Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 137
7
Effectiveness of Temporary
Jobs Programs
Programs providing temporary subsidized employment have been a
mainstay of youth employment and training efforts in the United States
since the War on Poverty. The committee reviewed five reports on YEDPA
programs that provided temporary jobs for youths. Three served
out-of-school youths exclusively: Ventures in Community Improvement
(VICI), Supported Work, and the Public Versus Private Sector Jobs
Demonstration Project. Two were designed primarily for in-school
youths: the Summer Youth Employment Program (SYEP) and the Youth
Incentive Entitlement Pilot Projects (YIEPP).
PROGRAMS FOR OUT-OF-SCHOOL YOUTHS
Table 7.1 presents the characteristics of the three
serving out-of-school youths. Table 7.2 summarizes the
and results of the evaluations of those programs.
Ventures in Community Improvement
programs
research design
Ventures in Community Improvement was a demonstration project
operated under the Youth Community Conservation and Improvement
Projects (YCCIP) part of YEDPA. The target population consisted of
youths aged 16-19 who were out of school, had employment difficulties,
or were economically disadvantaged (eligible for CETA). The program
provided participants with up to 12 months of work experience (on
average, participants stayed for 6 months) on construction projects to
improve public or low-income housing. Participants were supervised by
union journeymen at a ratio of 6 to 1. Job placement assistance was
provided to those completing the program, and participants were actively
encouraged to complete a General Equivalency Diploma (GED) program.
One objective of the project was to determine the impact of the program
on participants' subsequent labor market outcomes. Other objectives
were to test the feasibility of replicating the program model on a
broader scale and to find a way to measure the value of the community
housing improvements produced under the program. In all, there were
eight sites involving a total of 1,500 participants. The demonstration
137
OCR for page 138
138
a, U1
n ~
:,
Z U)
U]
.,,
U]
._
a)
S
C~
o
·e
U2
S
o
o
o
S
V
~n
1
1°
o
o
~n
o
U]
o
0
E~
.
00 U~
0 ~
~ 0
P4
q~
0 ~
.~'
s V
4U
P4
U]
v
~1
Q4 ~
~ ~n
0 ~
CS
4~
V
a) ~
s
U]
~ a
V
.,. ~
L4 0
a' ~
U) P4
s
~ 0
0 ~
~ P4
P.
4U
a
rl
0 ~
L' O
s
0 ~
l:5
a
a'
U] 4
s ~
~ U) 0
U) ~V
s ~n O
4~ s ~ 0
0 C 00 =) ~
_ a,
~ ~0
U)
kD
a) aJ
0 11 0 ~a) a,
~ ~3 s
X ~4U
Q]U~
~n
U] :s U]
s ~O ~S ~5
~ U] o O O ~ O
0 4 ~V O U]
~ U) ·- ~ U]
O ~ ~
~a 0 Y
0 0 a, V ~ ~0 ~ v ~ ~0 ~ P4 v a
O ~ ~ ~ UO ~O ~ ~ U] ~O ~ O
S ~ _~ '- tl5 S -~ rd ~S ~ ~1
V `5` Q ~ £ V o Q tc: ~V ~ ~ m
U] dP ~U) ~U) ~
1 ~ 1 dP dP dP 1 dP dP dP 1 1 dP dP dP
kD ~ u~ C ~S ~ rn ~ oo ~ ao
O ~ ~ ~ OD ~ ~ ~ ~ CX)
1 -1 1
..... ~ ~
~:s
0 0
o
. - ~
~ 0
V U]
~ U)
k4 ~ ~ 0 a' ~
U] · - ~ U]
U2 ~ ~ ~ 0
~ ~ s
a) O ~ ~ ~ ~5 v
V ~ ~ ~V ~ ~V
0
a) ~ C: 0 a)
~ ~ ' 4U V . - ( ~, - , -
h U] U1 V L' ~
~ ~ J~ 5 ~a) . - 4~ -- - a'
O ~ V
X ~ O U] s ~X
a) I~ ~0 4~ Q] ~ a ~a
~ ~ O
O ~ ~ O ~Y
·- U] ~ V ~ q~ U) U] U] ~
3
U] Y
s~
~O O
.,. 3
U2
~Y C~ ~U] ~ ~ U) O
V ~ ~ O
~ ~ ~ 0 0 ~Q, v N U] V
0) '- ~ ·- 3 0) J.) U) N ~ 0) ·~1 ~ ~ Q) tt5 J~
·- ~ a) ~v ~ (IS ~ ~ ~ ~ ~ V
54 ~ ~ V ~ ~ ~ ~ O ~ ~ ~ a
a'
p4 ~ ~ ~N ~ ~ X ~ ~ ~ Q ' -
X O 0 4U ~ Ll ~ O ~ Q O q~l ~ ~l. - ~S ~ O
' ~ a) ~ ,~ ~5 3 ~1 0 U) Q, ~ ~ ~ ~
Q4 ~ Q4 ~ p4 ~ U) P4 ~ Q -1 U] Q,
~ ~ ~ O ~ X ~ ~ O ~ ~ ~ ~ X ~ ~ O ~
k4 0 ' - V Q O ~ td V ~ 0) U) O aJ ~, Q, ~, aJ
0 3 ~
3 Cq E~
O
~ ~4 ~0
4 ~ ~ ~ V - -
O V O ~ V
~ ~ 0) ~ --, 3 ^ ~U] U) ~U] ~
.,' ~ ~ ~ ~ ~ W- - ~
-1 Q) ~ ·0 Q Q} 41) ' ~t ~4-~ (~5 ~J 0) 4~ 4~ ~U2
U] ~ ~ _ ~ ~ ~ ~ ~ 0 3 ~ ~ V
Q) ~ 0 H Q a' ~s ~ 4~ 0 0 a) ~ ~ ~
s ~ s ~V ~ ~ O ·=
H ~ ~ ~ ~ O V
J~ O ~ ~ O O ~ ~ ~ ~ ~ ~ ~ ~ O
_ ~ ~ ~ ~ ~ _
U]
OCR for page 139
139
U]
3
U]
a
._'
U]
aJ
S
tn
·e
U]
S
3
o
o
o
S
C)
U)
1
o
1
o
Ld
o
U]
o
P4
U)
Qo
o
E~
U]
a
Q]
a)
1 u'
O O
~q
O a
h ~
U1
4
q~
O
O
U)
O
O
~ O
O S
4
~:
O
~ O
O
C)
O
O
a) ~
N
.,.
U)
-~1
a) ~
.,'
u, a.
S ~
~ O
O ~
I O
·~ L~ O ~
~ ~1 S ~ ~JU
, ~.,~ ~ O
0 1 ~ u] :~. ~ a ~::
U] ~ ~ ~ ~ S ~ ~ ~ ' -
_t a~ (0 -~ ~ c: ~ u~ ~S
~ ~ Q. ~ c~ ~ 5: ~ a ~ .- J~
~ ~ u, m- ~ ~ ~ ~o n
:' ·' ~ ~ u, O I ~ ~ Q ,. t ~>
~ ~ ~ ~ 0 3
0 0 ~ a) ~4 ~ ~ 0 0 ~ s ~ ~
O ~ U] ~ o ~ ~O
a,
UD q~ tU C5, CD ~ O
0~ ~5 ~ ~ -/ G~ q~ 3 U] O U] dP
4J ~ ~ ~ U'
ns a~ s 0 c ~ ~ ~ :5 ·~ u~ ·~-) u~
a) ~ ~ ~ ~ u' O ~ s ~ ~,' ~ .-' _
~ ~ a' ~ ~ ~ ~ ~ ~
4~ U] tO - ~ ~d ~ O O c: ~S
m
- ~ q~ ~, ~ R5 u] ~ S Q1 ~, ~·,
O O ~ ~ ~ ~ ~ ~ ~ ~ ~S
~ U] ~ ~ O ~ ~ ~ ~ ~S O
0 ~ a) ~ ~5 ~ tlS ~ Q, ~ S ~ V O u~
U] ~.,'
~:
dP dP
dP dP ~C~
Lf) ~
°D 1 l- U]
1 ~c;~ 1 11 1 11
' U] ' O
O ~ kD ~q U] U] U2
O ~ ~ ~O ~O ~
~ ~ ~ ~Q~ ~ a' o
dP dP dP dP dP d~
rn c~ U] u~ ~ U] ~ ~U] ~ S ~ ~U] ~ ~ ~U] ~ ~ a~
s ~ ~ s ~ ~ s ~ ~s ~ ~ ~ ~s ~ ~ ~s S" ~ ~a'
4 ~ O ~ O ~ ~ ~ ~ ~S
~11 11 ~11 11 ~11 11 ~ ~ O ~ ~ 0 11 ~ 0 11
O O O O S: ~ C ~O ~ O O ~ O
V~ ~ ~ ~ ~ ~ V ~
oo ~.. _'
OD a, ~rn OD
~U]
.,, 4 ~
S V O
~n a ~
a) ~.d
C V
a) ~ u ~u
~ ~ ~O 3 O
O ~ ~c Q. a
~1
V q ~d~
V ~ ~ Y ~S ~O ~ ~ ~ ~O
a) a) a' s:: ~ ~ ~ s ~0
0 ~n 0 0 ~U] ~ 0 ~ tn ~u~
~5 ~q~ ~ ~ ~ 3 ~ 0 3 ~ S ~11 ~11 ~11
a) ~a' ~ ~c: ~s ~ ~ 0 ~ ~ ~11 ~11 ~11
u] U] u] 3 a~ O ~ to u ~ O ~ ~ u
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ O ~ S ~ ~ ~ ~ ~ ~ V ~ S ~ ~ V
a' ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 0 ~ ~ ~ a, ~ ~ ~ ~ O ~ ~ ~ ~ ~ ~ ~ ~ O ~ · -
k4 ~ ~ ~ U) ~ ~ ~ ~ O ~ ~ ~ ~ ~ ~ O ~ O ~ ~ ~ ~ ~ O
v v v c ~ s~ ~ + ~ + S + ~ + 0) + s + + ~ 0 Q ~1 ~Q ~ ~ ~: Q ~ O ~ ~ Q
~ ~ `:: ~O ~oo ~ ES ~ 3 s~ ~ ~ 3 ~ EE ~ Q~ 3
H H H ~ q ~I I I ~I I _' I ~ e ~ ~ ~ ~ ~
~0 ~0
· e ~+ I e e e e V e 4 ~e V
C 4 ~J~
H ~
U]
~U'
P. V ~ V
q~ H (IJ O U] ~ t ~V ~1
O t-) ~m ~15 ~1 '- O
c: 0 ~ ~Ql ~ ~ ~ 4~
u' ~ ~ ~ ~ ~ Q V
~, ~`: s a) ~ ES
~ ~ U2 ~a' ~ ~ ~ 0 ~ ~n
0 0' ~ ~ ~ 3 ~ tlS
S O O ~C ~ O ~ ~ O
' - ~ ' - ~ ~ ~ ~ ~ JJ
0 4~ JJ ~ O ~
~: v c: ~u ~u~ a~ ~ :>
O ~ :3 0 ~u' ~ ,-' ~ s:
u2 c: ~ V 3
' - ~ 4~ ~ ~ s a
U] H ~ ~ V ' ~ ' -
·a a ~ ~ ~0 0 ~ ~ x 0 U1 ~5 Q
~, ~ O ~ H ~ ~ ~ ~ V ~ ~ O
~ ~ V ~ ~ ~ ~ ~ ~ ~ ~ .=
0 G5
C: P: P;
~r ~0
a) 0
o
E~
U ~O
O
~Lt,
o
U] U] ~
3 Q 43
~: ~ a ~_ u ~O
.,'
-~d a. ~ ~ a) 4J ~
u' ~ ~a' O ::5 ~ ~ ~ ~0 V
0 ~0 0 ~ 0 ~ a
S ~ ~ ~ O
O ~ V O ~ ~ ~ ~ O
° ~ 0 U] ~ ~ a' a,
Q H ~3- ~n~ mo
~:5
:> u
OD o
<3\ -'
,,
,~
. 4J
4
a) v
~:
u'
L.
a'
a, 0
ra
_
0 ~
i4 a
al ~
~ 0
a) v
JJ
q~ 0
~ s3
U]
ra Ul
a, 0
~ S
C~
a
O ~
1 q~ a;
s
O
~ aJ ~
3 ~1 0
~ a
u2
4
O u
:' ~ ~
1 0
3
O O
q~
0~ ~
q~ ~ O
V
~ a
a) u'
4~ a
I
O
c:
O
~ q~
s: O
O u2
eq
4
oo u2 ~
~ ~ ~q
m~
mn
s
OCR for page 140
140
was implemented between September 1978 and February 1979 and concluded
in all sites in September 1980.
Comparison groups were derived from three sources: participants in
YCCIP programs run by the Department of Housing and Urban Development
(HUD; operating under an interagency agreement with the Department of
Labor) in four sites that overlapped VICI; youths in formula-funded
YCCIP construction programs in three sites; and youths randomly selected
from VICI waiting lists after the programs were fully enrolled with
youths judged to be the most motivated. According to the evaluation
report, objections raised by the Department of Labor and a pressing
need to launch the VICI project without further delay were cited as the
reasons for not following a true randomized evaluation design. Unfor-
tunately, few detailed data on participants in non-VICI programs are
provided and no data are provided on comparison group members drawn
from the VICI waiting lists.
Postprogram follow-up data were obtained from interviews adminis-
tered to comparison groups as well as the VICI participants at 1 month,
3 months, and 8 months postprogram. Final samples used for estimating
effects varied from 160 to 500 VICI participants and from 160 to 650
comparison group members.
Although long-term program effects could not be estimated, the
statistically significant estimated short-term effects of VICI relative
to individuals drawn from VICI waiting lists were (1) an increased
probability of employment; (2) an increased probability of being in an
apprenticeship position or on an apprenticeship waiting list; and (3)
higher quarterly earnings (a maximum of $1,050 with an average effect
of $3221.~ Not surprisingly, comparisons among VICI, HUD, and other
YCCIP participants failed to identify a dominant program. All of the
effects of VICI participation were estimated after controlling for
demographic characteristics, geographical location, and the date of the
participant's most recent interview. The evaluation report states that
personal characteristics did not exhibit statistically significant
effects on outcomes, while site differences did, but these results are
not shown in the report.
The evaluation design of the study is seriously flawed in several
respects. The nature of the various programs used to make comparisons
with VICI, as well as the participants, differed from VICI. The YCCIP
The analyses of outcomes are based on various multivariate tech-
niques, including binomial logit, log-linear regression, and tobit
analysis. The interpretation of the program effects therefore varies.
For example, the finding with respect to employment is that the
likelihood of employment is 111 percent higher among VICI participants
than among "controls." This finding implies that for a participant
whose probability of employment was 0.5 prior to the program the
probability would be about 0.68 after participating; one whose
employment probability was 0.9 prior to participating would have ~
probability of 0.95 of being employed after participating in the VICI
program.
OCR for page 141
141
.
programs often served 10 or fewer youths, and the HUD and VICI programs
usually enrolled more, up to 120 and 60, respectively. In addition,
both the supervisory ratios in the programs and the nature of the work
differed from VICI. VICI programs were construction oriented; the HUD
and YCCIP programs included less skilled activities, usually
landscaping and neighborhood cleanup (in YCCIP). VICI participants
were older and less likely to be enrolled in school.
Because of high sample attrition (only 37 percent of the VICI
participants and 35 percent of the combined comparison groups were
interviewed at the 8-month follow-up), the data actually used for
assessing program effects consisted of the most recent interview data
for each participant who was interviewed at least once after program
completion. In addition to the lack of equivalence in geographical
coverage, the follow-up sample for VICI participants differed not only
from the general VICI participant population but also from the HUD and
YCCIP comparison groups. The study could not determine what, if any,
biases would be present as a result of differences between the
follow-up samples and the total client populations.
Given the severe shortcomings of the evaluation design, what, if
anything, can be learned? The report may provide reliable insights and
documentation regarding implementation and program delivery issues.
Some attention was paid in the report to the optimal degree of site
adherence to a standard plan and latitude for across-site variations to
~ ~ ~ ~ ~ _ ~ ~ _ ~ _ me: ~ : ~ _ -_
~lIUII~U"~= lU~1 ~11~1~1~1~. The report clearly identifies the fact
that referral agencies had little incentive to refer potential clients
to the competing VICI project. The educational link is shown to be
weak: educational institutions had little incentive to play an
important role in the project. Moreover, participants lacked the
energy and motivation to pursue adult education at the end of a day on
a construction project. The involvement of union journeymen as crew
supervisors turned out to be particularly helpful in job placements
because of the journeymen's knowledge of the informal labor market and
their contacts; their referrals and recommendations carried more weight
than comparable activities by CETA job developers.
Although an evaluation of the long-term program effects is ruled
out by the small sample size and limited follow-up period, the study
does attempt a benefit-cost analysis of the program from a societal
point of view. Under a variety of alternative assumptions, the present
value of the benefits consistently exceeds the present value of the
costs. However, benefit-cost calculations are only as credible as the
underlying estimated program effects, and the evaluation study fails to
provide reliable evidence on the effectiveness of the VICI program in
changing the employment and earnings prospects of disadvantaged young
people.
Supported Work
Supported Work was a national demonstration program begun in 1975.
The program concentrated on four target groups: women who had been
receiving Aid to Families with Dependent Children (AFDC) for several
OCR for page 142
142
years; ax-addicts; ax-offenders; and young (17- to 20-year-old) high
school dropouts. Five of the 15 sites enrolled dropouts. Dropouts
participating in the program had no immediate plans for further educa-
tion and were without immediate employment prospects; many had a
history of delinquency. Supported Work sought to inculcate participants
with the necessary work habits, desire for employment stability, skills,
etc., for future labor market success; these were to be achieved through
subsidized work experiences that would be gradually more demanding and
approximate regular unsubsidized employment.
Three aspects distinguish the Supported Work program: (1) peer
group support; (2) graduated stress; and (3) close supervision. The
peer group aspect was implemented through the assignment of individuals
to work crews consisting mainly of program participants, and this
aspect was accomplished in a more or less consistent fashion across
sites. Graduated stress was intended to expose the participant to
increasingly higher performance standards that eventually corresponded
to those typical of regular, unsubsidized jobs. There was a good deal
of variation across sites in how this program aspect was actually
implemented. Close supervision was designed to facilitate the transfer
and then development of skills, proper work habits, and proper
attitudes. Supervisors could be either program staff or nonprogram
supervisors from the host employer.
All five Supported Work sites that had youth enrollees participated
in the evaluation. Applicants were randomly assigned to experimental
and control groups. Those in the experimental group could participate
in the program for a maximum period of 12 to 18 months, depending on
the site. Both experimentals and controls in the sample were
interviewed at baseline and then at 9-month intervals that continued
for up to 36 months for some sample members .2 For purposes of the
evaluation study, the enrollment period started in the second quarter
of 1975 and lasted until the second quarter of 1977. The maximum
length of the postprogram period covered by interviews was determined
by when an individual enrolled. On average, youths in the program left
it well in advance of the maximum period allotted for participation;
only about 18 percent of the enrollees left to take other jobs or to
enroll in an educational or training program (Maynard, 1980:Table
III.2).
The statistically significant positive effects of Supported Work on
employment rates, hours of work, and earnings were largely confined to
the period of participation in the program (see Table 7.3~. During the
first 3 months following enrollment, program participation increased
average monthly earnings of participants by $289 (389 percent), average
hours worked by 112 hours per month (459 percent), and the probability
2A special resurvey was conducted over the period July 1980-January
1981. This resurvey provided data covering a period of 38-67 months
following initial enrollment. The results of the study did not
substantially alter the conclusions reached in the original study
(Maynard et al., 19821.
OCR for page 143
143
TABLE 7.3 Selected Supported Work Outcome Differences For Young
Dropouts
Months
After Percentage Hours Worked Average Earnings
Enrollment Employed Per Month Per Month
1-3
4-6
7-9
10-12
13-15
16-18
19-21
21-24
25-27
28-30
31-33
34-36
37-39
40-42
43-45
46+
68*
43*
27*
19*
4.3
-3.4
-0.2
1.4
0.1
-5.3
-4.9
-5.2
2.4
-4~1
-1.4
2.1
112*
76*
52*
29*
5
-1.8
-2.7
2.3
-2.5
4.2
-6.2
-5.9
0.3
-8.7
-2.7
4.3
$289*
200*
146*
92*
8
18
13
29
_9
-26
~33
-21
9
-31
-20
18
NOTE: Adjusted participant group mean minus control group mean.
*Statistically significant at the .05 level or below in a two-tailed
test.
SOURCES: Data for months 1-15 from Maynard (1980~; data for months 16
and beyond from Maynard et al. (1982~.
Of employment by 68 percentage points (336 percent) relative to
controls. By 10 to 12 months following enrollment, these positive
effects attenuated dramatically: program participation raised average
monthly earnings by $92 (56 percent), average hours worked per month by
29 hours (58 percent), and the probability of employment by 19 per-
centage points (52 percent). Beyond 13 months from the date of initial
enrollment, program participation had no overall statistically
significant effects on labor market outcomes as compared with members
of the control group. The fact that experimentals had longer job
tenure than controls because of program participation had no impact on
postprogram employment rates, hours of work, or wage rates (Table 7.3~.
Statistically significant program gains in hours of work during the
period of actual participation tended to be larger among younger
participants, females, whites, the more educated, those who left school
because they wanted a job, those living with their parents, those
OCR for page 144
144
raised by both parents, and those with greater job training in the year
prior to enrollment. Supported Work had no long-term impact on educa-
tion and training decisions or on drug use. Similarly, there were
minimal long-term program effects on welfare dependence for youths, and
criminal activity was not reduced by the program.
Results of the benefit-cost analysis of the program indicate that
from the societal viewpoint estimated costs exceeded estimated benefits
by $1,465 per youth participant. While net costs were found to be quite
sensitive~to the method of estimating the value of output and project
costs, none of the alternatives reversed the benefit-cost result .3
Overall, the evaluation study appears to have been very careful in
its attention to conditioning factors, random assignment, and the use
of appropriate statistical techniques. We are therefore confident in
the stated finding of no postprogram effect for the severely dis-
advantaged youths who participated in Supported Work.
Supported Work Youth Variation
Starting in July 1979 four Supported Work sites were selected to
participate in a special variation of the program, the Supported Work
Youth Variation (Scharfman, 1981~. The special variation was directed
toward 17- to 20-year-old high school dropouts, many of whom had a
history of delinquency. Their experiences in the conventional Supported
Work program were unfavorable compared with the outcomes for other
targeted groups participating in the program. The variation sought to
incorporate features not generally provided in the regular program,
e.g., counseling, vocational education, skills assessment, and
training; to extend the length of the program to 24 months; and to
establish a tangible link to long-term labor market success and thereby
improve in-program performance. Unfortunately, there was no comparison
group for this follow-up study and therefore we did not find the
reported results meaningful.
30n the basis of a special verification study conducted in three
sites among ax-addicts and ax-offenders in the sample, self-reported
arrests were found to be underreported by 46 percent, on average, by
both experimentals and controls. The study notes that the measured
underreporting among those subgroups may or may not be generalizable to
the youth sample (Maynard, 1980~. Before estimating the value of
reduced criminal activity, a factor of 1.7 is applied to the control-
experimental arrest differential to correct for underreporting in the
benefit-cost calculations. As noted in regard to the Job Corps
evaluation (see Chapter 6), no underreporting factor is generally
accepted in the criminal justice literature.
OCR for page 145
145
Public Versus Private Sector Jobs Demonstration Project
The Public Versus Private Sector Jobs Demonstration Project focused
on differences in postprogram employment and earnings between partici-
pants in fully subsidized public sector jobs and those in fully sub-
sidized private sector jobs. The target population was 16- to
21-year-old, out-of-school youths eligible to participate in the Youth
Employment and Training Program (YETP) portion of CETA. Five YETP
sites that were operating from January 1979 to April 1980 were selected
for the demonstration project. Eligible youths who had completed
preliminary forms were matched in pairs within each site on the basis
of age, race, sex, and their scores on a reading test and then randomly
assigned to a fully subsidized job slot in either the public or private
sector. The subsidized jobs paid the minimum wage and the 100 percent
subsidy lasted 25 weeks. A total of 2,100 participants began the
program. As is typical, there was participant heterogeneity across the
five sites.
Information provided for our review documents the degree of effort
required at each site to develop job slots with public and private
employers. This information was disaggregated by industry for private
employers and by functional areas for public sector employers. In
general, more effort was required to place youths in private sector
subsidized jobs than in public sector subsidized employment.
Forty-four percent (921) of the participants completed 25 weeks in
the program. Completion rates were higher among females, blacks, and
those in public sector jobs (49 percent) than in private sector jobs
(38 percent).
Immediate postprogram information was gathered at the end of the
program period and at 3 and 8 months after program termination. At the
3-month follow-up only about 43 percent of the original sample could be
located and interviewed (54 percent of completers and 34 percent of
noncompleters). Among completers the full-time job rate was 50 percent
for public sector participants compared with 64 percent for private
sector participants. On the other hand the part-time job rate was 16
to 17 percent for both public sector and private sector participants.
Public sector participants exhibited a higher rate of enrollment in
educational or training classes (26 percent) than private sector
participants (18 percent). After controlling for various individual
characteristics, private sector participation continued to be
associated with higher postprogram employment rates.
Results from the 8-month follow-up are derived from the approxi-
mately 35 percent of the original sample who were located and inter-
viewed (42 percent of completers and 29 percent of noncompleters).
Although multiple regressions are not available for this period' the
results continue the earlier patterns among program completers: the
full-time job rate was higher among private sector participants (61
percent) than among public sector participants (52 percent), and the
part-time job rates were much the same for public sector (23 percent)
and private sector (25 percent) participants. By the time of the
8-month follow-up, none of the private sector participants and only 4
percent of the public sector participants was in an educational or
OCR for page 146
146
training program. At the time of the 8-month follow-up, there was
virtually no difference in the employment rate for completers (80
percent) and noncompleters (78 percent), though the high attrition
makes even this statement problematic.
We cannot conclude on the basis of the evaluation that there was
any difference in the effects of subsidized employment in the public
versus the private sector. Although private sector participation was
consistently associated with higher rates of subsequent employment,
adjustments for nonprogram-related characteristics considerably
narrowed the private sector advantage. The evaluation of this
demonstration project offers no basis on which to decide whether the
effort required to secure subsidized jobs in the private sector was
worth the additional cost. Although the study appeared to be very
promising in terms of both the nature of the project it described and
its research design, sample attrition severely limits the reliability
of the reported findings. We are not confident, therefore, about
drawing conclusions about the effectiveness of the project or about the
issue of the desirability of subsidizing public sector compared with
private sector jobs.
PROGRAMS SERVING IN-SCHOOL YOUTHS
Table 7.4 presents the characteristics of the two programs serving
in-school youths. Table 7.5 summarizes the research design and results
of the evaluations of those programs.
Summer Youth Employment Program
The objective of SYEP was to provide economically disadvantaged
youths (14- to 21-year-olds) with summer work experience in order to
"assist these youths to develop their maximum occupational potential
and to obtain employment not subsidized under CETA (P.L. 95-524, Sec.
481 ibid. Program emphasis varied across sites. Some sites offered
vocational training, others provided job counseling, some a combination
of both. Time and resource constraints were cited as the reasons for
not recording the precise program elements to which each participant
was exposed. Consequently, nothing can be learned about what sorts of
interventions were particularly effective or ineffective in accomplish-
ing program objectives. Sites also differed in terms of geographic
characteristics, i.e., urban, suburban, and rural, and in terms of
adherence of eligibility criteria.
The evaluation we considered is of the SYEP conducted at eight
sites chosen by the Department of Labor in the summer of 1979. It is
based on data for approximately 2,000 youths who were ostensibly
randomly chosen to participate or, if not accepted into the program on
grounds other than eligibility, to be in the comparison group. Approxi-
mately 250 youths were divided between the participant and comparison
groups in each site. The treatment and comparison groups differed
significantly in terms of some personal characteristics both across and
OCR for page 147
147
o
aJ u:
Q Q'
:5
~ U)
U]
·,1
U]
._1
h
C)
~:
o
S~
·e
V]
o
o
o
ru
U]
1
H
S"
o
V]
o
U]
Q
o
S~
o
E~
.
m
E~
O ~
O
· -
.d
O Q
.,~,
s
4U
a'
U]
,4
~ U]
O
L4
C'
4
a'
~ ~:
E~ C)
U] ~5
~ a,
.,' ._,
O
~ s~
U]
O
~ ~4
O
P4 <:
JJ
C)
a)
.m
O ~
~ O
s
4
:25
O
:^
4
U]
ss
~r
_
oo ~
U]
U] ~O
·a
Y ~
Ln
O
3 Q. 11
o
X
U] Ul
4U 4U
~:,
O O q:5
O O
O O
C)
~n ~n ~
1 1 (d
~4 ~4 U]
O ~O ~S
1 ~1 1
4~ 0 4~ 0
:25 ~:D
0 ~n ~0 u'
Y ~ ~ ~ ~ ~ a
F: ~ C)
U] ~ ~ ~ ~ ~ Ul
t~ ~ ~ ' - S
Q ~ ~ ~ ~ O Q ~ 3
O ~O ~ C:
0 1 dP dP d~ 0 1 0 dP dP dP d~
S ~ ~ ~ ~r S ~ ~ ~ ~ ~r
~ ~ ~r ~ ·n ~ ~ a
U] U2
I · e e · I e ~e e e
`:: ~
H H
~ 1
u, ~ ~ a
R
:,
~ S ~ q~ ~
43 V ~Ul
U) ~ ~ CJ,
O
0~ ' tn 3 a
U] ~ ~
Q, o ~, a)3
Q] ~ O
~ ~ ~ O
U2 ~ O
k4 Q I S
aJ a) 0 ~ v - -
0 ·a
U)
a,
a ~C: 1
N ~O
.,' ~
~ ~ ~ 0 a,
''] ~a) t(S ~o
U] ~ ~ ~
Q Q4 ~ O a) ~ ta
x m- - c:
~n ~L1 4J ~ ~ ~
O O
Y z5 ~ O ~ q~
s
O O v
3 Q
~O
cn
~_ U]
P~
' ~
U] ~ U] ~ C: ~. -
.~.) C~a~ ~ ~ 0
~ a) ~ 4~ a' ES ~4 ~ ~ 0
O
~ ~ ~ Z ~ ~ ~ ~ ~ ~
O ~ ~) H 4~ 4~ p~ ,<
~ ~1 ~ · O ·- O 1
0} Q~ 0 ~ U) =: ~ ~ H
F: ~ ~ · ~ ~ ~ . - ~ ~ Q
-
:5 0
U)
OCR for page 150
150
SYEP's goal of encouraging youths to return to school. The data
indicate that the program was particularly successful in raising the
part-time employment rate of severely disadvantaged 16-year-old black
males during the subsequent school year and particularly unsuccessful
in bringing about part-time employment for severely disadvantaged
15-year-old females.
Overall, the program seemed to have no significant effect on the
likelihood of contact with the criminal justice system. Also, there is
no evidence of program effects on attitudes toward such contact. The
results did indicate mutually negative attitudes by participants and
program personnel toward each other.
It is not legitimate to treat the evaluation results as if they
were produced by a random experimental/control group design. The
manner in which participants and nonparticipants were chosen varied
across sites. Also, the evaluation report does not explain why the
youths who were eligible for SYEP, who constituted the comparison
group, were not accepted for participation in SYEP. Although the study
suggests program participants gained in employment during the program
and in part-time employment after the program, the evaluation design
does not allow reliable inferences to be drawn about the effectiveness
of the SYEP.
As with other public employment and training programs in the 1970s,
the question of displacement received some attention with regard to
SYEP. The number of jobs provided in a public employment program may
not be the net number of jobs created as a result of the program
because some of the jobs would have existed even if there had been no
special jobs program and because the participants who have jobs may
displace others who would have had the jobs had the program not
existed. To the extent that displacement occurs, it is argued, the
usual estimates of the social benefits of the program may be
overestimates.
The determination of whether displacement occurs for employment and
training programs and, if so, to what degree and in what circumstances,
is quite complex at both a practical and a theoretical level. It
should be noted that while theory suggests that there may be displace-
ment leading to overestimates of benefits as conventionally measured,
there is also a theoretical possibility of replacement, if program
participants are moved from a labor surplus market to a labor shortage
market, in which case conventional methods would overestimate costs
and, thereby, underestimate the net social benefits of the program
[see, for example, Johnson (1979) and Kemper (1980) for discussion of
this issue]. While estimating the degree of displacement is par-
ticularly relevant in the context of estimating the net social benefits
and costs of a given program, society may still ask what proportion of
program output is over and above what would have existed in the absence
of the program.
We reviewed two attempts to estimate displacement in the Summer
Youth Employment Program; one (Zimmerman, 1980) relied on data col-
lected from personal interviews with program operators, and one (Crane
and Ellwood, 1984) was based on aggregate data relating state-level
employment to enrollment in SYEP and other variables. The study by
OCR for page 151
151
Zimmerman is based to a considerable degree on supervisors' judgments
about what projects would have been undertaken in the absence of the
program in the eight sites. It concludes that "in 30 percent of the
cases the output produced by SYEP project participants would have been
produced at the same scale by alternative suppliers in the absence of
the project" (Zimmerman, 1980:77~.
The study by Crane and Ellwood takes an entirely different approach.
The authors used unpublished data from the Current Population Survey
for the 12 largest states for 1972-1978, for the months of April, July,
and October, in order to measure employment by race and age group, and
unpublished program data for SYEP placements by race and age for the
same states over the same time period. The analysis relates employment-
to-population rates for nonwhite 16- to 19-year-olds by state to the
number of SYEP jobs per civilian nonwhite 16- to 19-year-olds for the
state, using various other employment and school measures to control
for what employment would have been in the absence of SYEP.
In such a regression, the coefficient of the SYEP placement variable
provides an estimate of how much each SYEP placement increased the
employment rate for nonwhites aged 16-19 across the 12 states during
that time period. Theoretically, if the SYEP job caused total displace-
ment, the coefficient of the SYEP variable would be approximately zero;
if there were no displacement the coefficient would be close to 1. The
authors conclude (Crane and Ellwood, 1984:23~: "Regardless of the
specification, estimated supplementation effects of SYEP seemed to fall
between .5 an .75. Thus our best estimates is that for each SYEP job
provided to nonwhite youths, one-third of a job is lost in the private
sector for this group. n Despite their assertion, only one of the four
equations yields an estimate that is significantly different from both
1.0 and zero.
Because of reservations we have about the precision of the estimates
and their statistical significance, we are not inclined to accept the
estimates of displacement derived from this study. Overall, we do not
believe that either the Zimmerman or Crane and Ellwood study provide
reliable estimates of the magnitude of displacement in the Summer Youth
Employment Program.
Youth Incentive Entitlement Pilot Projects
The Youth Incentive Entitlement Pilot Projects, which was mandated
under YEDPA, was the largest YEDPA demonstration program. It cost
approximately $240 million--$224 million for stipends and local program
operations and $16 million for monitoring and research (Diaz et al.,
1982:150--and lasted 2-1/2 years (early spring 1978-August 1980) with
an additional phase-out period of a year (fall 1980-summer 1981~.
Low-income youths aged 16-19 who had not yet graduated from high school
constituted the eligible target population for the program. The key
innovation of the program was that all eligible youths who lived in the
target area were entitled to a job if they met enrollment conditions.
Eligible youths were guaranteed minimum-wage jobs, part-time during the
school year and full-time during the summer months.
OCR for page 152
152
To continue their participation in the program, participants were
required to be enrolled in school or in an approved alternative
educational program and to be making satisfactory progress toward a
high school diploma. The short-run goals of the program were to reduce
school dropout rates, raise high school graduation rates, provide work
experience, and provide income during the program participation phase.
The longer-term goal was to improve life-cycle labor market outcomes as
a result of staying in school and receiving work experience (Farkas et
al., 1984~.
In all there were 17 demonstration projects across the country, and
more than 70,000 youths participated. As implemented, the program had
four major characteristics:
1. the average 15- to 16-year-old was enrolled in the program for
15 months (13.4 in the full program plus 1.6 in the transition year);
2. 71 percent of the work experience jobs were in the public
sector;
3. beyond provision of the job itself, very few services and
little training was provided: two-thirds of the youths in the program
received orientation, one-fourth were tested, and one-half received
employment counseling;
4. the enrollment requirement was enforced but the school
attendance and performance requirements were generally not enforced.
YIEPP was not a skills training, job search, or behavior
modification program. Thus, any effects observed are due to the work
experience and school enrollment aspects of the program. In effect,
the youths were provided with jobs and then left on their own to
benefit or not.
The entitlement program was designed to saturate an area with
jobs. Consequently, the presence of the program could have an effect
on the employment of eligible youths even if they did not participate
in the program since the total number of jobs available in the local
area would have increased. To account for this, an innovative approach
was taken in designing the evaluation--use of matched sites.
Four of the large-scale sites were selected as pilot sites for
evaluation purposes. Four sites that did not have the entitlement
program were selected as comparison sites. Cincinnati, a program site,
was matched with Louisville; Baltimore with Cleveland; rural Mississippi
counties with other rural Mississippi counties; and Denver with Phoenix.
The evaluation technique was to estimate regressions on outcomes (e.g.,
employment); the independent variables were individual characteristics
(to control for factors not accounted for by the match) and a dummy
variable indicating whether the person was in a program site; the
coefficient on the dummy variable is the program effect. Clearly, in
using this approach the quality of the match becomes critical.
The evaluation was also designed to include all eligible youths,
both participants and nonparticipants, in the pilot-site study group.
This evaluation strategy can counter the selectivity bias that plagues
evaluations based on nonrandom selection of participants and controls
because those choosing not to participate can differ in significant
OCR for page 153
153
ways from those choosing to participate. This method also yields
information on what the program participation rate might be given a
permanent, ongoing program. Stratified random samples of eligible
youths from the pilot sites and youths from the comparison sites who
would have been eligible had YIEPP operated in their areas were
selected as the interview sample for the evaluation.
Youths who were aged 15-16 at the start of the program constituted
the study group for the final report so the steady state effects of a
permanent program for 16- to 19-year-olds could be determined. This
group was also selected because it was believed that older youths had
already made career decisions prior to being aware of the program and
that this would contaminate pure program effects. In addition, the
program participation rates of older youths were lower than those for
the younger group, e.g., through summer 1980 cumulative participation
rates were 66 percent among 15- to 16-year olds and 46 percent among
17- to 20-year olds.
The sample for the final analysis was limited to blacks aged 15-16
at the time of program enrollment because they constituted the over-
whelming majority of participants and because most Hispanic youths in
the final evaluation sample were residents of Denver, a site that had
had substantial implementation problems. In addition, white youths in
the sample were considered too small in number and too heavily concen-
trated in the Cincinnati/Louisville pair (where a school busing con-
troversy led to substantial changes in white school attendance) to
provide reliable separate estimates. The sample of black youths num-
bered about 1,400 (excluding Denver/Phoenix); about 40 percent were
from Baltimore/Cleveland and 30 percent each from Cincinnati/Louisville
and rural Mississippi.
Because other large YEDPA programs were operating in the comparison
sites, the test was not one of the entitlement program compared with
nothing. While the evaluation provides no way of judging how signifi-
cant a factor this characteristic is, as in other evaluations it would
probably lead to an underestimate of program effects since youths in
the matched comparison sites were able to participate in other
employment and training programs.
As noted above, the program was run as a full--fledged entitlement
between spring 1978 and August 1980. During the transition period
between August 1980 and August 1981, the program operated at a reduced
level with a limited number of openings. The true postprogram
follow-up period, as defined for purposes of the evaluation, was the
postoperation period during the fall semester of 1981.
The follow-up period is troublesome in several regards. First, it
is not a long period in which to observe postprogram effects, and it
raises serious concerns about the extent to which effects that are
observed may persist over time. Second, the final interview covered
less than 2 months of the postprogram (i.e., postoperation) period for
62 percent of the sample. While most youths eligible for the program
did not receive entitlement jobs during the transition period, the
analysis design depends on defining the postoperation period as the
postprogram period, since in theory the employment of eligible youths
might be affected by the availability of an entitlement job. The
OCR for page 154
154
evaluation design does not permit examination of changes in program
effects over time because time since leaving the program cannot be a
variable in this design.
There are statistically significant in-program and postprogram
effects on weekly earnings, largely attributable to enhanced employment
rates, but also due in a modest way to small but statistically sig-
nificant increases in hours worked and wage rates received (in the
postprogram period). In-program earnings effects during the school
year were estimated to range between 46 and 161 percent higher than
weekly earnings in the absence of the program. Comparable earnings
effects during the summer periods vary between 48 and 65 percent.
During operation, the entitlement program significantly lowered
unemployment rates and raised employment and labor force participation
rates for young blacks as well as for all youths. The magnitude of the
effect was sufficient to eliminate substantially the employment and
unemployment differentials between black and white youths eligible for
the program. Thus, employment-to-population rates for blacks increased
from 21.1 to 41.3 percent, and those of ~ ~
37.4 percent during the program (Farkas et al., 1982~. Unemployment
rates decreased during the program, from 72.1 to 51.7 percent among
black eligibles and from 61.1 to 54.8 percent among white eligibles.
An important finding of the YIEPP evaluation is that approximately
two-thirds of the youths eligible for the program did participate at
some time. This finding means, in part, that youths are willing to
work at the minimum wage but that in the absence of a program like
YIEPP employers are unwilling to hire as many (at the minimum wage) who
wish to work. It may also mean, in part, that in the absence of such a
program in-school youths are not as likely to be in the labor force.
YIEPP demonstrated that a system can be found to employ significant
numbers of disadvantaged youths.
YIEPP jobs were largely in the public sector; private sector
involvement increased over time, but the participation rate of private
business was generally low. The percentage of all youth job hours
spent working for private work sponsors increased from 14 percent in
September 1978 to 23 percent in June 1980.
Although not empirically tested under entitlement, the effect of
varying subsidy rates on private employers' willingness to participate
was estimated in Baltimore and Detroit using employers' responses to
hypothetical questions: "Would you be willing to act as a work sponsor
at a 50 percent subsidy, a 75 percent subsidy, etc."? No comparison
was made between expressed willingness to participate and actual
whites increased from 31.2 to
behavior (Ball et al., 1981~. Only 5 percent of businesses said they
would participate if offered a 50 percent subsidy; 10 percent said yes
with a 75 percent subsidy. Even at a 100 percent subsidy, only 18
percent of private employers surveyed said that they would
participate. While the elasticity of the employer participation rate
with respect to changes in the subsidy rate appears high, i.e., a
doubling of the subsidy rate more than doubles the indicated
OCR for page 155
155
participation rate, this was not a true experimental test of
participation since there were no observations of actual behavior.
The overall effects of YIEPP on secondary school enrollment and
graduation were generally inconsequential. Virtually all of the
estimated schooling effects are statistically insignificant. However,
for black females in 1981, there was a sizable, statistically signifi-
cant reduction in college enrollment; otherwise, the overall college
enrollment effects are statistically insignificant, with some variation
across sites. This particular program effect is troubling and deserves
further scrutiny.
In the postprogram follow-up semester, program earnings effects
were estimated to be 39 percent above weekly earnings in the absence of
the program. These figures took the sites as the unit of observation
and are based on average earnings increases for all youths in the site
regardless of participation. If one assumes that the observed effects
persist for a year, there would be an estimated increase of $545 in
annual earnings for eligibles.
During the postprogram period in the fall of 1981, labor force
participation rates of the full youth cohort were higher than those of
the comparison group, but unemployment rates were not significantly
different. Among young black eligibles) unemployment and labor force
participation rates were not significantly different from those of the
comparison group.
Because of the scale of the entitlement effort and its potential as
a model for future programs serving a substantial portion of the youth
population, we devoted considerable effort to our review of this
evaluation. We concluded that the YIEPP evaluation was a sound one and
that meticulous attention was paid to the problems inherent in the
quasi-experimental (matched sites) design used in the evaluation. The
drifting apart of matched pairs of pilot and comparison sites over
time, i.e., the Baltimore/Cleveland and Cincinnati/Louisville pairs,
was recognized. The potential bias induced by attrition was sys-
tematically investigated through a special attrition sample and found
not to change the essential conclusions of the study.
Although not enough information is available to determine the
long-term effects of YIEPP, the finding that there are noteworthy
positive in-program effects on the employment rates of black and white
youths is convincing. Correction of program effects for individual
characteristics using regression analysis does not change the
results. Unfortunately, it is not possible to be as confident about
other in-program effects or about effects generally for Hispanic or
white youths because problems with the Denver/Phoenix and
Cincinnati/Louisville pairs severely reduced the number of Hispanics
and whites in the study.
Related to the discussion of the magnitude of the employment
effects of the entitlement program is the issue of displacement or net
job creation. The report on in-program effects (Farkas et al., 1982)
OCR for page 156
156
estimates that one new job was created for each 1.4 entitlement jobs
funded in the public sector and for each 2.2 jobs funded in the private
sector. The report also indicates that these assessments may be con-
servative in the sense that they only measure the extent of displacement
among program-eligible youths and do not include a measure of displace
ment of ineligible people (e.g., nondisadvantaged or out-of-school
youths or adults) who might otherwise have been hired.
If these estimates are accurate, approximately 40 percent of the
measured employment change resulted from shifting people who would have
been employed in the absence of the program. Such a finding would have
important implications for the net effects of the program in terms of
job creation, as well as obvious cost implications.
We discussed earlier the complexities involved in estimating dis-
placement in connection with our review of the Summer Youth Employment
Program above, and many of those concerns apply here as well. Because
samples of youths were drawn both in the entitlement pilot sites and in
the matched comparison sites, one could in theory capture one major
element of the degree of displacement by using the comparison site
figures to estimate what the employment of entitlement-age youths would
have been had there been no entitlement project.
This is essentially what Farkas et al. (1983) did by comparing the
employment rates in the pilot sites with those in the comparison sites
during the period of program operation for similar age-race groups and
then dividing the differences in employment by the number of entitlement
jobs in the pilot sites to measure net job creation displacement. Their
estimate of the magnitude of net job creation is about 70 percent,
implying displacement of around 30 percent. At the same time, control
sites probably had substantial numbers of youths in YEDPA and other
federal employment and training programs, and so the meaning of YIEPP
displacement, even if accurately measured, is not clear.
Another study of displacement in the entitlement program (Gould et
al., 1982) surveyed a sample of private firms, both those that had
actually provided jobs for entitlement participants and those that had
not. Information was gathered on the levels of output and employment
in these firms before, during, and after the period of the entitlement
program. Using the data from the period before entitlement and for
those firms not directly hiring participants, the authors developed
econometric models that provided estimates of what employment in the
firms would have been had there not been an entitlement project. (The
analysis was complicated by the fact that even in the preprogram period
many of the employers had other subsidized workers, presumably from
CETA programs, in their firms.) The authors estimate that about 40
percent of the jobs created through entitlement resulted in displace-
ment.
While this was an imaginative and interesting effort, several
features of it could lead one to question the precise magnitude of the
estimate of displacement. First, the response rate to the initial
survey was rather low (54 percent). Second, we have some doubts about
the ability of the econometric model based on limited preprogram data
to estimate what employment would have been in the absence of the
program when operating with a small sample of highly individualized
OCR for page 157
157
firms in particular localities. Third, and perhaps most important,
this method provides no indication of what happened to those workers
and resources that were displaced from these particular firms. Did
they find alternative employment or remain unemployed? Since there is
no market-wide measure of the degree of shortage, this important element
in the determination of the final degree of displacement is missing.
We believe that the technical problems with each of the studies
estimating displacement in the entitlement program are sufficient to
prevent confidence in accepting any given point estimate of its
magnitude.
There remain several other concerns about the entitlement evalua-
tion's findings. The in-program and postprogram findings vary con-
siderably by site. As noted above, we are convinced that the positive
in-program effects are sound, but we are not so convinced with regard
to postprogram effects. The postprogram results for weekly earnings
show a negative effect for Baltimore (relative to Cleveland), while
Cincinnati had an equal absolute value positive effect (relative to
Louisville) with a smaller sample size than Baltimore, and the effect
for rural Mississippi was positive. Hence, the finding of an overall
positive postprogram effect is influenced considerably by the results
for Mississippi.
In other words, if the analysis is limited to urban sites, the
average effect is zero or slightly negative. The researchers explain
the negative Baltimore effect by noting an unexpectedly healthy
Cleveland economy that might diminish the quality of the match over
time. However, Baltimore was described as the best run of the
programs. Because of across-site disparities in results and because it
does not seem plausible to average rural and urban sites, the
entitlement program results are most appropriately viewed as separate
case studies. It is therefore difficult to see how to extract results
on postprogram effects that are generalizable to the nation as a whole.
The estimated postprogram effects of the entitlement program vary
considerably across demographic subgroups as well as sites. The
postprogram earnings effects for young black males are (nominally)
twice those for young black females, $13.66 and $6.13 respectively, but
the estimates for females are not significantly different from zero.
The postprogram effect on weekly earnings for older (17- to 20-year-old)
black youths was $4.14, and the effect for 15- to 16-year-old whites
and Hispanics was one-sixth the size of the effect for comparable
blacks, approximately $1.53 per week. Both effects were statistically
significant. 4
4 This result is based on the weighted averages of estimated weekly
earnings effects. The weekly earnings effect of $9.11 for the young
black cohort (Denver/Phoenix included) declines to $7.45 for the full
young cohort, i.e., when Hispanic and white youths are added (Farkas et
al., 1984~. Using the sample weights, we computed the estimated
earnings effect for white and Hispanic youths (combined) as
9.11 n1 + X n2 = $7 45'
OCR for page 158
158
Besides weekly earnings effects, other outcomes also differed
considerably by sex. The school enrollment/graduation outcomes for
females did not appear to follow any pattern. Overall the program
appears to have reduced college enrollment for females in two of the
three sites. There is no readily apparent explanation for this result,
since high school graduation rates are not significantly lower in those
sites. Between the first interview and the last, the proportion of
young black females with one or more children increased from 6 to 48
percent in one site and varied substantially across the sites. The
estimated employment-related effects for young black women relative to
men are probably influenced to some degree by the high rates of
childbearing that characterized young black females in the sample.
SUMMARY
The evaluations of temporary jobs programs consistently found
evidence that in-program earnings and employment were higher as a
result of the program. The findings of the Summer Youth Employment
Program evaluation tentatively suggested in-program gains in employ-
ment, but we have only limited confidence in the evaluation. The
Supported Work and entitlement evaluations provided the strongest
evidence on this issue.
In the case of Supported Work, a program serving severely dis-
advantaged dropout youths, monthly in-program earnings of participants
were $289 above those earned by the control group during the initial 3
months after enrollment in the program. Youths eligible for the
entitlement program earned, on average, up to $9 more per week during
the School year than the comparison group and up to $10 more per week
during the summer. Earnings of the black youth cohort were as much as
$12-$13 higher than the comparison group's during both the school year
and the summer months (Farkas et al., 1982~.
For participants in Supported Work, 97 percent were employed during
the first 3 months after enrollment, compared with 29 percent of
nonparticipants. Employment rates of young blacks in the entitlement
program were up to 26 percentage points (235 percent) higher during the
program than those of youths in the comparison sites.
From this evidence we can generally conclude that temporary jobs
programs effectively increased employment for participants and, hence,
served an income transfer goal that has been an underlying rationale in
many such programs. Without regard to the merits of this particular
.
where, n1 equals black youth cohort weight, n2 equals white and
Hispanic youth cohort weight, and X equals weekly earnings effect for
white and Hispanic youths:
$9.11 (.781) + X (.219) = $7.45
X (.219) = $7.45 - $7.115
X (.219) = $0.335
X = $1.53
OCR for page 159
159
program goal, the extent to which the estimated employment effects of a
jobs program translate into an increase in the total number of jobs
available, an increase in the number of employed persons, or a decrease
in the number of unemployed persons may vary.
While there is considerable disagreement about the proper way to
estimate displacement in job creation programs, and a corresponding
distrust of any given point estimate of its magnitude, most researchers
and policy makers acknowledge the displacement problem. 5 From
society's point of view the nature of displacement may be as important
as its level. Thus, if a program displaces people who are more
advantaged than participants, and who could easily find alternative
employment, it may be considered less a problem than if the program
displaces individuals who are equally disadvantaged (Masters, 1981~.
The estimated in-program effects on labor market outcomes other
than employment rates and earnings were variable among the programs and
for different target groups within most of the programs. The entitle-
ment program significantly lowered unemployment rates and raised labor
force participation rates for all eligibles in the young cohort. Among
Supported Work participants, gains in hours worked tended to be larger
among those who were younger, females, white, and more educated.
Neither the Public Versus Private Sector Jobs Demonstration Project nor
VICI provided much information on in-program effects.
Because the research design of the entitlement program required the
ability to measure postprogram effects in an entire youth labor market
(not only for participants), the postprogram period was defined as the
time following the close of the program. Thus, the postprogram period
for the entitlement program was the fall semester of 1981, and 62
percent of the black youth sample was interviewed within 2 months of
the time the program terminated.
The entitlement program increased postprogram earnings of eligible
black youths by the equivalent of $545 per year (assuming measured
postprogram gains persisted) and raised postprogram hours worked for
employed black youths by 6 percent (from 32 to 34 hours per week).
Employment rates were higher among blacks and labor force participation
was higher for the full youth cohort. Neither labor force participation
s For example, the YEDPA authorizing legislation and the 1978 CETA
amendments required the Secretary of Labor to make periodic reports on
various aspects of the entitlement program, including displacement.
Reports were to include findings with respect to enrollment; costs; the
degree to which out-of-school youths returned to school or others
remained in school; the percentage of eligible youths participating;
the kinds of jobs provided and a description of the employers--public
and private; the degree to which on-the-job or apprenticeship training
was offered; the estimated cost of extending the program to all areas;
the effect of the program in reducing youth unemployment in the pilot
areas; and the effect of program job opportunities on other oppor-
tunities for youths in the area [P.L. 95-93 (YEDPA), Sec. 329; Report
No. 95-1765 (CETA Amendments of 1978), Sec. 4201.
OCR for page 160
160
rates nor unemployment rates of the young black cohort of entitlement
eligibles were significantly different from those of the comparison
group in the short postprogram period (Farkas et al., 1982~.
There is limited evidence on the long-term effects of participation
in temporary jobs programs. The Supported Work program clearly
indicated no such effect for the severely disadvantaged dropouts it
served. The evidence from the other studies is less clear, for a
variety of methodological reasons.
a maximum of 8 months of postprogram experience, some as little as 3
months. The VICI evaluation was designed to measure effects 8 months
after program participation, but was generally unconvincing.
The studies of temporary jobs programs that we examined were not
very encouraging about the goal of raising school attendance rates,
lowering drug abuse, or reducing negative encounters with the criminal
justice system. With respect to school retention, the summer jobs
program evaluation offered questionable evidence in support of
increased school participation. The entitlement program had no effect
on either school retention of youths already in school or school
completion by dropouts who had returned.
Most of the other studies analyzed
Representative terms from entire chapter:
weekly earnings
