
the two-year follow-up period.
About
67 percent of experimental group members worked for pay at
some point during the follow-up period, compared with 58 percent
of control group members (see Table 2).
As is typical of Work First programs, Jobs-First GAIN had
a larger impact on employment in year 1. Its impact shrank
somewhat in the following year but remained substantial, averaging
between 6 and 7 percentage points in each quarter of year
2. Employment gains will almost certainly continue into year
3.
During
the two years of follow-up, control group members earned an
average of $6,385, whereas experimental group members earned
an average of $8,012 — a gain of $1,627, or 26 percent. (Both
averages include zeros for those who did not work for pay
in year 1 or year 2.) As would be expected of a Work First
program, Jobs-First GAIN increased earnings primarily by helping
recipients who would not have found jobs on their own find
work and by helping recipients who would have eventually found
employment start working sooner. An analysis comparing only
those people in the experimental and the control groups who
found jobs shows that Jobs-First GAIN led to only small increases
in the number of quarters of employment and in average earnings
per quarter.
·
Earnings gains for AFDC-FGs grew from $759 in year 1
to $869 in year 2. The program increased full-time employment
and employment with fringe benefits at the end of year 2.
Quarterly
earnings gains reached their highest level ($237 per experimental
group member) at the end of year 2 and will almost surely
continue into year 3. Sample members’ responses to survey
questions about the characteristics of the jobs they held
at the end of year 2 underscore these positive trends. Jobs-First
GAIN produced a large (11 percentage point) gain in full-time
employment (of 30 hours per week or more) at the end of year
2. In addition, a higher percentage of employed experimental
than control group members reported receiving paid vacation
or medical coverage from their employer at the end of year
2 (see Table 3). Despite their earnings
gain, relatively few experimental group members were working
at jobs that provided fringe benefits at the end of year 2.
For instance, only 14 percent of experimental group members
(just under one-third of those who were employed) were working
at jobs with full-time hours and medical coverage, an important
indicator of successful employment.
·
Jobs-First GAIN reduced welfare and Food Stamp expenditures
and receipt of these benefits in the two-year follow-up period.
Over two years, experimental group members received an
average of $972, or 10 percent, less in welfare payments than
control group members (see Table 2).
The percentage reductions in welfare payments grew over the
course of follow-up, and will almost certainly continue into
year 3. Although most of the AFDC/TANF savings resulted from
reductions in the number of months during which people received
welfare, a substantial proportion of the savings stemmed from
lower welfare payments in months when people were still receiving
welfare. California’s Work Pays financial incentives (which
encouraged people to combine work and welfare in the short
term) and Jobs-First GAIN’s relatively high sanction rate
most likely contributed to this outcome. Two years after random
assignment, 66 percent of control group members were still
on welfare. Jobs-First GAIN reduced this proportion to 62
percent, an impact of 5 percentage points (rounded). The program
also produced two-year reductions in Food Stamp receipt and
expenditures that were similar in magnitude to the reductions
in AFDC/TANF expenditures.
·
Many experimental group members combined work and welfare.
At the end of year 2, the increase in the percentage of people
working and off welfare was small.
Figure 2 illustrates how Jobs-First
GAIN affected self-sufficiency by breaking down the experimental
and control groups into four categories based on their employment
and welfare status at the end of years 1 and 2. As shown,
Jobs-First GAIN reduced the proportion of sample members in
the most dependent group — those who were jobless and on welfare
— by 8 percentage points, from 45 percent to 37 percent, at
the end of year 2. The figure also indicates that the proportion
of experimental group members who were working and off welfare
increased over time (20 percent at the end of year 2, compared
with 11 percent at the end of year 1). Still, at the end of
year 2, most employed experimental group members were combining
work and welfare, and Jobs-First GAIN only slightly increased
the percentage of recipients employed and off cash assistance.
California’s relatively high welfare grants and Work Pays
earnings disregards helped produce these results. The earnings
of employed experimental group members reduced their welfare
grant amounts, but usually did not render them ineligible
for assistance.
·
Jobs-First GAIN led to a small increase in total income
from earnings (minus payroll taxes), estimated EITC payments,
AFDC/TANF, and Food Stamps in year 2. The impact on total
income appears to have grown over time.
Jobs-First GAIN led to the replacement of welfare dollars
with dollars from earnings. In year 2, the program increased
total yearly income from earnings (minus payroll taxes), estimated
EITC payments, AFDC/TANF, and Food Stamps by only $206, or
2 percent.
An alternative way to assess whether Jobs-First GAIN
affected income is to compare the proportions of experimental
and control group members whose income from these sources
exceeded the federal poverty threshold. (This calculation
of income differs from that used to define the poverty threshold
because it includes Food Stamps and excludes other income
sources.) The program produced mixed effects on this measure.
About 29 percent of experimental group members received income
that put them above the poverty threshold in year 2 compared
with 24 percent in the control group. AFDC/TANF and Food Stamps
remained important sources of income for many of these more
advantaged sample members. Only about 13 percent of experimental
group members overall were able to stay above the poverty
threshold on earnings alone. Jobs-First GAIN also increased
the proportion of people with very low incomes (defined as
receiving income below 50 percent of the poverty threshold)
by 3 percentage points. (None of the findings with respect
to the poverty threshold is shown.)
Another estimate of total income in the last month of
year 2, calculated from survey responses and administrative
data, shows Jobs-First GAIN’s effect to be larger and more
positive. This estimate includes child support payments, Supplemental
Security Income and disability benefits, Social Security and
pension benefits, and any other reported income. By this measure,
Jobs-First GAIN led to an average increase in total income
of $86, or 9 percent of control group members’ average income
in the last month of follow-up, which was $1,001. The program
also had a large effect on the proportion of people who received
income in the final month of follow-up that put them above
the poverty threshold, increasing this proportion by 10 percentage
points relative to the control group level of 32 percent.
·
Jobs-First GAIN did not affect medical coverage or use
of other noncash benefits.
At random assignment, all sample members received welfare
benefits and medical coverage through California’s Medi-Cal
program. Jobs-First GAIN did not affect the proportion of people with
medical coverage at the end of year 2. About 92 percent of
experimental group members in the survey reported receiving
coverage for themselves from Medi-Cal, from their employer,
or from another source — just 1 percentage point below the
control group level (the difference was not statistically
significant). Coverage levels for both respondents and their
children were similar in the two research groups and slightly
lower than adult coverage levels (around 90 percent). Despite
Jobs-First GAIN’s employment effect, the relatively high proportion
of respondents who combined work and welfare helped keep Medi-Cal
coverage levels relatively high. In addition, a higher proportion
of experimental than control group members received medical
coverage from their employers. Only about 3 percent of sample
members in each research group received Transitional Medi-Cal,
which provided one year of extended coverage to former welfare
recipients who earned too much to remain on welfare. (CalWORKs
provides two years of extended coverage.) Jobs-First GAIN
did not affect the levels of receipt of other types of noncash
assistance (aside from Food Stamps), such as federally subsidized
school lunches for children, public housing, government rent
subsidies, and energy assistance.
·
For reasons that are unclear, experimental group members
reported a higher incidence of food insecurity than control
group members.
As noted earlier, Jobs-First GAIN increased
total income by a small amount in year 2, although income
for most experimental group members remained below the poverty
threshold. Respondents to the Two-Year Client Survey provided
additional information about their problems in providing for
their families on a limited income. A little more than half
(53 percent) of experimental group respondents indicated that
they had experienced difficulty obtaining adequate and nutritious
food owing to lack of money at some point in year 2. This
level of food insecurity exceeds the national average for
U.S. households with incomes below the poverty threshold.
Furthermore, about 19 percent of experimental group members
reported experiencing a more severe type of food insecurity
that involved skipping meals, compared with 13 percent in
the control group. It is not clear why Jobs-First GAIN produced
this negative effect. Jobs-First GAIN did not have any other
statistically significant effect on indicators of health and
well-being.
VII.
Impacts on Child Care, Home Environment, and Child Outcomes
for Children of Single Mothers
This section estimates Jobs-First GAIN’s
impacts on certain indicators of children’s safety and cognitive,
social, and behavioral development. It presents results for
all children under 19, for boys and girls separately, and
for children in different age groups. Although Jobs-First
GAIN provided no special services for children, it may have
affected children indirectly through its impacts on mothers’
employment, earnings, welfare dependency, or the other outcomes
discussed above. This section also presents findings on additional
outcomes that could have affected children: (1) the use and
reliability of child care to support maternal employment (see
the full report for effects on the use of child care for other
reasons), (2) the mother’s marital status and living arrangements
at the end of year 2, and (3) the frequency with which the
mother and her children engaged in learning experiences or
played together. Other effects of Jobs-First GAIN that may
have affected children, such as changes in the mother’s self-esteem
and stress level, were not measured.
The impact estimates discussed in this section
were calculated from single parents’ responses to the Two-Year
Client Survey. They provide only a broad-brush picture of
Jobs-First GAIN’s effects on children’s well-being. In particular,
the survey did not include several measures of school readiness
and behavioral adjustment often used in research on children.
Furthermore, some of Jobs-First GAIN’s effects on children
may not have manifested themselves within the relatively short
follow-up period. Finally, the analysis does not examine whether
the program’s effects on children varied by family size or
by the mother’s age, race/ethnicity, educational attainment,
or other background characteristics.
·
Jobs-First GAIN increased the use of child
care — both care that recipients paid for out of pocket and
care that friends or family members provided free of charge.
Very few respondents in either research group reported using
subsidized child care or receiving transitional child care
benefits.
Nearly 48 percent of experimental group members
reported using child care during their current or most recent
job, an increase of 13 percentage points relative to the control
group level (Table 4). Nearly all of
the increase was in child care used while parents were working
at a full-time job. Jobs-First GAIN increased by 7 percentage
points the proportion of parents who used paid child care.
Most respondents who used this type of care paid for it out
of pocket, and Jobs- First GAIN did not affect whether payment
was made out of pocket. Only 5 percent of experi-mental group
members and 6 percent of control group members reported that
DPSS or another agency helped them pay for care. (The last
three results are not shown.) These findings are generally
consistent with anecdotal information on child care use among
welfare recipients in Los Angeles. However, given that all
welfare recipients were eligible for child care assistance
from DPSS, these results may underestimate receipt of such
assistance. Before April 1998, recipients could submit records
of out-of-pocket child care expenses to DPSS and have their
welfare check for the following month adjusted upward to compensate
them for those costs. Starting in April 1998 (and during the
period when respondents were interviewed), recipients no longer
submitted records of expenses but could arrange for DPSS to
pay child care providers directly.
The
program also increased the use of unpaid child care arrangements.
As would be expected, respondents with children under 6 at
random assignment were most likely to report having used child
care while they were working. The impacts of Jobs-First GAIN
on child care use were somewhat larger for this group than
for parents of school-aged children.
·
Jobs-First GAIN increased the proportion
of single parents who reported experiencing problems with
child care that caused them to miss or be late for work. Such
problems were most prevalent among families with preschool-aged
children.
About
27 percent of experimental group respondents reported missing
or being late for work at their current or most recent job
for one or more days in an average month because of child
care problems (Table 4). The program
led to an 11 percentage point increase in missing or being
late for work in the experimental group relative to the control
group. The effect was most pronounced for experimental group
members whose youngest child was under 6 at random assignment.
·
Jobs-First GAIN had no effect on marriage,
family composition, or amount of recreational time spent with
children.
Fifteen percent of control group respondents
reported being married and living with a spouse or living
with a partner at the end of year 2. In addition, one-fifth
reported living with extended family members. Jobs-First GAIN
had no effect on these household composition and marital status
outcomes for the whole survey sample or for families with
children in different age groups. Approximately 40 percent
of control group members stated that they played with their
children every day of the week. Jobs-First GAIN also had no
effect on this outcome, nor on the frequency with which mothers
and children engaged in learning experiences such as going
to a museum. (These results are not shown.)
·
Jobs-First GAIN had no systematic effect
on children’s outcomes overall or on outcomes for girls or
boys when separately examined. There is some evidence that
Jobs-First GAIN unfavorably affected outcomes for preschool-aged
children, although it is difficult to assess the extent of
these effects without a larger sample and more age-appropriate
measures.
Of the 10 child outcomes examined, Jobs-First
GAIN only had one statistically significant impact for children
overall, and this impact was positive: The program decreased
the proportion of children who were expelled or suspended
from school (results not shown). When child outcomes were
examined separately for girls and boys, no statistically significant
impacts were found. However, a larger proportion of preschool-aged
children in the experimental than in the control group, particularly
of those aged 4 to 5 at the time of random assignment, were
reported to have repeated a grade (most likely kindergarten
or first grade) once they entered school and or to have had
a condition that made their mothers’ going to work difficult.
Jobs-First GAIN had only one impact (out of 19 comparisons;
15 are shown in Table 5) on school-aged
children’s or adolescents’ academic achievement and schooling,
behavioral and emotional adjustment, or safety: a 6 percentage
point increase in the proportion of children aged 6 to 9 at
random assignment who attended a special class for a physical,
emotional, or mental condition.
VIII.
Cross-Program Comparisons
·
Jobs-First
GAIN’s net cost fell well below that of Los Angeles’s earlier,
education-focused GAIN program.
As
expected, the net cost of Jobs-First GAIN was much lower than
— in fact, was less than one-quarter of — the net cost of
Los Angeles GAIN. The difference in cost resulted in large
part from DPSS’s switch from high-cost basic education and
training to lower-cost job search services.
·
Jobs-First GAIN produced larger employment and earnings
increases than Los Angeles GAIN.
Table
6 shows how the two-year impacts of Jobs-First GAIN for
single parents compare with those of its basic-education-focused
predecessor. Members of the Los Angeles GAIN research sample
underwent random assignment from July 1989 through March 1990.
Unlike Jobs-First GAIN sample members, nearly all of them
had received welfare continuously for three or more years
at the time of random assignment, were jobless, and had no
children under 6. To allow for more meaningful comparisons
between the two programs, in this analysis the impacts of
Jobs-First GAIN and the county’s previous GAIN program were
estimated only for people with these characteristics. In addition,
all dollar impacts were converted into 1998 dollars.
Table 6 shows that Jobs-First GAIN
was more successful overall than Los Angeles GAIN. Jobs-First
GAIN’s two-year employment impact is 4 percentage points larger
than that of the earlier program, and its total earnings impact
is $1,516 larger. In contrast, the two programs had similar
impacts on welfare payments and receipt. These results indicate
that a Work First program can be more effective in boosting
employment and earnings than a basic-education-focused program,
even in a major metropolitan area with high unemployment and
in a welfare population that includes many recipients with
low educational attainment, limited work history, and limited
proficiency in English.
IX. Subgroup Impacts for Single Parents
·
Jobs-First GAIN benefited a broad cross section of the
welfare caseload, producing impacts for recipients with the
most as well as the fewest barriers to employment, for people
of different racial/ethnic backgrounds and different levels
of English proficiency, and for recipients in all regions
of Los Angeles County. Such consistency in impacts has rarely
been found for Work First programs.
Jobs-First GAIN produced impacts for three subgroups
that are typically considered the least job-ready: nongraduates
(people who had neither a high school diploma nor a GED certificate
at random assignment), people who did not work for pay in
the year prior to random assignment, and the “most disadvantaged”
recipients. (These subgroups are not mutually exclusive.)
The most disadvantaged subgroup, which consists of nongraduates
who did not work in the year prior to random assignment and
who had received welfare payments for at least two years cumulatively
before random assignment, faced more barriers to employment
than any other subgroup examined. As Figure
3 illustrates, Jobs-First GAIN raised two-year earnings
and reduced welfare payments for each of these subgroups,
demonstrating that Work First programs can work for recipients
who lack education credentials, job skills, and work experience.
Jobs-First GAIN also benefited recipients who faced fewer
serious barriers to employment, including people with a high
school diploma or GED and people who worked in the year prior
to random assignment, but it did not affect recently approved
applicants for welfare.
As shown in Figure 3, Jobs-First
GAIN led to employment and earnings increases for all four
racial/ethnic subgroups in the single-parent sample: non-Hispanic
whites, African-Americans, Hispanics, and Asians. Moreover,
for single parents with a limited command of English, Jobs-First
GAIN increased two-year earnings by $1,800 (or 45 percent)
per experimental group member and reduced two-year welfare
payments by more than $1,000 (9 percent; results not shown).
These impacts represent a notable achievement for the program
because lack of English proficiency often limits job prospects.
X.Impacts for Two-Parent Families (AFDC-Us)
The evaluation’s research design permitted only one adult
member of each two-parent household to be included in the
research sample; it was decided that this would be the first
adult in the household to show up for a program orientation
during the sample intake period. Nearly half of the AFDC-Us
in the sample are women.
·
Overall,
Jobs-First GAIN produced large two-year impacts on both employment
and earnings.
In the follow-up period, 55 percent of control group
members in the AFDC-U group worked for pay (see Table
7). The average control group member earned $6,598 (including
zeros for people who never worked). Jobs-First GAIN produced
a 10 percentage point increase in employment and an earnings
gain of $2,050, or 31 percent. These results acquire particular
importance in light of TANF’s work requirements, which are
much stricter for two-parent families than for single parents.
TANF requires a higher percentage of two-parent families than
single parents (in 1998, 75 percent versus 30 percent) to
work or participate in employment-related activities and requires
that they work more hours per week (35 hours versus 20 hours)
to be counted as participants.
·
Jobs-First
GAIN led to large reductions in two-year AFDC/TANF and Food
Stamp expenditures.
The program reduced welfare expenditures by $1,429, or
12 percent, over the two-year follow-up period (see Table
7). As was the case for single parents, these savings
resulted mostly from case closures, but also from lower average
monthly grants for those still on welfare. At the end of year
2, 60 percent of experimental group members versus 66 percent
of control group members were on welfare. The majority of
experimental group members who were employed at the end of
year 2 also received a welfare check. These results, while
promising, demonstrate once again that it is difficult to
move large numbers of recipients off assistance by instituting
financial incentives to combine work and welfare. Jobs-First
GAIN also reduced Food Stamp expenditures by $606, or 13 percent,
over two years.
·
The
earnings gain for members of two-parent families was offset
by reductions in AFDC/TANF and Food Stamp payments, leaving
average total income unchanged.
In year 2, both experimental and control group members
received an average of about $11,400 in earnings (minus payroll
taxes), EITC payments, AFDC/TANF, and Food Stamps. Although
this finding means that Jobs-First GAIN did not boost income
for AFDC-Us, it may nevertheless be viewed as relatively positive
because most previously evaluated welfare-to-work programs
actually reduced overall income for members of two-parent
families. (The effect of Jobs-First GAIN on income is uncertain,
however, because earnings and other income from spouses and
partners were not measured.)
·
Jobs-First
GAIN led to larger increases in employment for women than
for men. The two-year earnings gain, however, was nearly twice
as large for men as for women. The reduction in welfare expenditures
was also larger for men.
Over two years, Jobs-First GAIN produced a large (12
percentage point) employment gain for women and a somewhat
smaller (8 percentage point) increase for men. However, the
two-year earnings gain was nearly twice as large for men ($2,645)
as for women ($1,486), despite men’s smaller gain in employment
(see Figure 4). This is because male
experimental group members who worked earned more per quarter
(on average) than their control group counterparts, whereas
female experimental group members did not (results not shown).
Jobs-First GAIN reduced two-year AFDC/TANF payments by
a significantly larger amount for men ($1,750, or 14 percent)
than for women ($1,005, or 9 percent). For both subgroups,
the program continued to reduce AFDC/TANF receipt and payments
at the end of follow-up, but savings were larger for men.
·
Jobs-First
GAIN led to large reductions in two-year AFDC/TANF and Food
Stamp expenditures.
The program reduced welfare expenditures by $1,429, or
12 percent, over the two-year follow-up period (see Table
7). As was the case for single parents, these savings
resulted mostly from case closures, but also from lower average
monthly grants for those still on welfare. At the end of year
2, 60 percent of experimental group members versus 66 percent
of control group members were on welfare. The majority of
experimental group members who were employed at the end of
year 2 also received a welfare check. These results, while
promising, demonstrate once again that it is difficult to
move large numbers of recipients off assistance by instituting
financial incentives to combine work and welfare. Jobs-First
GAIN also reduced Food Stamp expenditures by $606, or 13 percent,
over two years.
·
The
earnings gain for members of two-parent families was offset
by reductions in AFDC/TANF and Food Stamp payments, leaving
average total income unchanged.
In year 2, both experimental and control group members
received an average of about $11,400 in earnings (minus payroll
taxes), EITC payments, AFDC/TANF, and Food Stamps. Although
this finding means that Jobs-First GAIN did not boost income
for AFDC-Us, it may nevertheless be viewed as relatively positive
because most previously evaluated welfare-to-work programs
actually reduced overall income for members of two-parent
families. (The effect of Jobs-First GAIN on income is uncertain,
however, because earnings and other income from spouses and
partners were not measured.)
·
Jobs-First
GAIN led to larger increases in employment for women than
for men. The two-year earnings gain, however, was nearly twice
as large for men as for women. The reduction in welfare expenditures
was also larger for men.
Over two years, Jobs-First GAIN produced a large (12
percentage point) employment gain for women and a somewhat
smaller (8 percentage point) increase for men. However, the
two-year earnings gain was nearly twice as large for men ($2,645)
as for women ($1,486), despite men’s smaller gain in employment
(see Figure 4). This is because male
experimental group members who worked earned more per quarter
(on average) than their control group counterparts, whereas
female experimental group members did not (results not shown).
Jobs-First GAIN reduced two-year AFDC/TANF payments by
a significantly larger amount for men ($1,750, or 14 percent)
than for women ($1,005, or 9 percent). For both subgroups,
the program continued to reduce AFDC/TANF receipt and payments
at the end of follow-up, but savings were larger for men.
·
The
program positively affected many subgroups of two-parent families,
but not as consistently as it did single-parent subgroups.
As
shown in Figure 4, Jobs-First GAIN increased two-year earnings
and reduced welfare payments for AFDC-Us who entered the program
with relatively severe disadvantages in the labor market:
no high school diploma or GED, no recent work experience,
and long-term welfare receipt. Less positively, earnings remained
low for experimental group members in these subgroups, averaging
from $2,500 per year (for the most disadvantaged) to a little
more than $4,000 per year (for those with no high school diploma
or GED). (Both averages include zeros for subgroup members
who did not work for pay in year 1 or year 2.) Jobs-First
GAIN produced especially large two-year earnings gains and
welfare savings for the most job-ready AFDC-Us: those who
worked for pay in the year before random assignment and were
short-term welfare recipients. For reasons that are unclear,
the impacts on earnings and welfare payments were smaller
for high school graduates and GED recipients than for nongraduates.
The
earnings gain for Hispanic AFDC-Us averaged more than $1,900
per year, an unusually large impact. Asian AFDC-Us experienced
a more moderate gain of about $700 per year. In light of the
fact that about half of Hispanics and almost three-fourths
of Asians in the AFDC-U sample lacked English proficiency
at random assignment, these results are impressively positive.
Non-Hispanic white AFDC-Us did not benefit as much from the
program. Jobs-First GAIN raised their employment level (result
not shown), but produced only a small and not statistically
significant increase in their earnings. The program reduced
welfare expenditures for all three racial/ethnic groups analyzed.
The savings were largest for Hispanics (21 percent) and smallest
for non-Hispanic whites (7 percent). There were not enough
African-American AFDC-Us to allow for reliable analysis.
XI.
Estimates of Jobs-First GAIN’s Cost-Effectiveness
The full report provides a limited
benefit-cost analysis of Jobs-First GAIN both for single parents
and for members of two-parent families. First, it considers
whether experimental group members became better off financially
as a result of their exposure to Jobs-First GAIN’s services,
messages, and mandates. The analysis also assesses whether
the program saved the federal, state, and local governments
money. Whether made from the perspective of the welfare sample
or from the perspective of the government budget, the benefit-cost
estimates rest on the assumption that Jobs-First GAIN continued
unchanged through the two-year follow-up period (that is,
that Los Angeles County did not implement CalWORKs) and that
control group members continued to be barred from the program’s
services, messages, and mandates.
The primary benefit-cost estimates
cover a five-year time horizon starting with the first quarter
after random assignment (quarter 2), which includes an observation
period and a projection period. The observation period for
each sample member encompasses the first two years after his
or her random assignment. Based on several assumptions about
trends over time, the gains and losses observed at the end
of year 2 were then projected through the end of year 5. The
analysis presents two estimates, each based on a different
set of trend assumptions. The more conservative estimate is
based on the assumption that Jobs-First GAIN’s effects at
the end of year 2 will decay over time, hitting $0 at the
end of year 5. The more optimistic estimate is based on the
assumption that the program’s effects will remain constant
from the end of year 2 through year 5. (Both patterns have
been observed in earlier evaluations of welfare-to-work programs.
For some programs, however, the effects actually increased
in later years.) For both estimates, the dollar value of each
projected effect was discounted at a rate of 5 percent per
year to reflect the fact that dollars earned or saved in the
future cannot be invested and are, therefore, worth less than
dollars earned or saved in the present. All benefit-cost estimates
are expressed in 1998 dollars.
·
Over five years after random assignment,
AFDC-FG experimental group members will have achieved a small
financial gain relative to their control group counterparts,
whereas AFDC-U experimental group members will have incurred
a small financial loss.
A program produces a net gain from
the standpoint of the welfare sample if experimental group
members’ earnings gains, fringe benefits,
and EITC payments exceed the value of reductions in their
transfer payments and increases in the taxes and child care
costs that they pay. Considering only the observed (that is, two-year)
effects on single parents, experimental group members’ increased
income from earnings, fringe benefits, and EITC payments exceeded
by about $400 their loss from paying payroll and sales taxes
and out-of-pocket child care costs and from receiving less
in welfare, Food Stamps, and Medi-Cal. Single parents in the
experimental group continued to realize a small net gain in
income at the end of year 2. Projected through the end of
year 5, these results increase the size of the welfare sample’s
net gain by an additional $200 to $400 per experimental group
member, depending on trend assumptions. AFDC-U experimental
group members did not fare as well in the observation period
and will not fare as well in the projection period: Depending
on trend assumptions, they are expected to incur a net loss
of between $300 and $700 over five years. (None of the benefit-cost
results discussed above are shown.)
·
Jobs-First GAIN will lead to substantial
savings to the government budget for both AFDC-FGs and AFDC-Us.
From the perspective of the government
budget, a welfare-to-work program is cost-effective if the
value of increases in tax revenues (minus EITC payments) and
savings in welfare, Food Stamps, and Medi-Cal payments and
administrative costs exceeds the net cost of providing employment-related
services to experimental group members. For AFDC-FGs, Jobs-First
GAIN will very likely realize between $2 and $3 in increased
revenues and savings for every additional dollar spent on
experimental group members, a substantial return to the budget.
The net gain to the government budget is likely to be even
higher for AFDC-Us (results not shown).
XII.
Discussion and Implications
As noted above, many TANF-era programs, including Jobs-First
GAIN’s successor, CalWORKs, are offering post-employment services
and more generous financial incentives to increase employment,
job retention, and advancement among welfare recipients. The
two-year findings presented in this summary of the final report
on the Jobs-First GAIN Evaluation confirm that pre-employment
services remain an important part of a welfare-to-work program’s
strategy for fostering self-sufficiency. Los Angeles County’s
transition from a basic-education-focused welfare-to-work
program to a strongly employment-focused program was successful:
Jobs-First GAIN achieved relatively large two-year employment
and earnings gains exceeding those for Los Angeles GAIN by
a wide margin. Even more impressive is the fact that Jobs-First
GAIN’s overall impacts remained large at the end of year 2
and will almost certainly continue. Moreover, Jobs-First GAIN
demonstrates that a welfare-to-work program can succeed in
one of the country’s large urban areas, where welfare-to-work
programs have traditionally fared poorly, and for many different
types of welfare recipients, including people with low English
proficiency and people with other severe disadvantages in
the labor market.
The
findings of the evaluation also point to areas where CalWORKs’
additional services and incentives may be instrumental. One-third
of experimental group members in the Jobs-First GAIN Evaluation
never worked for pay in the two-year follow-up period. Before
TANF, people who could not find jobs after enrolling in a
welfare-to-work program faced the greatest risk of experiencing
long-term welfare dependency. Under TANF, they risk reaching
time limits in welfare eligibility. In Los Angeles, these
people may benefit from CalWORKs’ wider range and more targeted
mix of services, including greater access to skill-building
activities combined with part-time employment, more aggressive
case management and supportive services to help people overcome
barriers to employment, and special services to address domestic
violence, mental health problems, and substance abuse. Other
experimental group members who appear not to have benefited
from Jobs-First GAIN’s services and mandates (at least not
within the first two years of follow-up) include those who
worked sporadically or worked at low-paying jobs without fringe
benefits. The evaluation also found that unreliable child
care arrangements caused many experimental group members to
miss or be late for work. Under CalWORKs, welfare recipients
with these problems may benefit from access to post-employment
services that increase job retention or speed the transition
to a new job and from more aggressive case management to ensure
that all who qualify for Transitional Medi-Cal and child care
benefits and for the EITC receive these types of assistance.
The two-year impacts of Jobs-First GAIN on employment
and welfare status also highlight issues that CalWORKs’ current
mix of services and financial incentives may not address.
Notably, the program was more successful at reducing welfare
expenditures than at moving recipients off assistance. This
finding may be expected, because the program encouraged enrollees
to take advantage of California’s Work Pays financial incentives
by combining work and welfare in the short term. As noted
earlier, CalWORKs provides a somewhat stronger financial incentive
to combine work and welfare, but at the same time puts a five-year
limit on adults’ eligibility for assistance. At some point,
the program will need to increase the rate of exits from assistance
— otherwise, many adult recipients will begin to encounter
lifetime limits on eligibility without a means of compensating
for the lost income. The key issue is whether welfare recipients
advance to jobs that offer stable, full-time employment at
hourly wages exceeding the relatively high eligibility cutoff
for continued assistance. However DPSS modifies its welfare-to-work
program to meet this challenge, Jobs-First GAIN’s mix of services,
messages, and financial incentives will likely continue to
be an integral part of DPSS’s efforts to promote employment
and self-sufficiency.