A Profile of ISKUR Trainees

ISKUR Vocational Trainees
Profile, job-search behavior and expectations
June 15th 2011
This note presents the results of a baseline survey of about 5,300 people participating in a study to evaluate
the impact of vocational training provided by the Turkish Employment Agency (ISKUR). The study is a joint
effort by ISKUR and the World Bank aimed at identifying ways to improve vocational training programs in
the context of the rapid expansion of these programs. Evaluation participants are a representative sample of
ISKUR trainees in the 23 provinces selected for the study to represent the spectrum of labor market
conditions across Turkey. Data was collected between September 2010 and January 2011 before training
courses started. This is the most comprehensive information on ISKUR trainees to date.
The note provides a comprehensive analysis of ISKUR trainees in terms of (i) their characteristics and
differences relative to the unemployed and jobless population, (ii) their job search behavior, and (iii) their
expectation about ISKUR vocational training in terms of improved labor market outcomes. Ultimately, this
note aims to provide policy relevant information for the final evaluation of these programs in early 2012 by
(1) making an ex-ante assessment of how well ISKUR training is targeted (i.e. whether training is being
offered to individuals who could potentially benefit the most from it), and (2) analyzing the extent and nature
of the demand for ISKUR vocational training.
The focus of ISKUR trainings on women and youth seems appropriate from a policy perspective.
ISKUR trainees are mostly women (63%) and young (average age of 27). International evidence shows that
training programs are more cost-effective for youth than for adults. Targeting youth is also justified by their
sheer numbers in Turkey and their difficult transition into the labor market (60% of the 20-24 year olds with
less than secondary education are jobless—defined as not working and not going to school). And women
make up 72% of the jobless in Turkey and are significantly less educated than men, justifying the targeting of
training programs to women, particularly young women.
ISKUR trainings target well-educated unemployed people, who are not most in need of training.
About 74% of ISKUR trainees have at least secondary education, being significantly better educated than
even the working 20-34 year olds in the population (let alone the jobless). This is a reflection of both the
demand for such courses and their supply, as many of these courses are designed for people with medium
levels of education, and training providers tend to select individuals with higher levels of education. And yet
52% of the working age population (15-64) in Turkey has less than basic education (lower secondary
education), accounting for 64% of the jobless and 65% of informal workers. Even among 20-34 year olds, the
proportion with less than high school is 57%, accounting for 53% of the jobless and 54% of the informal in
that age group. These disadvantaged segments of the working age population are not being served by ISKUR.
And yet activating them, particularly the young, into productive employment is crucial to enhance productive
employment in Turkey over the medium term.
ISKUR trainees engage in little job search, although most of those who do tend to use ISKUR
employment services. Only about half of ISKUR trainees had been actively searching for a job during the 4
weeks prior to the survey (and before training), with men and more educated people being more active than
1
women and less educated people, respectively. Not surprisingly, the percentage of people looking for a job is
significantly smaller in poor provinces (where job opportunities are more limited) than in richer provinces.
Although the main channel for job search is family and friends, 74% of trainees searching for a job have used
ISKUR services. Access to ISKUR employment services is partly determined by the availability of ISKUR
offices across different province. ISKUR services are regarded as the second best channel to find wage
employment after applying directly to employers. The percentage of women with a positive view of ISKUR
services is higher than the percentage of men, even though women make less use of ISKUR services.
The results on job search behavior point to the need to encourage more job search and expand
employment services to make them front and central to activation efforts. International evidence shows
that this low job search activity before providing vocational training is not a cost-effective way to activate the
jobless. Job seekers need to be encouraged to look for a job and be assisted in this task through employment
services before they get any training. Employment services (job placement, counseling, job search assistance),
which are limited in Turkey, should be front and central to activation efforts: employment services are more
cost-effective than other activation measures and an effective screening device for identifying individuals who
require additional services (e.g. vocational training). For example, in the UK all registered unemployed people
are offered minimal employment services (job placement) and required to take individual actions to find a job
(the market test) (unemployment benefits and participation in activation programs in conditional on that)
before determining whether they need additional employment services and training (the most expensive
activation service).
ISKUR trainees attach a great ex-ante labor market value to ISKUR training, even relative to other
trainings, particularly among the low educated. Although the actual impact of vocational training will be
revealed once data from evaluation participants are collected approximately 1 year after the trainings take
place, it is informative to look at the expected impact of trainings (as perceived by trainees before starting the
training) and the reasons behind these expectations. About 80% of trainees believe ISKUR training will help
them do better in the labor market. The expected impacts after 1 year are very large: relative to no training,
ISKUR training is perceived to increase the probability of finding a job by 32 percentage points and increase
wages by 39%. Interestingly, trainees with less than secondary education completed, which make up the
minority of ISKUR trainees, have higher expectations than trainees with higher education levels. Although
this result is based on expectations, it casts some doubt on the appropriateness of focusing ISKUR trainings
on people with mid to high levels of education. And the expected impact of ISKUR training is higher than
that of other vocational training (13.6 percentage point increase in the probability of finding a job).
The high expected value of ISKUR training is explained by the high perceived quality and relevance
of trainings and value attached by employers to ISKUR certificates. About 94% of trainees think
ISKUR training improves job-readiness skills. Almost all trainees are confident about the quality of ISKUR
training, particularly if provided by public institutions (50% are very confident about their quality). And the
quality of ISKUR training is perceived to be higher than the quality of other training. In addition to quality,
92% of trainees value the fact that employers do value the certificate provided by ISKUR. Despite the
genuine value attached to ISKUR training, one third of trainees admit that the main reason for taking the
training is the stipend provided to trainees (this proportion decreases with the level of education of the
trainee). And the free nature of ISKUR training also matters: two thirds of trainees say they would not pay for
other training if not accepted in ISKUR training, being the main reason the lack of savings/cash and limited
access to credit.
2
Although based on expectations and perceptions, these results point to the genuine value attached
to ISKUR trainings, which provides a good base on which to build reforms to improve the
effectiveness of these programs. The expected impact of ISKUR training on the labor market (both
relative to no training and compared to other training) is consistent across age groups, gender, provinces,
training courses and previous experience with ISKUR trainings. Although, it is very possible that people
overestimate the impact of ISKUR training, they have enough information on the program to make a
judgment on whether it leads to better labor market outcomes. Likewise, trainees have better information
about ISKUR trainings than other trainings so they may tend to overestimate the relative quality of ISKUR
training. By the same token, however, their perceptions about the quality, relevance and employers’ value of
ISKUR training need to reflect point to the genuine value-added of ISKUR training. The actual evaluation of
ISKUR training to be conducted in early 2012 will provide the evidence on their labor market impact and
how it differs for difference groups of trainees and training providers.
The rest of this note is organized as follows. Section 1 puts this note into context and provides a quick
description of the sample and the data. Section 2 provides a profile of ISKUR trainees and compares this
profile to other groups in the labor force. Section 3 looks at the job search behavior of ISKUR trainees.
Section 4 looks at the expectations and perceptions about ISKUR trainings.
1. Introduction
Despite a strong performance after the crisis, the Turkish labor market continues to be characterized
by low activity rates and low labor productivity, constraining Turkey’s high growth potential. Less
than half of the working-age population (15-64) (WAP) are working (less than a quarter of working-age
women), while unemployment has remained above 10%. Job informality (defined as jobs without social
security benefits) has come down remarkably but it still affects 41% of workers, contributing to the low labor
productivity in Turkey relative to competitors. Low activity rates and labor productivity constrain Turkey’s
high growth potential: Differences in labor productivity account for 80% of the Turkey-EU15 GDP per
capita gap, while 18% of the gap is due to differences in labor force participation rates.
Figure 1: Few people working and their productivity is low
Less than half of people are working
Labor productivity (relative to the US): increasing
but remains low
Jan_05
Apr_05
July_05
Oct_05
Jan_06
Apr_06
July_06
Oct_06
Jan_07
Apr_07
July_07
Oct_07
Jan_08
Apr_08
July_08
Oct_08
Jan_09
Apr_09
July_09
Oct_09
Jan_10
Apr_10
July_10
Oct_10
Jan_11
50
45
40
35
30
25
20
15
10
5
0
Total
Female
Source: Labor Force Survey: Seasonally-adjusted monthly data; population 15+. Labor productivity: OECD.
3
Raising activity rates and productivity is a long term challenge, but decisive policy action today is
essential to take advantage of the closing demographic window. Urbanization, agricultural shedding and
the increasing WAP until 2020 will continue to increase the number of people (mainly youth and low skill
workers) looking for non-agricultural jobs, making it harder to reduce unemployment and informality in the
non-agricultural sector and to increase female labor force participation. And as agricultural shedding and
urbanization slow down, more women will join the non-agriculture labor force.
Skills upgrading of the labor force is central to enhancing productive employment in Turkey. Skills
are best acquired the first time around and reforms to improve skills while in school are most cost-effective.
However, the impact of these reforms will materialize in the long run and Turkish growth potential is
currently impinged by the large share of the working-age population (WAP) (15-64) that do not have the
basic skills to find a job or to get out of informal/low productivity jobs: 52% of the WAP has less than basic
education (lower secondary education), accounting for 64% of the jobless and 65% of the informal.
The expansion and improvement ISKUR vocational training since 2008 have been instrumental to
the government employment reform agenda. The 2008 labor package reduced employers’ social security
contributions and other non-financial labor costs, and allowed all registered unemployed to benefit from
Active Labor Market Programs (ALMP). The coverage of ALMP has expanded significantly since 2009
(mainly vocational training) (the number of beneficiaries is expected to be around 400,000 in 2011, from just
over 17,000 in 2006), and a number of reforms to improve the quality and relevance of vocational training
have been introduced, including the development of a national qualifications framework for vocational
education and training (ongoing), the introduction in 2010 of quality criteria in the selection of training
providers and the launch of the UMEM project, which will provide quality vocational training in technical
schools and internships at Union of Chambers and Commodity Exchanges of Turkey (TOBB) businesses. A
comprehensive Employment Strategy (ES) has been drafted, including measures to improve labor market
flexibility, worker protection and the link between vocational education and jobs. It is expected that the new
ES will include a number of measures to increase the coverage the ALMP and improve their effectiveness.
This note presents the results of a baseline survey of about 5,300 people participating in a study to
evaluate the impact of vocational training provided by the Turkish Employment Agency (ISKUR).
The study is a joint effort by ISKUR and the World Bank aimed at identifying ways to improve vocational
training programs in the context of the rapid expansion of these programs in Turkey. In particular the study is
designed to answer the following questions: (1) What is the average impact of ISKUR training on the labor
market (as measured by the likelihood and quality of employment)? (2) Which trainees benefit the most from
training (in terms of gender, age, level of education and skills, work experience etc.)? (3) What are
mechanisms/processes through which training affects labor market outcomes (e.g. improved skills, reduced
search costs etc.) What provider characteristics make training most effective?
The evaluation sample. The sample of evaluation participants includes approximately equal numbers of
individuals that were randomly selected into taking vocational training (treatment group) and individuals that
were randomly selected out of training (control group). This experimental design required focusing the
evaluation on training courses that were sufficiently oversubscribed (i.e. the number of people eligible and
interested in these courses is at least twice the number of training spaces available). Evaluation participants
went through the standard selection process for ISKUR training until the final randomization: they were (i)
eligible for and interested in training and (ii) selected by the training providers among a larger pool of eligible
applicants. The data was collected between September 2010 and January 2011 before the training started. A
4
such, evaluation participants are a representative sample of ISKUR trainees in the 23 provinces selected for
the study to represent the spectrum of labor market conditions across Turkey (from an initial sample of 39
provinces with at least two oversubscribed courses in 2009) and the selected (oversubscribed) courses (130).
Baseline data was collected from 5,318 out of 5,700 evaluation participants.
The randomization of evaluation participants into treatment and control groups was successful,
which validates the impact evaluation strategy. The Annex 1 shows that, across many different
characteristics, there are no statistically significant differences between the treatment and control groups. This
validates the evaluation strategy that will compare the labor market outcomes of treatment and control groups
and attribute the difference to ISKUR training and ISKUR training only. This evaluation will mainly draw
from a follow survey of evaluation participants approximately one after the trainings have been completed
(early 2012). The final evaluation report will be available by June 2012. This is the first rigorous impact
evaluation of a large-scale publicly provided training program in a middle income country context. Further
details on the design and implementation of the impact evaluation study are available in a companion note.
Figure 2: Evaluation sample (baseline survey)
Selected provinces for the evaluation
Regional distribution of evaluation sample
Distribution of evaluation participants
ANKARA
ANTALYA
182
BAYBURT
283
DENIZLI
123
DIYARBAKIR
1771
DUZCE
ELAZIG
693
ERZURUM
ESKISEHIR
GAZIANTEP
HATAY
ISPARTA
327
ISTANBUL
IZMIR
KAYSERI
106
179
111
266
158
100
150
226
100
137
111
100 100
308
131
100
163
KIRIKKALE
KOCAELI
MANISA
MUS
SAKARYA
TEKIRDAG
TRABZON
USAK
Evaluation courses
2
3
3
2222
11 1
3
1
24
4
5
7
3
38
5
6
9
6
Computer /Computer programming
Accounting professionalist/Computerized Accounting
Babysitter
Cashier
Foreign Trade and Customs Professional
Fitter(natural gas)/Plumbery
Old and Sick People Nurse
Welder, Gas Arc
Retailing and Merchandising/Salesperson
Cook
Weaver, carpet-ehram
Modelist/Stilist
Coiffeur and Hair care/Skin care and beauty
Operators (forklift/sewing machine)
Medical Secretary
Human resources Management
Manufacturer, furniture
Applied Basic Electronics/Electronic technicians
Finalcut
Waiter, service
Moulder, grouting
5
2. Who is taking ISKUR trainings? A Profile of ISKUR Trainees
Most ISKUR trainees are women and young. About 61% of individuals in the evaluation sample are
females. This is partly explained by the courses offered: Out of the 130 courses in the evaluation, 12 courses
only have female trainees (babysitting, weaving, hair care, and sewing machine operator) and 17 courses have
no female trainees (applied basic electronics, welders, plumber, furniture manufacture, natural gas fitter, and
forklift operator). But women also make the majority of trainees in a wide range of other courses, including
computerized accounting, computer manager, foreign trade and customs professional, computer-aided
design, computer network design, salesperson, and web designer. For example, in accounting professionals
and computerized courses women account for 72%, 77% in Cashier/Clerk/Bookkeeping and 90% in medical
secretary or apparel related occupations (e.g. stylist). The average age among ISKUR is 27 and 60% are aged
20-29. Men tend to be a bit younger than women (26 versus 28). There is some variation in age across
courses, with the average age being lower in occupations were women are more dominant. Because of their
youth, only 12% of trainees are household heads and 34% are married.
Most trainees have completed secondary education and have some degree of
specialization/technical background. About 76% of men and 73% of women have completed at least
secondary education, and 30% of men and women have completed at least 2 years of tertiary education.1 This
high educational achievement is partly due to the large presence of youth in the sample. Younger cohorts
have indeed higher educational achievements than older cohorts. But the percentage of people with
secondary education completed is still high (relative to the population) among 34 years old and older (65%
for men and 59% for females). Most trainees have some degree of specialization or technical background.
The most common technical backgrounds are computer technologies/programming and accounting. About
one third of trainees have a background in the field of training they are applying to, while others have general
degrees (e.g. business) and are interested in ISKUR training to acquire some specialization. Interesting, 20%
of males and 29% of females have already taken vocational trainings over the past 5 years, mostly from
ISKUR (47%), and in areas closely related to the training they are applying to now.2
Figure 3: High educational achievement among ISKUR trainees
35
30 29 29
30
29
31 30
25
20
13
15
10
18
16
16
11
13
13
15
8
5
0
Primary or less
Lower sec.
Men
High school
Women
Voc HS
College
Total
About 15% of trainees are still in school.
Because ISKUR trainees cannot enroll in trainings if they had taken a course in the last two years, most of the trainings
tend to have been completed more than two years ago, between 2008 and 2010.
1
2
6
ISKUR trainees have had limited work experience, mainly in the form of formal private full time
wage employment. A large share of applicants (37%) has never worked before, and among those with prior
experience 20% have worked less than 1 year (average experience is 5 years). Men are significantly more likely
to have worked before than women (71% versus 53%). There is significant regional variation: Bayburt and
Mus have particularly low shares of trainees with work experience, while about 80% of trainees in Duzce and
Manisa have worked before. The limited work experience of trainees is partly explained by the age and gender
composition of trainees. More than 90% of people with previous work experience worked as private sector
wage employees and about 58% had social security coverage. Applicants worked on average 54 hours/week
and received an average of TL 677/month.
ISKUR trainees are significantly younger, have less work experience and are more educated than the
average unemployed or jobless individual in urban areas.3 Women are overrepresented among ISKUR
trainees compared to the pool of unemployed but not relative to the jobless (defined as not working and not
going to school—a measure of human capital utilization) (Table 1). ISKUR trainees are significantly younger
and have less work experience than the average unemployed or jobless individual. Interestingly, ISKUR
trainees are less likely to be looking for a job than the average unemployed in urban areas. The most striking
difference is that ISKUR trainees are significantly more educated than the average unemployed, jobless or
even employed individual in urban areas. And this is not only due to the younger profile of ISKUR trainees,
as the differences remain even after limiting the urban sample to 20-29 year olds. This difference is explained
by two factors: (i) the demand for training courses, as people self-select into these courses; (ii) and supply, as
many courses are designed for people with medium levels of education and training providers tend to select
individuals with higher levels of education. Interestingly, the incidence of job informality among ISKUR
trainees (for their most recent job) is higher than that for the average wage employee (42% versus 24%).
Table 1: Profile of ISKUR trainees relative to unemployed and jobless
% Female
% Married
% Aged 15-24
% High school or more
% High school or more (20-29)
% Worked before
% Looking for a job in last 4 weeks
ISKUR trainees
All Men Women
63
34
23
41
45
50
42
74
76
73
86
86
86
61
71
56
49
58
43
Unemployed (urban)
All Men Women
30
49
54
38
31
28
38
42
34
59
60
51
75
90
93
81
89
90
88
All
72
75
15
24
37
55
14
Jobless (urban)
Men Women
72
16
31
53
94
35
76
15
21
33
40
6
Source: Data for the unemployed and jobless is from the 2009 Labor Force Survey (urban sample). Jobless is defined as not working
and not going to school.
3. Job Search Behavior and the role of ISKUR
Only half of ISKUR trainees are actively searching for a job. Over the last four weeks prior to the face to
face survey(and before training), only about half of ISKUR trainees not currently working or going to school
were actively searching for a job. The share of applicants actively searching increases significantly with the
level of education: while 60% of those with at least some tertiary education are actively searching for a job,
only 37% of those who have completed primary education are searching for a job (Figure 4). This share also
The comparison with the unemployed and jobless is restricted to urban areas to make it comparable to ISKUR trainees
are training is provided in predominantly urban areas.
3
7
varies significantly across provinces: Bayburt and Mus have the lowest rates of job search (2% and 11%,
respectively) while Duzce and Tekirdag and Izmir are above 70%. Men are significantly more likely to be
looking for a job than women (61% versus 42%), a difference that remains even after controlling for
differences in education level, indicating that women face greater obstacles to search due to their household
responsibilities.
Figure 4: Limited job search among ISKUR trainees
(% actively looking for a job in the last 4 weeks, by gender and level of education)
80
72
70
55
60
50
40
30
39
53
45
43
57
60 58
63
49
42
39
32
29
20
10
0
Primary or less
Lower sec.
Men
High school
Women
Voc HS
College
Total
Trainees engaged in job search tend to devote significant time to job search and are looking for full
time wage employment. Most trainees looking for a job started searching in 2010, when they registered
with ISKUR (60%), although many a sizable number (25%) started searching before 2008. Job search is an
intensive activity: on average men search for about 10 hours per week, while women search for 15 hours per
week. Most job seekers looking for full time wage jobs (80%), even among women (79%). The reservation
wage of ISKUR trainees is high: only 38% would accept a monthly wage of TL 1,000 (the average monthly
wage in Turkey) (even only 40% among those with less than high school). About 40% would not accept a job
without social security (20% among those with less than high school) and, among those who would, the
reservation wage would be TL 1,800 per month (with no significant differences across levels of education).
Only 15% of the applicants are prepared to migrate if they do not find a job next year. A number of factors
may help to explain the high aspirations of ISKUR, including educational achievement, family background
(they come from families with significantly higher per capita household income than the average unemployed
in urban areas), and work experience (only half of jobseekers come from a previous work relationship).
Most job seekers use ISKUR services to help them find a job. Although the most common channel to
search for a job are family and friends (87% of jobseekers use it), the second most popular channel is ISKUR
(74%), followed by direct contact with employers (64%) (Figure 5).4 Only 12% have used private employment
agencies, reflecting in part their limited availability. Internet is also frequently used and, to a smaller extent,
newspaper advertisements. ISKUR tends to be less frequently used by women than men (72% versus 78%).
And job seekers with primary education or less tend to rely most heavily on ISKUR services, while college
graduates use it the least. There are significant differences in the use of ISKUR services across provinces,
reflecting in part the availability of ISKUR services: in Duzce, Gaziantep, Hatay and Tekirdag more than 92%
of job seekers use ISKUR, while in Mus, Eskisehir and Elazig the percentage is between 30% and 40%.
4
Note that multiple channels can be used.
8
ISKUR services are positively valued by job seekers. Most job seekers believe that applying directly to
employers is the best channel to find a wage job, with ISKUR ranking immediately after: 28% of jobseekers
view ISKUR as the best channel to search for a wage job. Family and friends, although more frequently used
than ISKUR, is viewed as a much less effective channel. Interestingly, although women use ISKUR services
less they tend to value them higher than men (32% versus 23%). There is also some regional variation,
possibly indicating the varying quality of services across provinces: more than 40% of jobseekers in
Gaziantep, Antalya, Hatay and Usak value ISKUR services as the best way to find wage employment, while
12% or less of jobseekers in Ankara, Trabzon and Kirikkale do.5
Figure 5: ISKUR services are used by most job seekers
(% use of different job search tools, by gender)
100
90
80
70
60
50
40
30
20
10
0
89 86 87
79
68
60
64
71
75
73 75 74
50
43 46
12 11 12
Employer
ISKUR
Friends and
Private
Printed ads
relatives employment
agencies
Men
Women
Internet
Total
4. Expectations about ISKUR Vocational Trainings
ISKUR trainees attach a great ex-ante labor market value to ISKUR training, even relative to other
trainings, particularly among the low educated. This section looks at the expected impact of trainings (as
perceived by trainees before starting the training) and the reasons behind these expectations. About 80% of
trainees believe ISKUR training will help them do better in the labor market. The expected impacts after 1
year are very large: relative to no training, ISKUR training is perceived to increase the probability of finding a
job by 32 percentage points (Table 2 and Figure 6) and increase wages by 39% (42% within two years). The
expected impacts are larger for women than for men, but there are no systematic differences by age.
Interestingly, trainees with less than secondary education completed, which make up the minority of ISKUR
trainees, have higher expectations than trainees with higher education levels.6 Although this result is based on
expectations, it casts some doubt on the appropriateness of focusing ISKUR trainings on people with mid to
high levels of education. And the expected impact of ISKUR training is consistently higher than that of other
vocational training (13.6 percentage point increase in the probability of finding a job). There is variation
across provinces (trainees in Ankara, Antalya and Isparta have the highest expected impacts) and courses
(waiter, cook, nurse and hairdressing have the highest expected impacts).
Many reviews show that employment services are the fundamental services offered by employment agencies in OECD
countries. In spite of this employment services currently play a very minor part in ISKUR’s portfolio. However,
expanding and upgrading their role can help improve the overall effectiveness and efficiency of ISKUR’s trainings.
6 A similar pattern appears for the impact on wages although the differences are smaller.
5
9
Table 2: Large expected impact of ISKUR training on the labor market
(Probability of finding a job after 1 year under different scenarios)
Less than
high school
High school
Voc. HS
College
No
training
27
All
Training
31
33
35
41
ISKUR
training
62
No
training
24
45
46
48
64
65
64
29
30
35
Women
Training
38
ISKUR
training
59
No
training
35
44
44
48
63
65
65
33
37
36
Men
Training
48
ISKUR
training
68
46
48
48
64
65
62
Figure 6: Large expected impact of ISKUR training on the labor market
(Implicit expected impacts relative to no training and other training, in percentage points)
40
35
33
35
32
29
30
25
21
20
19
19
16
15
10
5
0
Less than high
school
High school
Impact relative to no training
Voc. HS
College
Impact relative to other training
The high expected value of ISKUR training is explained by the high perceived quality and relevance
of trainings and the value attached by employers to ISKUR certificates. About 94% of trainees think
ISKUR training improves job-readiness skills and the knowledge of the profession (Figure 7). About one
third of trainees pick a specific course because they already have some background in that area, and so they
aim to upgrade their skills to help them find a job in that area. About half of trainees pick the training course
because they think there are many jobs available in that area, and so they aim to get the skills necessary to get
a job in that areas. Almost all trainees are confident about the quality of ISKUR training, particularly if
provided by public institutions (50% are very confident about their quality). And the quality of ISKUR
training is perceived to be higher than the quality of other training (particularly for privately-provided
training) (Figure 8). In addition to quality, 92% of trainees value the fact that employers do value the
certificate provided by ISKUR. Despite the genuine value attached to ISKUR training, one third of trainees
admit that the main reason for taking the training is the stipend provided to trainees (this proportion
decreases with the level of education of the trainee). And the free nature of ISKUR training also matters: two
thirds of trainees say they would not pay for other training if not accepted in ISKUR training, being the main
reason the lack of savings/cash and limited access to credit.
10
Figure 7: ISKUR trainees attach genuine labor market value to ISKUR training
(% agreeing ISKUR training is useful for different things)
100
90
80
70
60
50
40
30
20
10
0
95
94
92
90
79
46
41
38
37
34
28
11
Job-readiness
skills
Knowledge of
the profession
Employers'
value of
certification
Agree
Networking to
find jobs
Changing
profession
Stipend
Strongly agree
Note: Agree category includes agree and strongly agree
Figure 8: The quality of ISKUR training is perceived to be higher, particularly if publicly provided
(% confident that the quality of different type of training will be of good quality)
120
100
97
94
87
80
60
67
53
36
40
43
21
20
0
ISKUR, public
Other, public
Confident
ISKUR, private
Other, private
Very confident
Note: Confident includes somewhat confident and very confident
Although based on expectations and perceptions, these results point to the genuine value attached
to ISKUR trainings. The expected impact of ISKUR training on the labor market (both relative to no
training and compared to other training) is consistent across age groups, gender, provinces, training courses
and previous experience with ISKUR trainings. Although, it is very possible that people overestimate the
impact of ISKUR training, they have enough information on the program to make a judgment on whether it
leads to better labor market outcomes. Likewise, trainees have better information about ISKUR trainings than
other trainings so they may tend to overestimate the relative quality of ISKUR training. By the same token,
however, their perceptions about the quality, relevance and employers’ value of ISKUR training need to
reflect point to the genuine value value-added of ISKUR training. The actual evaluation of ISKUR training to
be conducted in early 2012 will provide the evidence on their labor market impact and how it differs for
difference groups of trainees and training providers.
11
Annex 1: A Successful randomization of the ISKUR vocational trainings
The high capacity of the ISKUR IT services, together with the strong implementation capacity of
the provincial offices, allowed that the randomization of the applicants into the ISKUR vocational
trainings to be carried out by computer. The impact evaluation will compare the labor market outcomes
for the trainees and the non trainees before and after the completion of the courses. At the core of this
identification strategy is the capacity to randomize the individual participation within each of the courses. The
randomization worked as follows: each training provider provided a list of up to 50% more applicants than
they had capacity to train. These individuals were then randomly offered a position in the training course
using the special information system set up to register applicants into the program. As a result, any
differences between treatment and control groups in the full assigned population are due purely to chance.
Because the effects of the trainings in the applicant’s labor market outcomes are likely to be different
depending on the vocational courses, gender, or age of the beneficiaries, the randomization was stratified by
these variables.7
Although the response rate for the face to face interviews was high, the 5% non-response may have
affected the balance between trainees and non-trainees. Since less than 100% of those assigned to
treatment and control completed the face to face baseline survey conducted between September 2010 and
April 2011, it is worth checking whether treatment and control groups have balanced observable
characteristics. The final sample available has a total of 5,318 individuals. Table A1 presents the means for
treatment and control group and a p-value for testing the difference in means for 37 different characteristics.
The findings clearly show that the sample is well-balanced on a number of important characteristics
that likely also will affect the labor market outcomes of the applicants to the ISKUR trainings. In
particular, the treatment and control groups do not statistically differ in terms of gender, educational levels,
participation in prior training courses, ever having worked for pay, beliefs about the likelihood of being
employed in one year with and without taking training courses, household asset ownership and household
income, and numeracy and risk attitudes. However, there are statistically significant, although small in
absolute magnitude, imbalances in some variables: Those in the treatment group are slightly younger on
average, and get slightly higher scores on a Raven test (a measure of non-verbal IQ). Since some of the
questionnaires were completed after individuals knew their treatment status, some of the other small
differences might reflect changes in attitudes and knowledge as a result of treatment assignment. This could
explain, for example, why the treatment group is slightly less likely to have sought work in the last 4 weeks, to
say they would accept a job at 600 lira, to have worked for pay in the last 4 weeks, and to know where a
similar course could be taken privately – individuals who have just found out they are taking the course might
be less likely to look for work or to work, and might learn the course is offered privately also. The small
significant difference in mental health could also arise from a short-term increase in stress from those starting
the new course. As a result, the randomization was successful. It created two largely similar samples in terms
of observable characteristics. This finding is reassuring that any differences in outcomes to be observed in the
tracking (summer 2011) and in the follow up surveys (2012) will be solely driven by the ISKUR vocational
trainings.
7
See Donmez et al (2011).
12
Table A1: Test of Randomization for ISKUR Sample
Means by Assignment Status Diff. in Means
Control
Treatment
28.24
27.92
Female
0.63
0.63
Youth (age <25)
0.38
0.40
Male Youth
0.16
0.17
Female Youth
0.21
0.23
Male >=25
0.21
0.20
Female >=25
0.42
0.40
Less than High School Education
0.26
0.26
University Education
0.14
0.14
11.36
11.42
Has done prior training course in last 5 years
0.26
0.26
Household Size
4.02
4.02
Married
0.32
0.32
Has been unemployed since 2009
0.40
0.38
Receives unemployment insurance
0.05
0.05
Has sought work in last 4 weeks
0.51
0.46
Says would accept a job at 600/month
0.34
0.31
Says would accept a job at 1000/month
0.59
0.58
Worked for pay in last 4 weeks
0.14
0.11
Has ever worked for pay
0.60
0.59
Total years working for pay
3.36
3.28
Agrees that mainly interested in course for stipend
0.35
0.33
Know where course could be taken privately
0.34
0.37
Percent chance would pay for course if not selected
33.51
34.94
Percent chance will be employed if take another course
45.05
45.07
Percent chance will be employed without course
31.79
31.22
Percent chance will be employed if take ISKUR course
63.64
63.55
Risk-seeking score (higher = more risk seeking)
6.52
6.42
Financial risk-seeking score (higher=more risk seeking)
4.77
4.76
Has health problem that prevents physical work
0.02
0.02
Mental health index (higher = worse mental health)
11.72
11.98
Durable asset index
-0.00
0.00
Household has a computer
0.68
0.69
Age
Number of Years Schooling
Household has internet connection
0.53
0.54
14,360
14,253
Gets all 4 numeracy questions right
0.57
0.58
Raven test score
5.64
5.94
Household annual income
p-value
0.096
0.741
0.066
0.811
0.052
0.537
0.187
0.938
0.961
0.548
0.543
0.952
0.984
0.126
0.907
0.000
0.026
0.229
0.008
0.439
0.563
0.234
0.037
0.125
0.984
0.395
0.893
0.164
0.960
0.720
0.003
0.978
0.304
0.240
0.761
0.420
0.001
Source: Authors. Based on the 5,318 observations of ISKUR applicants from the ISKUR-WB baseline survey (2011).
Note; Differences in treatment and control tend to be not statistically significance across a wide range of socio economics characteristics.
There are two exceptions: age and probability of being actively looking for a job)
13