Fostering Non-Cognitive Skills in Active Labor Market Programs: Evidence from an RCT in Israel Analia Schlosser and Yannay Shanan Tel Aviv University
Preliminary results from an ongoing experiment June, 2017
MOTIVATION • Active labor market policies (ALMPs) use a set of tools to increase working skills and employment among unemployed and welfare recipients • They have become a standard feature of modern welfare policies in most OECD countries (Figure 1) • However, we still know little regarding what works for whom and why
IN THIS STUDY ✓We evaluate an ALMP implemented in Israel using an experimental design ✓ALMP program targeted at unemployed income support claimants ✓Each participant receives a personalized treatment + group coaching ✓Program focuses on enhancing non-cognitive skills
✓We analyze a wide range of outcomes and estimate dynamic and heterogeneous effects ✓We examine the possible channels behind program impact
✓We analyze program externalities (not today) ✓Long term ongoing evaluation (short term results so far)
RELATED LITERAURE
Types of ALMPs • Productivity enhancing • Training • Wage subsidies either to employers or financial incentives to workers • Direct employment in public sector • Matching • Services and Sanctions (i.e. counseling and monitoring, job search assistance, and sanctions in case of noncompliance)
RELATED LITERAURE: ALMPs Effectiveness
• Card et al. (2015) meta analysis of ALMPs (only 25% based on RCTs) • Kluve (2010) meta analysis (only 9 of 141 studies based on RCTs) ✓ Most evidence based on non-experimental studies ✓ Little but growing evidence based on RCTs ✓ Substantial heterogeneity in the impact of such programs ✓ Less evidence on the long term effects of ALMPs ✓Outcomes examined: earnings, employment, hours worked, exit rate from unemployment
RELATED LITERAURE: ALMPs Effectiveness • Review by Crépon and van den Berg (2016): we still know very little regarding • Underlying mechanisms • What works for whom • Cost benefit • GE effects • Traditional ALMPs may be ill-designed for reintegrating most unemployed into society. • One may consider alternatives such as mentoring, therapy, and group treatments to boost self-esteem and other personality traits. ➢Overall, no evidence so far on programs that focus on enhancing non-cognitive skills and their potential effect on labor market success
RELATED LITERAURE: The role of Non-Cognitive Skills • Increasing interest in the effect of non-cognitive skills on success in life • Importance of non-cognitive skills for test scores, schooling, and labor market outcomes (Brunello and Schlotter, 2011; Heckman and Kautz, 2013; Heckman and Rubinstein, 2001; Borghans et al., 2008; Jacob, 2002) • The variance of many later-life outcomes explained by non-cognitive abilities rivals that explained by cognitive skills (Heckman el al., 2006, Lindqvist and Vestman, 2010). • Some evidence that early life and school interventions can have an impact on non-cognitive skills (Carneiro and Heckman 2003, Heckman et al. 2013, Martins, 2010, Felfe et al. 2011).
• Little evidence on the returns to investments in non-cognitive skills among adults (e.g. Blattman et al. (2015) on criminals in Liberia) • Not much is known regarding which skills are malleable later in life.
INSTITUTIONAL BACKGROUND: Income support in Israel • Approx 100,000 households receive income support in Israel (5% of population)
• Avg. monthly allowance for couple with 2 children in 2013: 2,843 NIS (32% of avg. monthly income or 66% of min. wage) • Eligibility based on age, income, assets, and ‘Employment test’ • Employment test: those considered able to work (healthy, age<60, age of youngest child>2 for mothers) need to report weekly to local employment office • Treatment at employment office is minimal: weekly fingerprint in automatic machine and once a month (or when relevant) meeting with employment officer who provides employment referrals • Not reporting to employment office or rejecting a relevant employment offer results in denial of monthly income support payment
INSTITUTIONAL BACKGROUND: Employment Circles Program •
‘Employment Circles’: mandatory active labor market program designed to integrate unemployed income support claimants in the labor force, preventing long-term chronic unemployment
•
Start date: February 2014 • 20,000 participants until Nov. 2016 • 3,400-5,700 participants until February-June 2015 (sample of today’s talk)
•
Target group: unemployed income support claimants aged 20-50 in 15 local employment offices participating in the pilot • New and re-registered claimants - ‘flow’ • Existing claimants in the system for 2 years or less - ‘stock’
•
Program expansion: (not part of today’s talk) • Individuals with longer welfare history • Additional employment offices • Older individuals
INSTITUTIONAL BACKGROUND: Employment Circles Program • Claimants
are randomized weekly in each of the employment offices into treatment and control groups
• One
of the first RCTs implemented by a government agency in Israel
• Path-breaking
step towards the use of evidenced-based policy making in Israel
PROGRAM DETAILS Week 1:
Income support claimant signs up at employment office (flow)
Income support claimant is already in the system (< 2 yrs.) (stock)
Randomized into treatment or control group by central office (block randomization: week-employment office-flow/stock)
Week 2:
Control: continues with usual procedure reporting weekly to employment office
Treated: Informed that is part of a special mandatory program
PROGRAM DETAILS • Jointly developed with “Tevet”, based on the American STRIVE model, which emphasizes personal development and the improvement of soft skills needed to integrate and excel at a job.
• Participants are required to attend employment office 3 times a week • Program details • Overall program duration: 3-7 months • Avg. annual cost per participant: 1,400 NIS
DATA •
The Israeli Employment Service (IES) operational database (updated June 2015) • Report to employment office, job referrals, refusals, welfare claim status: before, during and after program implementation • Socio-demographic information
•
National Insurance Institute of Israel (NII) data (updated February 2015) • Social benefits, disability status: before, during and after the program. • Employment and earnings (only before program – waiting for data of post-program period)
•
Survey data (February-June 2015) • Employment history, socio-demographic information • Wages, work hours and employment status • Non-cognitive skills (work self-efficacy, job search skills, self-efficacy) and barriers to employment • Overall satisfaction and satisfaction from income and work
DESCRIPTIVE STATISTICS Characteristics of Employment Offices in pilot program First wave
Second wave
All pilot employment offices
Non-pilot employment offices
(1)
(2)
(3)
(4)
Number of active job-seekers Age Education Number of supported children Women Married Arab Immigrant Previous work experience
8,380 38.4 10.2 2.3 0.62 0.46 0.47 0.16 0.64
15,227 38.1 8.8 3.1 0.60 0.55 0.74 0.11 0.58
23,607 38.2 9.3 2.8 0.61 0.52 0.64 0.13 0.60
33,031 38.4 9.5 2.5 0.64 0.48 0.56 0.15 0.62
Income support claimants
0.96 4.0 0.058 7
0.97 6.0 0.072 7
0.97 5.0 0.065 14
0.96 5.1 0.071 59
Locality S.E.S Local Unemployment rate N
Weighted averages by number of job seekers in employment office.
DESCRIPTIVE STATISTICS: Number of participants by employment office
IDENTIFICATION STRATEGY: MAIN EQUATION
Outcomeijtp: outcome of individual i, from employment office j, randomized at week t, who belongs to stock/flow group p
Treatmenti: treatment dummy Xi: Vector of individual characteristics: gender, marital status, number of children, new immigrant, age, health status, ultra-orthodox, Arab
μjϒtδp: Employment office * randomization week * stock/flow Fixed Effects
MAIN EQUATION ✓We include individual characteristics to increase precision and to control for any differences between T and C units that derive from randomization in finite sample. Results identical (but less precise) without controls. ✓Additional model with individual fixed effects for welfare recipiency outcomes (and in the future for earnings) ✓We also estimate models for dynamic effects: monthly outcomes based on subsample randomized >=8 months.
IDENTIFICATION STRATEGY: additional design issues
• 73 individuals from the treated group (1.2%) were removed from the program due to serious physical or mental health limitations.
• We include them in our analysis. So our estimates measure intent to treat (ITT) rather than treatment and are a bit diluted. • We could do IV. But given their low share, we just see the estimated effect as ATE.
RESULTS • Descriptive statistics of program participants and Balancing tests ✓Flow sample is “stronger” than stock sample ✓No differences between treated and comparison groups ✓Randomization worked!
RESULTS • Main results based on administrative data (effect after 7 months in the program) ✓25% reduction in likelihood of receiving income support ✓1,640 NIS (24%) reduction in total welfare payments vs. cost per participant per year of 1,400 NIS ✓No substitution with other social assistance payments
• Results based on individual Fixed effects models ✓Virtually identical to main results
RESULTS • Dynamic Effects (based on subsample with program duration >=8 months) ✓Impact of stick vs. carrot: Some immediate impacts after program enrollment and further impacts after active participation in program • No differences in benefits denial due to refusals of job offers It might be that threat effect is enough Decrease in income support payments does not derive from benefit denials • No effect on number of job referrals No evidence for resource diversion (could go either way)
SURVEY RESULTS • Survey response rate: 33% • Individuals with different duration in the program (median=8.7 months)
• Selection into survey: women, married, and health limitation more likely to respond. Immigrants, less likely. • However, no evidence for differential selection according to treatment status We use weights to reflect program population We are able to replicate main results based on survey weighted population
SURVEY RESULTS • Welfare to work? Results on traditional labor market outcomes (individuals with different duration in the program) ✓Increase in LFP (16%) and employment (25%) ✓Most of the increase in employment comes from part time jobs (hours effect 1.8 / employment .079)=22.8 hours per week ✓Hourly wages a bit above min. wage (23.12 NIS) ~ 204/(1.8*4.3)=26.3 ✓Increase in earnings for marginal workers who started to work (percentiles 62-80)
SURVEY RESULTS • Program tools: Results on non cognitive skills and search capacity ✓Non-cognitive skills are relevant for labor market outcomes ✓Positive impact on non-cognitive skills related to program content ✓No effect on less malleable non-cognitive skills (falsification test) ✓Program tools or employment effect? • Well-being and expectations ✓No effect on well-being ✓Higher expectations on future employment among those who have been longer in the program
HETEROGENEOUS EFFECTS
• Heterogeneous effects by welfare and employment history ✓Stronger impact on less attached to labor market (no recent employment history / with recent welfare history) • Heterogeneous effects by demographic characteristics ✓All groups affected to some extent ✓Stronger impact on Arabs and the less educated • Heterogeneous effects by local unemployment rate ✓Stronger impact in areas with higher unemployment rates
COMPARISON WITH OTHER PROGRAMS Flow
Stock
Employment circles Lights to employment 2014-2015 2007-2010 (1) (2)
Employment circles 2014-2015 (3)
Lights to employment 2007-2010 (4)
Mehalev 2005-2006 (5)
Evaluation method
RCT
Quasi-experimental
RCT
Quasi-experimental
Quasi-experimental
Program effect on Income support receipt
-0.09
-0.08
-0.16
-0.19
-0.27
Mean of control group
0.42
0.40
0.63
0.77
0.80
Change in %
-21%
-20%
-25%
-25%
-34%
7 months
7 months
7 months
8 months
15 months
Program effect on employment
0.05
0.05
0.15
0.08
0.10
Mean of control group
0.33
0.43
0.28
0.44
0.28
Change in %
15%
12%
54%
18%
36%
5-16 months
7 months
5-16 months
8 months
15 months
1,400 NIS
12,000 NIS
1,400 NIS
12,000 NIS
6,000 NIS
Measurement
Measurement Avg. cost per participant
SUMMARY • Program increased employment and decreased welfare recipiency • Effect seem to be partly driven by an improvement in non-cognitive skills • Larger impact for: • long term unemployed • longer history on welfare dependence • Unskilled (<=12 yrs of schooling) • Areas with high local unemployment rates
• Program pays itself already in the first year after implementation
NEXT STEPS • Effects on employment and earnings based on Social Security records • Heterogeneous effects on employment and earnings: • What works for whom • Exiting welfare vs. entering employment • Additional outcomes (job ‘quality', job persistency, total household income) • Disentangle effects of different program tools (increased monitoring, workshops, and sanctions) • More on non-cognitive skills (second survey) • Long term effects • Externalities • Does the program have any displacement effects (i.e. crowding out other jobseekers)? • Has the program deterred new job seekers from claiming welfare, lowering the incoming flow of welfare claims?
Public expenditure on active labor market policies - % of GDP among OECD Countries (2011) 2.5
2.0
1.5
1.0
0.5
0.0
Source: Labour market programs: expenditure and participants, OECD Employment and Labour Market Statistics (database)
DESCRIPTIVE STATISTICS: Number of participants by employment office Flow
Jerusalem Beitar Elit Beit Shemesh Bnei Brak Tel Aviv Nazarath Haifa Shfaram Tveria Tzfat Afula Um El Fachem Ashdod Beer Sheva Total
Stock
Control
Treated
Control
Treated
(1)
(2)
(3)
(4)
606 49 12 29 116 356 293 302 101 110 142 388 182 1021 3707
452 48 167 154 333 309 420 245 254 216 275 215 326 557 3971
68 9 15 15 53 64 62 77 62 65 69 79 55 50 743
121 30 65 109 140 110 128 188 194 127 129 139 192 58 1730 30
Balancing tests All treated (1)
T-C (2)
female
0.548
age
35.519
married
0.481
children
2.205
single parent
0.229
immigrant
0.208
health limitation
0.368
-0.008 (0.011) 0.001 (0.001) 0.009 (0.015) -0.002 (0.003) 0.004 (0.016) -0.011 (0.014) 0.001 (0.010)
Flow treated T-C (3) (4) 0.555 34.633 0.479 2.131 0.222 0.221 0.316
-0.010 (0.012) 0.001 (0.001) 0.016 (0.016) -0.001 (0.003) 0.017 (0.019) -0.009 (0.016) 0.013 (0.012)
Stock treated T-C (5) (6) 0.534 37.553 0.486 2.375 0.245 0.177 0.488
-0.000 (0.022) 0.001 (0.002) -0.011 (0.036) -0.005 (0.007) -0.031 (0.032) -0.013 (0.034) -0.032 (0.021)
Balancing tests (cont.) All
Flow treated T-C (3) (4)
treated (1)
T-C (2)
arab
0.380
0.385
ultra orthodox
0.181
2010-2012 work experience
0.485
2013 work experience
0.335
0.028* (0.017) 0.029 (0.019) 0.002 (0.012) 0.011 (0.012)
2010-2012 income support recipiency
0.337
0.291
2013 income support recipiency
0.355
-0.015 (0.013) 0.005 (0.015)
F-stat for joint significance of covariates N
0.164 0.497 0.370
0.233
1.0 5701
10151
Stock treated (5)
T-C (6)
0.038** (0.018) 0.013 (0.022) -0.005 (0.013) 0.005 (0.014)
0.369
-0.015 (0.044) 0.071* (0.039) 0.026 (0.027) 0.033 (0.026)
-0.012 (0.014) -0.001 (0.016)
0.440
0.220 0.457 0.256
0.635
1.1 3971
7678
-0.020 (0.026) 0.014 (0.029)
1.6 1730
2473
.1 0 -.1 -.2 -.3
-.3
-.2
-.1
0
.1
Dynamic Effects Percent reporting to employment office (Treatment â&#x20AC;&#x201C; Control)
0
1
2
6 5 4 3 Months since randomzied
Flow N=2,925
7
8
9
0
1
2
6 5 4 3 Months since randomzied
Stock N=1,457
7
8
9
-.3
-.3
-.2
-.2
-.1
-.1
0
0
.1
.1
Dynamic Effects Percent receiving full income payments (Treatment â&#x20AC;&#x201C; Control)
0
1
2
5 4 3 Months since randomzied
Flow N=866
6
7
8
0
1
2
3 4 5 Months since randomzied
Stock N=727
6
7
8
Sanctions and referrals
Benefits denial Job referrals per month (1)
(2)
Controlled difference
-0.003 (0.006)
0.037 (0.031)
Outcome mean
0.076
0.560
Program effect 7 months after randomization Mean control group
Flow & Stock
(1)
(2)
Employment office activity
0.531
-0.144*** (0.022)
Income support recipiency
0.493
-0.125*** (0.022)
Unemployment benefits recipiency
0.011
-0.001 (0.003)
Disability payments recipiency
0.034
-0.001 (0.010)
Average monthly Income support payments
861.651
-204.352*** (36.164)
Total acumulated income support payments
6,893.207
-1,634.813*** (289.315)
Selection into survey treated female age
-0.004 (0.008) 0.015* (0.008) -0.001 (0.001)
health limitation arab
0.021** (0.009) -0.020 (0.014)
married
0.024** (0.012)
ultra orthodox
0.028* (0.016)
children
-0.002 (0.002)
Pool
0.009 (0.010)
single_parent
-0.007 (0.013)
immigrant
-0.042*** (0.011)
Selection into survey treated female age
-0.020 (0.067) 0.001 (0.021) -0.001 (0.001)
treat_female treat_age
0.027 (0.032) -0.000 (0.002)
married
-0.002 (0.026)
treat_married
0.048 (0.041)
children
-0.000 (0.005)
treat_children
0.003 (0.008)
single_parent
-0.026 (0.029)
treat_single_parent
0.060 (0.043)
immigrant
-0.052* (0.030)
treat_immigrant
-0.017 (0.040)
health_limitation
0.071*** (0.019)
treat_health_limitation
-0.040 (0.029)
arab
-0.056 (0.034)
treat_arab
0.065* (0.038)
ultra_orthodox
0.047 (0.047)
treat_ultra_orthodox
0.040 (0.052)
Balancing test survey All treated (1)
T-C (2)
female
0.563
age
35.363
married
0.508
children
2.229
single parent
0.214
immigrant
0.171
health limitation
0.368
0.008 (0.029) 0.001 (0.002) 0.046 (0.040) -0.007 (0.008) 0.043 (0.046) 0.005 (0.041) -0.045 (0.028)
Flow treated T-C (3) (4) 0.575 34.440 0.537 2.298
0.200 0.183
0.308
-0.010 (0.034) 0.001 (0.003) 0.049 (0.049) -0.008 (0.009) 0.049 (0.056) 0.031 (0.048) -0.036 (0.033)
Stock treated (5)
T-C (6)
0.539
0.059 (0.056) 0.000 (0.004) 0.042 (0.065) -0.003 (0.019) 0.030 (0.081) -0.064 (0.081) -0.072 (0.052)
37.322 0.447 2.081
0.242 0.144
0.497
Balancing test survey (cont.) All
Flow
Stock
treated (1)
T-C (2)
treated (3)
T-C (4)
treated (5)
T-C (6)
arab
0.366
0.385
0.172
0.092 (0.056) 0.037 (0.055) -0.003 (0.007) 0.023 (0.034) -0.021 (0.036) -0.016 (0.044) 0.007 (0.054)
0.325
ultra orthodox
0.068 (0.049) 0.020 (0.051) -0.002 (0.006) 0.019 (0.030) -0.007 (0.031) -0.032 (0.035) 0.028 (0.040)
0.014 (0.102) -0.021 (0.114) 0.004 (0.010) 0.026 (0.060) 0.016 (0.064) -0.050 (0.064) 0.055 (0.065)
11.229 2010-2012 work experience
0.518
2013 work experience
0.339
2010-2012 income support recipiency
0.325
2013 income support recipiency
0.375
F-stat for joint significance of covariates N
0.169 11.217 0.508 0.355 0.297 0.255
0.6 1124
1949
0.178 11.253 0.539 0.306 0.383 0.631
0.5 360
519
0.5 764
1430
40
Main results: full sample vs. survey sample
Income support recipiency
Unemployment benefits recipiency
Disability payments recipiency
Full sample
Survey sample
(1)
(2)
-0.110***
-0.090***
(0.015)
(0.033)
0.516
0.514
-0.001
-0.005
(0.004)
(0.007)
0.019
0.025
0.0004
-0.009
(0.006)
(0.010)
0.029
0.028
Main results: full sample vs. survey sample Full sample (1)
Survey sample (2)
-135.392*** (24.514) 850.667
-79.045 (54.681) 862.247
Total accumulated income support payments
-1035.104*** (168.256) 4971.147
-781.374** (365.427) 5834.327
Monlthly average NII transfers
-140.150*** (26.476) 1113.123
-161.087*** (54.365) 1167.883
Total accumulated NII transfers
-1098.575*** (206.454) 6642.562
-1245.479*** (402.408) 8038.804
6635
1949
Average monthly Income support payments
N
Survey results on labor market outcomes Outcome mean
Number of observations
Treatment effect
(1)
(2)
(3)
LFP
0.566
1949
0.091*** (0.031)
Employment
0.319
1949
0.079** (0.035)
Full time employment
0.159
1943
0.007 (0.026)
Hours worked
8.534
1863
1.805* (1.091)
Earnings
952.972
1825
204.009 (130.768)
Total income (from work and welfare payments)
2131.384
1825
29.097 (127.357)
-.05
0
.05
.1
.15
Program effect on hours worked
0
10
20
30
Treatment-Control Difference in CCDF
40
50
-.05
0
.05
.1
.15
Program effect on earnings (NIS)
0
1000
2000
3000
4000
5000
-1000
0
1000
2000
Quantile treatment effects on the distribution of earnings
0
.2
.4
.6
.8
1
Relationship between non-cognitive skills and labor market outcomes control group LFP
employment
fulll time employment
(1)
(2)
(3)
(4)
(5)
Job search self efficacy score
0.240*** (0.034)
0.107*** (0.029)
0.057*** (0.022)
367.647*** (118.883)
3.198*** (0.880)
Work self efficacy score
0.157*** (0.023)
0.094*** (0.018)
0.039*** (0.014)
243.625*** (82.684)
2.110*** (0.606)
Self efficacy score
0.098*** (0.027)
0.063*** (0.024)
0.051*** (0.016)
192.195* (105.239)
1.793** (0.830)
ADHD score
-0.147*** (0.031)
-0.104*** (0.026)
-0.063*** (0.023)
-296.493** (122.524)
-2.871*** (0.959)
Note: The table reports the average effect of the individual items included in each index.
earnings
weekly hours
Program effects on non-cognitive skills (z-scores)
Job search self efficacy score
N
Treatment effect
(1)
(2)
1748
0.101** (0.045)
Work self efficacy score
1724
0.122*** (0.046)
Self efficacy score
1754
0.015 (0.047)
ADHD score
1765
0.024 (0.05)
Program effects on non-cognitive skills (z-scores)
Job search self efficacy score
N
Treatment effect
(1)
(2)
1748
0.101** (0.045)
Work self efficacy score
1724
0.122*** (0.046)
Self efficacy score
1754
0.015 (0.047)
ADHD score
1765
0.024 (0.05)
Note: The table reports the average effect of the individual items included in each index.
Program effect on job search self-efficacy (z-scores)
I am confident in my abilities to search for a job
N Treatment effect (1) (2) 1772 0.110 (0.090)
I am confident in my ability to use the internet in order to find a job
1748
0.281*** (0.085)
I am confident in my ability to write a resume
1797
0.145 (0.089)
I am confident in my ability to pass a job interview
1755
0.115 (0.080)
Program effect on work self-efficacy (z-scores) N (1)
Treatment effect (2)
Achieve goals that will be assigned
1729
0.148** (0.070)
Respect schedules and working deadlines
1751
0.205*** (0.078)
Learn new working methods
1737
0.124* (0.073)
Concentrate all energy on work
1735
0.210*** (0.075)
Collaborate with other colleagues
1743
0.132* (0.069)
Have good relationships with my superiors
1733
0.154** (0.065)
Be courteous to customers
1724
0.131** (0.067)
Get to work on time
1746
0.194*** (0.070)
How well can youâ&#x20AC;Ś
Program effect on general self-efficacy (z-scores) N
Treatment effect
(1)
(2)
I can always manage to solve difficult problems if I try hard enough
1772
-0.014 (0.072)
If someone opposes me, I can find the means and ways to get what I want
1754
0.044 (0.071)
It is easy for me to stick to my aims and accomplish my goals
1772
0.030 (0.083)
I can usually handle whatever comes on my way
1758
0.060 (0.065)
Program effect on ADHD (z-scores) N
Treatment effect
(1)
(2)
I find it hard to finish things I have started
1784
0.114 (0.085)
I often get distracted by activity or noise around me
1795
0.055 (0.097)
I often misplace or have difficulty finding things
1803
-0.098 (0.084)
I find it hard to concentrate at work
1765
-0.050 (0.073)
I often feel restless or fidgety
1780
0.015 (0.084)
I find it hard to wait in line
1789
0.103 (0.092)
Suggestive evidence: Employment vs. program tools effects Outcome means and differences Employed Control Treated (1)
(2)
Unemployed diff (3)
Control Treated (4)
(5)
diff (6)
Job search self efficacy score
0.266 0.203 -0.063 [0.728] [0.767] (0.062)
-0.259 -0.094 0.164*** [0.894] [0.850] (0.050)
Work self efficacy score
0.269 0.253 -0.016 [0.629] [0.676] (0.054)
-0.303 -0.113 0.190*** [1.080] [0.948] (0.060)
Self efficacy score
0.199 0.153 -0.045 [0.686] [0.763] (0.059)
-0.194 -0.119 0.075 [0.918] [0.908] (0.054)
ADHD score
-0.232 -0.144 0.087 [0.599] [0.675] (0.052)
0.168 0.124 -0.044 [0.758] [0.776] (0.045)
Program effects on non-cognitive skills (z-scores) â&#x20AC;&#x201C; falsification test Non Placebo N Treatment effect
Placebo Treatment effect
N
(1)
(2)
(1)
(2)
Job search self efficacy score
1748
0.077** (0.039)
193
-0.037 (0.099)
Work self efficacy score
1724
0.109*** (0.041)
192
0.003 (0.117)
Self efficacy score
1754
0.021 (0.040)
193
-0.092 (0.122)
ADHD score
1765
-0.028 (0.038)
192
0.172* (0.095)
Program effect on satisfaction and expectations <8 months since randomization Treatment N effect
>=8 months since randomization Treatment N effect
N
All Treatment effect
(1)
(2)
(3)
(4)
(5)
(6)
Overall satisfaction
1763
-0.033 (0.062)
877
-0.030 (0.071)
886
-0.023 (0.121)
Job satisfaction
667
0.121 (0.138)
333
0.292 (0.177)
334
-0.212 (0.262)
Employment expectations
1668
0.087 (0.067)
800
-0.004 (0.093)
868
0.233** (0.114)
Program effects on income support recipiency by work and welfare history Not employed 2010-2012
Employed 2010-2012
Not employed 2013
Employed 2013
(1)
(2)
(3)
(4)
-0.128*** (0.020) 0.574
-0.100*** (0.024) 0.445
-0.140*** (0.018) 0.575
-0.040 (0.031) 0.391
3167
3468
4451
2184
Received income support in 2010-2012 (5)
Did not receive income support in 2010-2012 (6)
Received income support in 2013 (7)
Did not receive income support in 2013 (8)
-0.123*** (0.027) 0.701
-0.097*** (0.020) 0.428
-0.162*** (0.028) 0.751
-0.080*** (0.021) 0.415
2822
4486
2150
4485 57
Program effects on income support recipiency by demographic groups men
women
jews
arab
(1)
(2)
(3)
(4)
Non Ultra Orthodox Jewish (5)
-0.098*** (0.024) 0.463
-0.110*** (0.022) 0.553
-0.094*** (0.019) 0.446
-0.136*** (0.025) 0.591
2907
3728
3884
no health limitations (9)
with health limitations (10)
non single parents (11)
-0.105*** (0.018) 0.500
-0.122*** (0.029) 0.550
4399
2236
Ultra Orthodox native born Jewish
immigrant
(6)
(7)
(8)
-0.101*** (0.021) 0.423
-0.069 (0.054) 0.531
-0.096*** (0.026) 0.459
-0.096*** (0.036) 0.425
2751
2899
985
2523
1361
single parents
12< educ
12 educ
12+ educ
(12)
(13)
(14)
(15)
-0.107*** (0.018) 0.506
-0.127*** (0.041) 0.552
-0.154*** (0.023) 0.588
-0.093*** (0.021) 0.470
0.003 (0.118) 0.299
5224
1412
2836
3478
321 58
Program effects on income support recipiency by local unemployment rate
Unemployment rate<7.5%
Unemployment rate>7.5%
(1)
(2)
-0.077*** (0.023)
-0.131*** (0.021)
0.428
0.569
2,827
3,808
59
PROGRAM DETAILS â&#x20AC;&#x201C; TREATED Introduction with designated caseworker : The caseworker informs the participant of being included in the program. He will continue to meet with the participant on a weekly basis and occasionally provide him with job referrals.
Diagnosis: Two meetings with an occupational trainer which include classification of participants according to employability and motivation levels; identification of specific barriers to employment; and designing a recommendation for a personalized program.
Allocation of participants: The head of employment office and\or the designated caseworker decide on the fitting course for each participant based on personal familiarity and the recommendation provided by the occupational trainers. 60
PROGRAM DETAILS â&#x20AC;&#x201C; TREATED Job placement focused preparatory workshop Purpose: preparation for job search phase; emphasis on skills and tools required to secure employment Target: high motivation participants; participants who completed the motivation workshop Duration: two hours, twice a week, for three weeks
Job search assistance (JSA) workshop Purpose: proactive job search Duration: two hours, twice a week, for two and half months
Process-focused preparatory workshop Purpose: preparation for motivational or JSA workshops. Identifying strengths, fostering motivation, work self efficacy and self-image Target: low\medium motivation participants Duration: two hours, twice a week, for three weeks
Motivational workshop Purpose: Provide a comprehensive picture of the modern workplace and the skills required to excel in it; focused on developing personal skills and a more proactive work attitude Target: low\medium motivation participants Duration: three to five hours, twice a week, for two months At the end of workshop: continued personal\group occupational training sessions 61
Job-Search Self-Efficacy Score 1-5 scale (from strongly disagree to strongly agree) 笨的 am confident in my abilities to search for a job 笨的 am confident in my ability to use the internet in order to find a job 笨的 am confident in my ability to write a resume 笨的 am confident in my ability to pass a job interview
Work Self Efficacy Score 1-5 scale (from not at all to very good) ✓Achieve goals that will be assigned ✓Respect schedules and working deadlines ✓Learn new working methods ✓Concentrate all energy on work ✓Collaborate with other colleagues ✓Have good relationships with my superiors ✓Be courteous to customers ✓Get to work on time
Self-Efficacy Score 1-5 scale 笨的 can always manage to solve difficult problems if I try hard enough 笨的f someone opposes me, I can find the means and ways to get what I want 笨的t is easy for me to stick to my aims and accomplish my goals 笨的 can usually handle whatever comes my way
ADHD score 1-5 scale ✓I find it hard to finish things I have started ✓I often get distracted by activity or noise around me ✓I often misplace or have difficulty finding things ✓I find it hard to concentrate at work ✓I often feel restless or fidgety ✓I find it hard to wait in line