Estimating the Employment and Educational Effects of

Vocational Training: the Role of School Quality

Cristiano C. Carvalho† Raphael Corbi†† Rodrigo De-Losso‡‡

March 19, 2021

Abstract

This paper provides new evidence of the short and long-run effects of vocational
training (VT) on labor market and educational outcomes, with a particular interest
in how school quality may confound estimates. VT schools may differ from regular
schools not only in terms of type of training, but also in the availability of resources.
We take advantage of a particular institutional arrangement in the state of Paraná,
Brazil, where a single private institution named FIEP provides both VT and regular
education under two separate but closely related entities, while non-FIEP institu-
tions provide regular education. As both VT and regular schools within FIEP have
more resources and better inputs than non-FIEP schools, simply comparing out-
comes of VT and regular students can be misleading even if students were assigned
randomly to schools. Using a unique survey applied to different cohorts of high
school graduates, we show that quality plays an important but nuanced role when
comparing the effects of general and VT in the short and long run. In particular, our
propensity score estimates indicate that FIEP VT graduates have higher short-run
employability than both FIEP and non-FIEP non-VT students. However, non-VT
graduates from the better-funded FIEP system are more likely to continue to higher
education, so that the short-run employment effect all but dissipates as they enter
the labor force in the long-run.

Keywords: Vocational education, short and long-run labor market outcomes, higher
education.

†
Corresponding author. University of Michigan. E-mail: crcarval@umich.edu
††
University of São Paulo. E-mail: rcorbi@usp.br
‡‡
University of São Paulo. E-mail: delosso@usp.br
We are grateful to FIEP, Instituto Paraná, Renata Narita, Leonardo Rosa, and conferences and
seminars participants for helpful comments. The authors acknowledge financial support from FIPE.
Raphael Corbi thanks the University of Chicago for their hospitality where parts of this work were
completed.

1
Electronic copy available at: https://ssrn.com/abstract=3808661
1 Introduction

Vocational education training has great potential for developing and improving
specific skills in the workforce. The benefits of this modality are often associated
with a smoother transition between school and the labor market, an increase in
labor productivity and a specific labor training (Souza et al., 2015). These positive
short-term effects of vocational education on income and on the likelihood of get-
ting a job are well documented in the literature (Malamud and Pop-Eleches, 2010;
Tansel, 1998; Hanushek, Woessmann, and Zhang, 2011; Almeida et al., 2015; Costa,
Fernandes, and Vasconcellos, 2010; Assunção and Gonzaga, 2010; Neri, 2010).

On the other hand, the effects on wages and employment are less clear in the
long-term. The rapid pace of technological change may favor skills that are more
adaptable and flexible in the long-term, making the specific skills developed in the
vocational training become obsolete more quickly than general skills learned in regu-
lar courses. Such trade-off brings up the question of what the effects on employment
and income over an individual’s life cycle are. A considerable number of papers in
the literature presents evidence that the short-term effects are usually greater than
the long-term and, in general, the whole cycle net effects are positive (Hanushek
et al., 2017; Brunello and Rocco, 2017; Golsteyn and Stenberg, 2017; Oswald-Egg
and Renold, 2021). Important exceptions are Attanasio et al. (2017) and Kugler
et al. (2020) who find positive long-lasting effects of vocational training on labor
market and educational outcomes in Colombia.

Another important branch of the literature analyzes the effects of vocational

training on educational outcomes. The main hypothesis is that, by increasing stu-
dents’ motivation and engagement, vocational training reduces dropout and increase
the probability of graduating from high school. Although most of the literature evi-
dence are based on programs implemented in developed countries, some recent works
that focus on developing countries provide comparable findings. For instance, Elac-
qua et al. (2019) find that vocational training reduces the dropout probability and
increases the math and Language standardized test scores in Brazil. Importantly to
our setup, they are also more likely to attend schools with better quality.

This paper contributes to this literature by uncovering new evidence of the

effects of vocational training on labor market and educational outcomes, with a

2
Electronic copy available at: https://ssrn.com/abstract=3808661
particular interest in how school quality may confound estimates. VT schools may
differ from regular schools not only in terms of type of training, but also in the
availability of resources. More specifically, quality improves the chance of a regular
education graduate to enroll in higher education and increases the probability of
a vocational training student to transit directly to the labor market in the short-
term. Therefore, it affects these two groups differently in an important channel
of skill accumulation: enrollment in higher education. Hence, simply comparing
vocational and non-vocational education students’ outcomes may be misleading due
to differences in institutional quality, even if students were assigned randomly to
schools.

We take advantage of particular institutional arrangement in the state of

Paraná, Brazil, where a single institution named FIEP (Paraná State Industries
Federation) provides vocational and non-vocational education under two separate
but closely related entities, namely SENAI (Industry’s national learning service) and
SESI (Industry’s social service). Both entities have more resources, better teachers
and infra-structure than public non-FIEP schools which provide only regular edu-
cation. Our analysis compares SENAI’s technical courses, designed to provide spe-
cialized skills for technical career using hands-on manufacturing training, to SESI’s
non-vocational high school and as well as to regular public high schools of the same
areas.

Based on two unique field surveys conducted by the Paraná Research Institute,
we focus on these educational modalities, SENAI and SESI, to compare the effects
of the vocational and traditional courses offered by the FIEP System on the em-
ployability, wages, overall satisfaction, and enrollment in higher education. The first
survey was held from August to October 2018 and captures the short-term effects
by interviewing a sample of students who graduated from 2015 to 2017. The second
survey, held from May to July 2019, captures the long-term effects by interviewing
a sample of students who graduated from 2011 to 2014. In both surveys, a sample
of students who attend mainly the regular public high school system (not FIEP)
was also interviewed. The questionnaires covered the following topics: general char-
acteristics of the interviewees, employability and performance in the job market,
satisfaction with their job, and with their professional status and educational back-
ground and enrollment in higher education.

3
Electronic copy available at: https://ssrn.com/abstract=3808661
 We proceed in two steps steps. We first select our sample by using a propensity
score matching to ensure comparability between students who graduated from the
FIEP System - including both vocational training and high school - and students
mainly from regular public schools.1 As a second step, we use the selected sample
to compare the results between FIEP and non-FIEP students, allowing for different
effects between vocational and non-vocational training among the FIEP graduates.
This two-step procedure allows us to (i) assess the overall effect of vocational training
(usually estimated in the Brazilian literature as discussed below) by comparing
outcomes of SENAI’s technical students with non-FIEP graduates, as well as (ii)
disentangling the overall vocational training effect due to differences in management
and institutional quality and to the actual vocational aspect of training.

Our main findings show that accounting for institutional quality matters for es-
timating the labor market and educational effects of vocational training. Regarding
short-run employability, we find that vocational training significantly increases the
probability of being employed in any job and also employed in formal jobs after we
control for quality. This comes at a price of decreasing the share of full-time stu-
dents. Both effects still significant in the long-run but controlling for quality makes
them smaller. In other words, accounting for quality has different implications over
time, it increases the employment effects in the short-term, but attenuates the ef-
fects in the long-term. These estimates are consistent with students that finished
high school in a better institution having a grater probability of going to higher
education, decreasing their likelihood of being employed in the short-run. On the
other hand, in the long-run most of them have already finished higher education
and are back in the labor force.

The effects on income are also persistent over time. In both time periods,
addressing the potential issue of differences in quality of the institutions has the
same implications, it increases the income results of vocational training. Former
students of SENAI’s technical courses are more likely to be employed in higher
paying jobs, both in the short- and long-run, and the effects do not seem to reduce
in the long-run. When we simply compare the SENAI’s technical courses former
students with students who have not been enrolled in the courses offered by the
1
Our main specification relies on the nearest neighbor propensity score matching with replace-
ment. As a robustness exercise, we also use an OLS model and a kernel propensity score matching
strategy

4
Electronic copy available at: https://ssrn.com/abstract=3808661
FIEP System, we still find a smaller effects in the short-run, but no significant effect
is found in the long-run.

Finally, the surveys include some questions that allow us to test the effects
of vocational training on the satisfaction with activity sectors, professional status
and educational background. After we account for quality, we find that SENAI’s
technical courses former students have a higher probability of being employed in
the activity sector that they consider to be the most beneficial for themselves in
the short-run and report lower satisfaction with their educational background both
in the short and long-term. This result is also linked with the fact that vocational
training students also have lower probability of enrolling in higher education in the
two analyzed periods. Without addressing the potential quality issue, none of these
results are observed.

We believe that comparing these two specific education modalities provide us

with a complementary evidence of the vocational training effects on the labor market
outcomes to the one that is provided in the literature. Attanasio, Kugler, and
Meghir (2011), Attanasio et al. (2017) and Kugler et al. (2020) causally analyze the
effects of large vocational program in Colombia by comparing participants with non-
participants selected using a randomized trial. Instead of comparing participants
with non-participants, we compare vocational training considering regular education
as the outside option, which is an important measure to decide between public
policies that focus on each of these modalities.

More precisely, the vocational program evaluated by Attanasio, Kugler, and

Meghir (2011), Attanasio et al. (2017) and Kugler et al. (2020) was targeted to young
people between the ages of 18 and 25, who were unemployed or out of school and
who were placed in the two lowest deciles of the income distribution. By comparing
participants with non-participants selected using a randomized trial, Attanasio et al.
(2017) finds a positive and persistent effect in employment and earnings. Therefore,
in this context, their findings indicate that vocational training can be an important
safety net program to support vulnerable groups. Our paper is interested in the
lifetime trade-offs of choosing vocational training over traditional education for a
broader range of students, the ones that are about to complete or that have just
completed high-school. The main idea is that the short-term benefits of vocational
training could change future decisions about skill accumulation, such as enrollment

5
Electronic copy available at: https://ssrn.com/abstract=3808661
in higher education.

In addition, we contribute to the existing literature on vocational training in

Brazil. Past studies relied on the National Household Sample Survey (PNAD) data
to measure the effects of vocational training (Almeida et al., 2015; Costa, Fernandes,
and Vasconcellos, 2010; Assunção and Gonzaga, 2010; Neri, 2010). The empirical
strategy consisted of using a propensity score approach on the individuals’ observable
characteristics, without accounting school quality due to lack of data. We address
the potential issue of differences in quality across institution by taking advantage
of a centralized management system. Our survey also allows us to analyze the
effects for different cohorts and provide a measure of short- and long-term effects
of the vocational training. Recent studies that better address the potential omitted
variables bias using experimental analysis do not make the distinction between time
period frames, and, therefore, provide no answers about the vocational training
effects over time (Camargo et al., 2018).

Our estimates also provide a relevant input for policy design in a moment when
vocational training has been gaining relevance in the country. In the recent years
Brazil has increased significantly the amount invested in vocational education. The
Federal Government expenditure has increased 5 times from 2003 to 2016 (0.04% to
0.2%) and the number of students enrolled in technical education during high school
increased by 45% between 2007 and 2013 (Elacqua et al., 2019). Vocational training
is offered by public and private institutions in Brazil. A major role is played by the
entities such as FIEP System across most other states, which are private institutions
partially financed with public resources. They are privately managed but they are
allowed to collect mandatory taxes on the payroll of firms in their activity sector.
Their main purpose is to provide technical and vocational training to address the
specific demands of skilled labor in their activity sectors.

This paper is organized into 5 sections beyond this Introduction. The following
section briefly describes the institutional background behind the FIEP education
system. Section 3 presents in detail the database and the methodology used in this
paper. Sections 4 and 5 present the results of participation in FIEP System courses
and robustness checks, respectively. Finally, section 6 concludes this paper.

6
Electronic copy available at: https://ssrn.com/abstract=3808661
2 Institutional Background: the FIEP Educational System

In Brazil, the vocational education training courses are offered by public and
private institutions. Even though, public institutions enroll more students than
private ones, a significant part of the technical and vocational education is provided
by the entities that constitute the so-called S System (private institutions partially
financed with public resources.): provide 43% of professional and technical education
in Brazil (Souza et al., 2015).

The S system is composed of a group of private and nonprofit entities classified

as autonomous social service. Those entities are sector specific and have two main
goals: i) to improve the quality of life of workers in each sector covered, ii) to provide
technical and vocational education to meet the demand for skilled work in their
respective sectors. The S System as a whole has 10 entities in total, namely: SENAI,
SESI, IEL, SENAC, SESC, SEBRAE, SENAR, SENAT/SEST, and SESCOOP. The
entities are privately managed, but they are allowed to collect mandatory taxes on
the payroll of firms in their activity sector. (Souza et al., 2015)

The structure of the S system varies across states in Brazil. In the state of
Paraná, some entities that integrate the S System, more specifically the Industry’s
national learning service (SENAI) and the Industry’s social service (SESI), work
collaboratively with the entity that represents the industrial business category (FIEP
- Paraná State Industries Federation). Those entities together constitute the FIEP
System. The advantage of having this sort of integration between the S System
and the entity that represents the industrial business category is to be able to offer
vocational courses that will address specific needs of the market.

This article focuses on two specific courses offered by the FIEP Educational
System: SENAI’s technical courses, and SESI’s high school course. Those modalities
can be divided in vocational and traditional education, since SENAI’s courses are
designed to prepare and train their students in specialized skills for specific career
fields using hands-on training on industry, and SESI’s high school focuses on broader
theory. The technical courses that we evaluate in this paper are not integrated in
the high school, they either require the student to be enrolled in (concomitant) or
have finished (subsequent) high school, but they do not offer it jointly with their

7
Electronic copy available at: https://ssrn.com/abstract=3808661
vocational training2 . It is also worth noting that even thought SESI’s high school is
more focused on broader theory, their students receive discounts in case they enroll
in a SENAI’s course.

It is important to highlight that private high-schools are widely recognized in

Brazil by their average superior quality. Therefore, comparing vocational train-
ing, that is often privately provided, with students that have not been enrolled in
the FIEP system could lead to misleading results, since participation in vocational
training would be confounded with better inputs. Table 1 presents the average re-
sults in ENEM, a standardized exam required to enroll in higher education, and the
mean difference test between the SESI high-school and public and private school in
Paraná. The descriptive results presented in table 1 provides some evidence that
SESI high-schools have on average a better performance in standard exams than
public schools and a worse performance than private schools. Unfortunately, these
variables are not available to SENAI, since we are focusing on concomitant and sub-
sequent courses, and we have to rely on the hypothesis that the quality inputs might
not significantly differ between SESI and SENAI since they are centrally managed
by the same institution. We also present some infrastructure measures in the ap-
pendix (tables A1 and A2) that show that SESI high schools usually have better
infrastructure than public high schools. In terms of infrastructure, although SENAI
technical courses are not significantly different from public institutions that offer
this same modality, they also have better structure than public regular schools.

3 Data and identification strategy

3.1 Data

The field research conducted by the Paraná Research Institute was designed to
estimate the effects of the courses offered by the FIEP System on the employabil-
ity, wages, overall satisfaction of its graduating students, and enrollemnt in higher
enducation. Data was collected in two complementary rounds. In the first of them,
students graduated from the courses between the years of 2015 to 2017 and stu-
dents who have not been enrolled in the courses offered by the FIEP System were
2
The SENAI’s courses that we analyze in this paper are not integrated to the regular high
school. They require the student to be enrolled or have completed the high school, but students
independently choose the institutions to take high school and vocational training.

8
Electronic copy available at: https://ssrn.com/abstract=3808661
interviewed from August to October 2018. This first sample covers 813 students
from SENAI’s technical courses, 523 from SESI high school and 873 students from
outside the FIEP System and is used to measure the short-term effects of vocational
training offered by the FIEP System. In the second round, students graduated from
the courses in the years of 2011 to 2014 and students who have not been enrolled in
the courses offered by the FIEP System were interviewed from May to July 2019.
Overall, 731 students from SENAI’s technical courses, 1375 from SESI high school
and 167 students from outside the FIEP System were interviewed in order to comple-
ment the first round and provide measures for the long-term effects of participating
in the System’s technical courses.

The sample of interviewed students was selected using the stratified sampling
technique, which consists of dividing the entire population into different subgroups
so that each individual is part of only one stratum. After defining the subgroups, the
selection of respondents can be performed by simple random sampling within each
defined stratum. The strata are defined according to characteristics observed for the
entire population, ensuring the complete representativeness of those characteristics
in the selected sample and reducing the sampling error. It is worth noting that the
selection process of the interviewees followed the same script in both stages, but was
carried out independently.

In this context, we first use the stratified sampling technique to ensure that
the sample of graduated students are representative of the population of graduate
students from the covered years, 2011 to 2017. Second, we use this technique to
select the sample of students who have not been enrolled in the courses offered by
the FIEP System and to ensure it presents similar characteristics to the sample of
graduate students. We use two data sources to implement this technique, namely:
the administrative data of all students who graduated from the courses provided by
FIEP, and the Paraná Research Institute database covering information of a broader
sample, people who have not been enrolled in the FIEP System.

Based on the information available at the Paraná Research Institute database3 ,

3
The variables available at the Paraná Research Database are: Sex (Male/Female), Age
brackets (16-24 years old/25-34 years old/35 or older), Schooling (Elementary School (Complete
or Incomplete)/High School (Complete or Incomplete)/Higher Education (Complete or Incom-
plete)), Main Occupation (Registered Employee/Unregistered Employee/Public Employee/Self-
employed (registered)/Self-employed/Entrepreneur/Free-lancer/Intern/Apprentice (Paid)/ Unem-
ployed (Job Seeker)/ Inactive (Not a Job Seeker) /Housewife/Retired/Only student/ Only living

9
Electronic copy available at: https://ssrn.com/abstract=3808661
we selected the students that are outsiders from the FIEP System in 3 steps:

1. All observations with the following main occupations were excluded: Public
Employee, only housewife, only retired or only living of some income. Ad-
ditionally, individuals who earned more than 10 minimum wages of monthly
household income were excluded.

2. Based on remaining population we kept only individuals with complete or

incomplete high school, and complete or incomplete higher education, which
were representative of SESI high school, SENAI technical and qualification
courses.

3. Within the remaining subsample, the strata were defined according to the age
group, the gender and the subregion of the state of Parana in a way that
replicates each stratum observed in the group of students graduated from the
courses offered by FIEP.

The sampling strategy presented above was also used when selecting the sample
of students that graduated from FIEP courses with a small difference, the subsample
was subdivided into two subgroups: SESI high school and SENAI technical courses.
We implemented this additional step to preserve the representativeness of these two
courses separately.

The questionnaire answered by the interviewees contained questions that cov-

ered the following topics: general characteristics of the interviewees, employability
and performance in the job market, satisfaction with their job, and with their pro-
fessional status and educational background and enrollment in higher education. All
questions included in the questionnaire were multiple choice. The full version of the
questionnaire is presented in the annex.

3.2 Identification Strategy

In order to assess causality, we first need to ensure that the sample of students
who have not been enrolled in the courses offered by the FIEP System is similar to
of some income), Monthly household income (Up to 1 minimum wage (mw)/Between 2 and 5
mws/Between 6 and 10 mws/ More than 10 mws) and City.

10
Electronic copy available at: https://ssrn.com/abstract=3808661
sample of students who graduated from these courses. We partially address this is-
sue in the survey sample selection by using the stratified randomization. We further
ensure comparability using the propensity score matching technique, the most com-
mon way to choose individuals with similar characteristics between different groups.
Based on observable characteristics relevant to the selection of program participants,
we select one or more units in the group of outsiders that are as similar as possible
to each unit in the group of graduates of the FIEP System courses.

The main issue that matching and other impact assessment methods try to solve
is the problem of selection bias in the participation of a given program. That is, it is
possible for program participants to be previously different from non-participants.
In this case, simply measuring the results between groups would be capturing prior
differences and not just the effect of participation in a particular program. Formally,
matching is based on the following identification hypothesis: conditional on some
covariate vector X, the outcome Y is independent of D, where D ∈ {0,1} is a dummy
variable of participation in FIEP courses. It is noteworthy that matching on X is
problematic if this vector is of high dimension (“curse of dimensionality”).

To address this dimensionality problem, we prefer to use the propensity score

matching method. When the probability of participating in a program depends
on several observable factors, rather than selecting similar individuals based on a
multidimensional measure, it is possible to establish a one-dimensional measure for
such a process (balancing score). This measure is given by the propensity score,
which is the probability of participation in the program conditional on covariates,
that is, the predicted probability of participation of individuals given their set of
observed characteristics. Rosenbaum and Rubin (1983) and Dehejia and Wahba
(1998) present the formal proof that allows the transformation of a multidimensional
measure into the propensity score.

The basic idea of the matching method is to search in a large group of non-
participants those individuals who are similar to the participants group in all relevant
observable characteristics. The selection bias is eliminated in the process as long as
it only occurs in the observable characteristics included in the model. In order words,
we must assume that the conditional hypothesis is valid, which means there can be
no unobservable characteristics that are associated at the same time with program
participation and potential outcomes. Formally based on Heckman, Ichimura, and

11
Electronic copy available at: https://ssrn.com/abstract=3808661
Todd (1997), where Y (0) is the potential results of non-participants in the courses
offered by FIEP, we should assume:

Y(0) ⊥ D | P(X), (1)

Therefore, the non-participant outcomes have, conditional on P(X), the same dis-
tribution that participants would have experienced if they had not participated in
the program.

In addition, the common support hypothesis should be valid, that is, although
observable characteristics may influence the likelihood of participating in a program,
participation cannot be completely defined by a set of variables. This hypothesis
ensures that it is possible to find an individual in the control group for each individual
in the treatment group after controlling for the influence of covariates. We should
assume:
Pr(D=1|X) < 1 ∀ X, (2)

Our dataset contains a rich set of socio-demographic variables. We assume

that they can potentially influence the participation in the courses and the labor
market and matching quality outcomes. Therefore, the variables incorporated in
the vector X were: education level, geographic region of the state of Paraná, gender,
age group, marital status, family condition (if the person is responsible for the
household), a dummy variable if the individual attended young adult education,
and a dummy variable if the person attended public school. The long-term survey
asks the individuals in retrospect about their employment status and income when
they started taking the course. We add this information as control variables for this
specific sample. In addition to the variables used in stratified randomization, we
ensure that the outsiders’ group is similar to the group that took a course in the
FIEP System also in these characteristics included in the model.

We use the propensity score estimates to select the closest neighbor of each
individual in the sample of students who graduated from all the courses offered by
FIEP4 . After estimating the participation probability using a Probit model, we allow
4
We run the propensity score considering all FIEP courses together to obtain the nearest
neighbor in the outsiders’ group. After obtaining the nearest neighbor, we allow for different
coefficients for students who graduated from SESI high school and from SENAI’s technical course,
and are able to test between them.

12
Electronic copy available at: https://ssrn.com/abstract=3808661
for replacement in the selection process, i.e., the same individual in the outsiders’
group can be selected as counterfactual for more than one individual in the FIEP
System group. To guarantee that we are selecting similar individuals, a caliper of
2p.p. is used – 2p.p. is the maximum level of tolerance for the distance between
the individual’s propensity score in the two groups considered. If the tolerance level
is exceeded, such individual in the FIEP System group is not considered in the
estimation.

After the final sample is defined, we compare the average of the outcome vari-
able weighting for how many times the same individual in the outsiders group was
used as counterfactual and also allowing for different effects of vocational training
and high school among the FIEP System graduates5 . This strategy permits us to
have two different control groups at the end - SESI high school and outsiders - to
measure the effects of SENAI vocational training. We expect that, by comparing
SENAI’s technical former students with students from outside the FIEP System,
we will have a measure similar to the one that is commonly used in the Brazilian
literature, that is totally based on the conditional on observable variables exogeneity
hypothesis. A more flexible measure is obtained by comparing SENAI’s technical
former students with students that graduated from SESI high school. We believe
that this measure of vocational training effect better addresses the potential issue
of differences in management and quality of the institutions that provide those two
different education modalities, since both courses are provided and managed by the
same institution - the FIEP System.

It is worth noting that, even though we believe that our estimation strategy al-
lows us to control for some unobservable variables, such as differences in quality and
management of the institutions, our results are still in some measure dependent on
the hypothesis that the selection into the groups must be determined by observable
characteristics. We rely on the assumption that no other unobservable characteristic
is driving the choice between high school and vocational training for the students
that decide to enroll in a course provided by the FIEP System. Additionally, the
difference in the results when using each of the control groups would only reflect
the effects of quality if no other omitted variable is affecting the selection into the
FIEP System courses. Based on that assumptions, our main specification relies on
5
According to Caliendo and Kopeinig (2006), the use of the nearest neighbor with replacement
is the method that minimizes bias but has a high variance.

13
Electronic copy available at: https://ssrn.com/abstract=3808661
the nearest neighbor propensity score matching with replacement. As a robustness
exercise, we also use an OLS model and a kernel propensity score matching.

Figure 1 presents the density function of the probability of participating in the

courses. It is possible to observe that the common support hypothesis is respected.
For all different probability of participation (x-axis), it is possible to observe some
mass in the density function for both groups. Moreover, the weighting performed
after selecting the nearest neighbor makes the density functions similar, in other
words, for each probability of participation the same mass of people is observed.

Figure 1: Density function of probability of participating in courses

Short run

a) Density not using weights b) Density using weights

Graduated from FIEP Outsiders Graduated from FIEP Outsiders
6

6
4

4
Density

Density
2

2
0

0 .2 .4 .6 .8 0 .2 .4 .6 .8
Probability of attending the courses Probability of attending the courses

Long run

a) Density not using weights b) Density using weights

Graduated from FIEP Outsiders Graduated from FIEP Outsiders
15

15
10

10
Density

Density
5

5
0

.4 .6 .8 1 .6 .7 .8 .9 1
Probability of attending the courses Probability of attending the courses

3.3 Difference in the average tests

The main purpose of the matching method is to ensure that the groups analyzed
are comparable to each other in their observable characteristics. To test if this was

14
Electronic copy available at: https://ssrn.com/abstract=3808661
achieved, we test the difference between the average of the variables in each of the
analyzed groups for both periods, short and long-term. We present the average
tests for the total sample of respondents and for the sample that is selected by
the matching method. Results are shown in the table 2. The “Difference” column
presents the results of the statistical tests that verify if the differences in the average
between the groups are significant.

Analyzing the table 2, it is possible to verify that the matching method cor-
rects almost all pre-existing differences in the socio-demographic variables, ensuring
comparability between the groups in the short-term. In the long-term, the matching
does not correct all the differences due to the small sample of outsiders. It is note-
worthy that stratified randomization was performed based on the variables of age,
gender and region of the state of Paraná. However, it is not possible to guarantee
that there are no differences between the groups in the other variables. The match-
ing method was performed with the inclusion of all variables presented in the table
2, and, as presented, it makes the groups more comparable in all these dimensions.

4 Results

The effects of vocational training on the variables that measure employment

status, income, overall satisfaction, and enrollment in higher education are presented
below. We compare the short and long-term effects of each course covered in our
analysis. In the tables, we report the average of the dependent variable for the
control group, consisted of students outside the FIEP System, and the difference
between them and the two groups of students graduated from the FIEP System
- regular and vocational education. In order to present a measure that accounts
for difference in institutions’ quality, we also present the p-value of a T test for
the difference between the coefficients estimated for students that graduated from
vocational and regular FIEP courses.

The effects of the vocational training on the professional status are presented
in table 3. An important finding is that controlling for quality has an important
impact on the vocational training results for most variables. More specifically, if we
simply compare the students graduated from the SENAI technical courses with the
students that are outsiders, we find positive significant effects only on employment in

15
Electronic copy available at: https://ssrn.com/abstract=3808661
the formal sector (23.3p.p.) in the short-run and on general employment (18.5p.p.)
and on employment in the formal sector (18.1p.p.) in the long-run.

More nuanced results are found when we test the estimated coefficients between
the students graduated from SENAI technical courses and SESI high school. The
employment effects become even more pronounced in the short-run. We find that
the former students of SENAI technical courses are 24p.p. (= 0.0942 + 0.1466)
more likely to be employed and 25.6p.p (= 0.2331 + 0.0232) more likely to be
employed in the formal sector in comparison to the students graduated from the
SESI high school, with both differences significant at a 1% significance level. A
higher employment probability comes at a price of decreasing the share dedicated
to only study in this group. A SENAI graduate is 23p.p. (=-0.0631 - 0.1673) less
likely to be only studying in the short-run relatively to a SESI high school graduate.

Differently from the short-run, controlling for quality attenuates all effects in
the long-run, but they are still significant. The former students of the SENAI
technical courses are 4.4p.p. (=0.1856 - 0.1408) more likely to be employed and 9p.p
(=0.1815 - 0.0917) more likely to be employed in the formal sector in comparison to
the students graduated from the SESI high school. The difference in the share that
is only studying persists in the long-run, a SENAI graduate is 4.2p.p. (=-0.0906 +
0.0487) less likely to be only studying in the long-run relatively to a SESI high school
graduate. A possible rationale for why accounting for institutions quality intensifies
the effects on the short-run and diminishes the effects on long-run is that students
that took regular education from high quality institutions have a higher chance of
going to college, and, as a consequence, to stay out of the labor force until they
complete their studies. In the long-run, when they return to the labor force, they
catch up part of the difference in the probability of being employed in comparison
to the students that graduated from vocational training.

The effects on income are also influenced by accounting for potential differences
in quality between educational institutions. Different from the pattern observed
when analyzing the employment status variable, controlling for quality accentuates
the positive effects of a vocational training in the short- and in the long-run. As
presented in table 4, by comparing the students who graduated from the SENAI
technical courses with the students that are outsiders, we find positive and significant
effects on income in the short-run, vocational training former students are 16.9p.p.

16
Electronic copy available at: https://ssrn.com/abstract=3808661
more likely to earn more than R$2000.00. However, this positive effect on income
totally fade out in the long-run.

The same is not true when we test the estimated coefficients between the stu-
dents graduated from SENAI technical courses and SESI high school. The differ-
ence in the income distribution between these two groups continues significant in
the long-run. In the short-run, the SENAI technical courses former students are
28.7p.p. (=0.1698 + 0.1169) and 4.2p.p. (=0.0251 + 0.0165) more likely to earn
more than R$2000.00 and more than R$4000.00, respectively. In the long-run, these
differences change to 22.3p.p. (=0.0397 + 0.1832) and 6.5p.p. (=-0.0457 + 0.1111)
and still significant at a 1% significance level. A possible rationale for these results
is that a higher employment inflow among the SESI high school graduates are also
associated with an increase in the share of the inexperienced workers in this group,
driving the income distribution towards low paying jobs.

In order to get a complete picture of the effects of vocational training, we mea-

sure the former students’ satisfaction with their activity sectors, professional status
and educational background. More specifically, we use different dummy variables
that indicate if the individual is employed in their preferred activity sector, and
if they are satisfied with their professional status and with their education back-
ground. The results are presented in the first three columns of table 5. The effects
are not very informative when quality is not taken into account. Comparing SENAI
technical graduates with the group of outsiders, we find significant effects only on
the share that is satisfied with their professional status in the long-run (15.2p.p.).

A more consistent analysis is provided when we account for institutions’ quality.

By comparing the SENAI technical courses and SESI high school, we find that
the SENAI technical graduates are more likely to be employed in their preferred
activity sector (12p.p. (= 0.1180 + 0.0019)), but are relatively less likely to be
satisfied with their educational background (-6.6p.p. (=0.1339 - 0.2008)) in the
short-run. In the long-run, only the negative effect in the satisfaction with the
educational background persists. More precisely, the SENAI technical course former
student is 6.7p.p. (=0.0209 - 0.0877) less likely to be satisfied with their educational
background.

Finally, the effects of vocational training on the probability of enrolling in higher

17
Electronic copy available at: https://ssrn.com/abstract=3808661
education (bachelors’ degree and postgraduate education) are presented in the fourth
column of table 5. The results are consistent with the argument that controlling for
institution quality increases the chance of going to college among the students that
took regular education from high quality institutions (SESI high school). As pre-
sented in the table, no effect in enrollment is found when we compared the SENAI
technical graduates with the group of students that are outsiders. However, rela-
tively to the SESI high school graduates, a SENAI technical course former student
is less likely to be enrolled in higher education in the short- (-16.2p.p. (= -0.0491
- 0.1129)) and long-run (-16p.p. (= -0.0129 - 0.1470)). In summary, the better
employment outcomes comes at a price of a lower enrollment in higher education.

5 Robustness checks

In this section we check the robustness of our results by re-estimating them

using two alternative econometric specifications: 1) we use an Epanechnikov kernel
propensity score matching procedure; 2) We run an OLS model using the same
vector of control variables that we use in the matching process.

It is important to highlight that exists a trade-off between bias and variance

among the propensity score models that we use in this paper. On one hand, the
Epanechnikov kernel matching increases the number of distinct non-participants
used to construct the counterfactual outcome and therefore decreases the variance
of the estimator. On the other hand, this non-parametric method uses the whole
distribution of individuals which likely decreases the average quality of matching
and increase the bias, especially in the case where the propensity score distribution
is very different in the treatment and control group (Caliendo and Kopeinig, 2008).
To partially address this problem, we also impose the same caliper used when es-
timating the nearest neighbor method, 2p.p. No relation, however, can be stated
when comparing OLS and propensity score method in terms of bias and variance.

The results using these two alternative methods are presented in tables 6 to 8.
The main results found on the employment status variables are maintained in our
robustness analysis. Vocational training courses increase the probability of being
generally employed and employed in the formal sector both in the short- and long-
run. By controlling for institutions’ quality, the employment effects become more

18
Electronic copy available at: https://ssrn.com/abstract=3808661
pronounced in the short-run and diminishes in the long-run. Importantly, in the
short-run a higher employment probability comes at a price of decreasing the share
dedicated to only study for the group of vocational training graduates. The degree
of attenuation in the long-run effects, however, varies with the estimation method.
When the results are estimated using a propensity score matching approach, the
effect on employment decreases, but still significant in the long-run. In the results
estimated using OLS, no difference between regular and vocational training grad-
uates persists in the long-run. Similarly, in both cases, ignoring the institution
quality dimension leads to an overestimation of vocational training long-run effects
on employment.

The effects on income are also consistent across all different estimation meth-
ods. Similarly to what is found in our main specification, controlling for quality
accentuates the positive effects of a vocational training in the short- and in the
long-run. Comparing the students who graduated from SENAI technical courses
and SESI high school, we find that the former is more likely to be employed in a
higher paying job both in the short- and long-run. The effects found using OLS are
smaller than the ones estimated using propensity score methods, but they are also
significant at 1% significance level.

Finally, the effects on the satisfaction variables and on the probability of en-
rolling in higher education are consistent between the different propensity score
matching methods, but differ from the results estimated using OLS. The estimates
of the Epanechnikov kernel matching approach also show that the SENAI techni-
cal graduates are more likely to be employed in their preferred activity sector, but
are relatively less likely to be satisfied with their educational background in the
short-run, and, in the long-run, only the latter effect persists. Regarding the higher
education enrollment, using this method we find that the vocational training gradu-
ates are less likely to be enrolled in short- and long-run. These effects, however, are
sensible to the estimation method. Using a OLS, the only effect that persists is the
greater activity sector satisfaction of vocational training graduates in the short-run.

19
Electronic copy available at: https://ssrn.com/abstract=3808661
6 Conclusion

This paper provides evidence about the short- and long-term effects of voca-
tional training on employment status, income, overall satisfaction, and enrollment
in higher education. We take advantage of particular institutional arrangement in
the state of Paraná, Brazil, where a single private institution named FIEP provides
both vocational and regular education under two separate but closely related enti-
ties, while non-FIEP institutions provide regular education. As both vocational and
regular schools within FIEP have more resources, better teachers and infra-structure
than non-FIEP schools, simply comparing outcomes of vocational training and reg-
ular students can be misleading even if students were assigned randomly to schools.

Using a unique survey applied to different cohorts of high school graduates, we

show that quality plays an important but nuanced role when comparing the effects
of general and vocational education in the short and long run. In particular, our
propensity score estimates indicate that FIEP vocational training graduates have
higher short-run employability than both FIEP and non-FIEP regular students.
However, regular graduates from the better-funded FIEP system are more likely
to continue to higher education, so that the short-run employment effect all but
dissipates as they enter the labor force in the long-run. We also find that vocational
training has a persistent positive effect on income both in the short and long-run,
especially when accounting for differences in quality.

Our estimates provide a relevant input for policy design in a moment when
vocational training has been gaining relevance in Brazil. In recent years, Federal
Government expenditure has increased 5 times from 2003 to 2016 (0.04% to 0.2%)
and the number of students enrolled in technical education during high school in-
creased by 45% between 2007 and 2013 (Elacqua et al., 2019). Industrial associations
such as FIEP play a major role in providing vocational education across Brazilian
regions. While privately managed, they are mainly funded via payroll tax revenues.
Hence understanding the effectiveness of such publicly-funded type of enterprises is
key to improving human capital in Brazil.

20
Electronic copy available at: https://ssrn.com/abstract=3808661
References

Almeida, Rita, Anazawa, Leandro, Menezes Filho, Naercio, and Vasconsellos, Ligia
(2015). “Investing in Technical Vocational Education and Training: Does It
Yield Large Economic Returns in Brazil?” World Bank WPS7246.
Assunção, J and Gonzaga, G. (2010). “Educação Profissional no Brasil: Inserção e
Retorno.”
Attanasio, Orazio, Guarı́n, Arlen, Medina, Carlos, and Meghir, Costas (2017). “Vo-
cational Training for Disadvantaged Youth in Colombia: A Long-Term Follow-
Up”. American Economic Journal: Applied Economics 9, pp. 131–143.
Attanasio, Orazio, Kugler, Adriana, and Meghir, Costas (2011). “Subsidizing Vo-
cational Training for Disadvantaged Youth in Colombia: Evidence from a Ran-
domized Trial”. American Economic Journal: Applied Economics 3, 188––220.
Brunello, G. and Rocco, L. (2017). “The labour market effects of vocational and
academic education over the life cycle. Evidence based on a British cohort.”
Journal of Human Capital 11.2, pp. 106–166.
Caliendo, M. and Kopeinig, S. (2008). “Some Practical Guidance for the Imple-
mentation of Propensity Score Matching.” Journal of Economic Surveys 22.1,
pp. 31–72.
Caliendo, Marco and Kopeinig, Sabine (2006). Some practical guidance for the im-
plementation of propensity score matching. Tech. rep. Institute for the Study of
Labor (IZA).
Camargo, Juliana, Lima, Lycia, Riva, Flavio, and Souza, André Portela (2018)
(2018). “Technical education, noncognitive skills and labor market outcomes:
experimental evidence from Brazil.” Textos para discussão 480, FGV/EESP -
Escola de Economia de São Paulo, Getulio Vargas Foundation (Brazil).
Costa, Fernanda, Fernandes, Julia Guerra, and Vasconcellos, Lı́gia (2010).
“Avaliação Econômica do Ensino Médio Profissional.” Fundação Itaú Social.
Dehejia, R. H. and Wahba, S. (1998). “Propensity score matching methods for non-
experimental causal studies.” NBER working paper 6829.
Elacqua, Gregory, Navarro-Palau, Patricia, Fernanda Prada, Maria, and Soares,
Sammara (2019). “Does technical education improve academic outcomes? Ev-
idence from Brazil.” IDB working paper IDB-WP-01057.

21
Electronic copy available at: https://ssrn.com/abstract=3808661
Golsteyn, B. H. H. and Stenberg, A. (2017). “Earnings over the life course: General
versus vocational education.” Journal of Human Capital 11.2, pp. 167–212.
Hanushek, E. A., Schwerdt, G., Woessmann, L., and Zhang, L. (2017). “General
education, vocational education, and labor-market outcomes over the lifecycle.”
Journal of Human Resources 52.1, pp. 58–87.
Hanushek, Eric A., Woessmann, Ludger, and Zhang, Lei. (2011). “General education,
vocational education, and labor-market outcomes over the life-cycle.” National
Bureau of Economic Research w17504.
Heckman, J., Ichimura, H., and Todd, P. (1997). “Matching as an Econometric
Evaluation Estimator: Evidence from Evaluating a Job Training Programme.”
Review of Economic Studies 64, 605––654.
Kugler, Adriana, Kugler, Maurice, Saavedra, Juan E., and Herrera-Pradas, Luis
Omar (2020). “Long-Term Educational Consequences of Vocational Training in
Colombia: Impacts on Young Trainees and their Relatives”. Journal of Human
Resources.
Malamud, Ofer and Pop-Eleches, Cristian (2010). “General education versus voca-
tional training: Evidence from an economy in transition.” Review of Economics
and Statistics 92.1, pp. 43–63.
Neri, Marcelo (2010). “A Educação Profissional e Você no Mercado de Trabalho”.
Rio de Janeiro: FGV/CPS.
Oswald-Egg, Maria Esther and Renold, Ursula (2021). “No experience, no employ-
ment: The effect of vocational education and training work experience on labour
market outcomes after higher education”. Economics of Education Review 80,
p. 102065.
Rosenbaum, P. R. and Rubin, D. B. (1983). “The central role of the propensity score
in observational studies for causal effects.” Biometrika 70.1, pp. 41–55.
Souza, Andre Portela, Lima, Lycia, Arabage, Amanda, Camargo, Juliana, Lucena,
Thiago de, and Soares, Sammara (2015). Vocational Education and Training in
Brazil. Tech. rep. Inter-American Development Bank.
Tansel, Aysit (1998). “General versus vocational high schools and labor market
outcomes in Turkey”. Economic Research Forum Working Paper 9905.

22
Electronic copy available at: https://ssrn.com/abstract=3808661
Table 1: Difference in the average ENEM (2015) variables between Sesi and other
high schools
Public Paraná Private Paraná Sesi Testing the mean differences
(1) (2) (3)
Mean Obs Mean Obs Mean Obs (1) - (3) (2) - (3)
Participation rate 62.2881 391 84.1577 273 83.6944 45 -21.4063*** 0.4632
Average in natural science 478.3771 391 539.3789 273 493.3822 45 -15.0052*** 45.9966***
Average in human science 559.7686 391 605.9794 273 575.0084 45 -15.2399*** 30.9709***
Average in Portuguese 506.0876 391 555.1439 273 523.2969 45 -17.2092*** 31.8470***
Average in math 469.9187 391 557.967 273 500.5831 45 -30.6644*** 57.3839***
Average in writing 534.2359 391 612.7274 273 564.8487 45 -30.6127*** 47.8788***
Index of faculty adequacy 78.5803 390 68.1136 272 75.4667 45 3.1136* -7.3531***
Student permanence index in high school 82.8856 391 68.9434 273 75.6313 45 7.2542*** -6.6880*
Students’ approval rate in high school 83.8453 391 96.1187 273 97.3119 42 -13.4666*** -1.1932**
Students’ failure rate in high school 10.1936 391 3.7264 273 2.4214 42 7.7722*** 1.3049**
Students’ dropout rate in high school 5.9611 391 0.1549 273 0.2667 42 5.6945*** -0.1117
Significance levels: * 10%, ** 5%, ***1%.

23
Electronic copy available at: https://ssrn.com/abstract=3808661
 Table 2: Difference in the average tests between treatment and control groups
Without matching With matching
Graduated from Outsiders Difference Graduated from Outsiders Difference
FIEP FIEP
Short-term
Age group
16-24 years old 0.8024 0.52 0.2823*** 0.8018 0.8011 0.0008
25-34 years old 0.1243 0.2096 -0.0854*** 0.1246 0.1186 0.006
35 or older 0.0734 0.2703 -0.1970*** 0.0736 0.0803 -0.0068
Share of females 0.3787 0.37 0.0088 0.3799 0.3836 -0.0038
Geographic region of the State of Paraná
Central 0.0689 0.0664 0.0024 0.0691 0.0503 0.0188
Curitiba 0.4057 0.362 0.0437** 0.4047 0.4122 -0.0075
North 0.357 0.362 -0.0049 0.3574 0.3761 -0.0188
West 0.0876 0.1031 -0.0155 0.0878 0.0743 0.0135
South 0.0808 0.1065 -0.0257** 0.0811 0.0871 -0.006
Education level
Elementary School (complete or incomplete) 0.0037 0.0435 -0.0398*** 0.0038 0.0015 0.0023
High school (complete or incomplete) 0.5479 0.5888 -0.0409* 0.5495 0.5375 0.012
Higher education (complete or incomplete) 0.4416 0.3379 0.1037*** 0.4399 0.458 -0.018
Postgraduate studies 0.0067 0.0298 -0.0230*** 0.0068 0.003 0.0038
Share that attended regular school 0.9386 0.89 0.0486*** 0.9392 0.9542 -0.015
Share that attended public school 0.7418 0.8396 -0.0979*** 0.744 0.7372 0.0068
Marital status
Single 0.8361 0.606 0.2301*** 0.8356 0.8378 -0.0023
Married 0.1557 0.3414 -0.1857*** 0.1562 0.1532 0.003
Share responsible for the household 0.2313 0.4147 -0.1834*** 0.2297 0.2095 0.0203
Observations 1336 873 2209 1332 184 1516
Long-term
Age group
16-24 years old 0.8884 0.8982 -0.0098 0.9191 0.9656 -0.0465***
25-34 years old 0.0693 0.024 0.0454** 0.0349 0.0079 0.0270***
35 or older 0.0423 0.0778 -0.0356** 0.046 0.0264 0.0196**
Share of females 0.3941 0.4072 -0.0131 0.4209 0.4807 -0.0598***
Geographic region of the State of Paraná
Central 0.1662 0.1737 -0.0075 0.1777 0.2031 -0.0254
Curitiba 0.4858 0.4371 0.0486 0.458 0.4659 -0.0079
North 0.2251 0.2036 0.0215 0.2348 0.221 0.0137
West 0.0665 0.1198 -0.0533*** 0.0693 0.0693 0
South 0.0565 0.0659 -0.0094 0.0603 0.0407 0.0196**
Education level
Elementary School (complete or incomplete) 0 0 0 0 0 0
High school (complete or incomplete) 0.2213 0.3593 -0.1380*** 0.2285 0.1645 0.0640***
Higher education (complete or incomplete) 0.7179 0.5389 0.1790*** 0.7076 0.7795 -0.0719***
Postgraduate studies 0.0608 0.1018 -0.0410** 0.064 0.0561 0.0079
Share that attended regular school 0.9691 0.9341 0.0350** 0.9693 0.9741 -0.0048
Share that attended public school 0.6918 0.8683 -0.1764*** 0.7308 0.7604 -0.0296
Marital status
Single 0.7612 0.7365 0.0246 0.7816 0.8123 -0.0307*
Married 0.226 0.2395 -0.0135 0.2052 0.1803 0.0249
Share responsible for the household 0.2702 0.3234 -0.0532 0.2681 0.2411 0.027
Observations 2106 167 2273 1891 125 2016
Significance levels: * 10%, ** 5%, ***1%.

24
Electronic copy available at: https://ssrn.com/abstract=3808661
 Table 3: Effect of each course on employment status variables
(1) (2) (3) (4) (5)
Unemployed Inactive Studying Employed Employed in a
formal job
Short-term Effects
Senai technical 0.0052 -0.0152 -0.0631 0.0942 0.2331∗∗∗
(0.0745) (0.0202) (0.0691) (0.0827) (0.0642)
Sesi high-school 0.0187 -0.0204 0.1673∗∗ -0.1466∗ -0.0232
(0.0751) (0.0204) (0.0711) (0.0839) (0.0644)
Mean dep. var - control group 0.1329 0.0473 0.2357 0.5631 0.2477
Coeff. equality test (T test - p-value) 0.496 0.582 0.000 0.000 0.000
Observations 1516 1516 1516 1516 1516
Long-term Effects
Senai technical -0.0943 -0.0019 -0.0906 0.1856∗∗ 0.1815∗∗
(0.0609) (0.0109) (0.0638) (0.0757) (0.0728)
Sesi high-school -0.0862 -0.0085 -0.0487 0.1408∗ 0.0917
(0.0604) (0.0102) (0.0636) (0.0750) (0.0714)
Mean dep. var - control group 0.1750 0.0132 0.1713 0.6351 0.3548
Coeff. equality test (T test - p-value) 0.552 0.158 0.003 0.021 0.000
Observations 2016 2016 2016 2016 2016
Significance levels: * 10%, ** 5%, ***1%. We use robust standard errors. The table presents the results using employment status variables
in the short and long-term. The columns display the results for different binary variables, and also the shares observed in the control group
consisted of outsiders. We first select our sample using the propensity score method. We select the closest neighbor of each individual by
estimating the participation probability using a Probit model, we allow for replacement in the selection process, i.e., the same individual
in the control group can be selected as counterfactual for more than one individual in the treatment group. To guarantee that we are
selecting similar individuals, a caliper of 2p.p. is used – 2p.p. is the maximum level of tolerance for the distance between the individual’s
propensity score in the treatment group and their nearest neighbor in the control group. If the tolerance level is exceeded, such individual
in the treatment group is not considered in the estimation. After the final sample is defined, we estimate by OLS the difference between
the groups in terms of the average outcome variable weighting for how many times the same individual in the control group was used as
counterfactual. We build the confidence intervals using robust standard errors.

25
Electronic copy available at: https://ssrn.com/abstract=3808661
 Table 4: Effect of each course on different income brackets
(1) (2) (3)
Less than R$2.000 More than R$2.001 More than R$4.001
Short-term Effects
Senai technical -0.1698∗∗∗ 0.1698∗∗∗ 0.0251∗
(0.0561) (0.0561) (0.0148)
Sesi high-school 0.1169∗∗ -0.1169∗∗ -0.0165
(0.0551) (0.0551) (0.0136)
Mean dep. var - control group 0.8062 0.1938 0.0310
Coeff. equality test (T test - p-value) 0.000 0.000 0.002
Observations 844 844 844
Long-term Effects
Senai technical -0.0397 0.0397 -0.0457
(0.0923) (0.0923) (0.0878)
Sesi high-school 0.1832∗∗ -0.1832∗∗ -0.1111
(0.0908) (0.0908) (0.0868)
Mean dep. var - control group 0.4874 0.5126 0.1817
Coeff. equality test (T test - p-value) 0.000 0.000 0.000
Observations 1498 1498 1498
Significance levels: * 10%, ** 5%, ***1%. We use robust standard errors. The table presents the results using income bracket variables in
the short and long-term. The columns display the results for different binary variables, and also the shares observed in the control group
consisted of outsiders. We first select our sample using the propensity score method. We select the closest neighbor of each individual by
estimating the participation probability using a Probit model, we allow for replacement in the selection process, i.e., the same individual
in the control group can be selected as counterfactual for more than one individual in the treatment group. To guarantee that we are
selecting similar individuals, a caliper of 2p.p. is used – 2p.p. is the maximum level of tolerance for the distance between the individual’s
propensity score in the treatment group and their nearest neighbor in the control group. If the tolerance level is exceeded, such individual
in the treatment group is not considered in the estimation. After the final sample is defined, we estimate by OLS the difference between
the groups in terms of the average outcome variable weighting for how many times the same individual in the control group was used as
counterfactual. We build the confidence intervals using robust standard errors.

26
Electronic copy available at: https://ssrn.com/abstract=3808661
Table 5: Effects of the courses on the share of satisfaction variables
(1) (2) (3) (4)
Activity Professional Education Enrolled in
Sector Status Background Higher Education
Short-term Effects
Senai technical 0.1180 0.1014 0.1339∗ -0.0491
(0.1026) (0.0851) (0.0739) (0.0817)
Sesi high-school -0.0019 0.0741 0.2008∗∗∗ 0.1129
(0.1059) (0.0860) (0.0740) (0.0828)
Mean dep. var - control group 0.4739 0.6564 0.6935 0.4437
Coeff. equality test (T test - p-value) 0.003 0.271 0.000 0.000
Observations 853 1497 1514 1516
Long-term Effects
Senai technical 0.0409 0.1520∗∗ 0.0209 -0.0129
(0.0894) (0.0771) (0.0652) (0.0724)
Sesi high-school 0.0262 0.1126 0.0877 0.1470∗∗
(0.0881) (0.0762) (0.0640) (0.0712)
Mean dep. var - control group 0.5595 0.6098 0.8070 0.3538
Coeff. equality test (T test - p-value) 0.586 0.065 0.000 0.000
Observations 1572 2004 2013 2016
Significance levels: * 10%, ** 5%, ***1%. We use robust standard errors. The columns display the results for different binary
variables, and also the shares observed in the control group consisted of outsiders. “Activity Sector” is a dummy variable that
indicates if the individuals are employed in their preferred activity sector. “Professional Status” and “Educational Background” are
dummies that equal 1 if the individuals are satisfied with their professional status and with their education background, respectively.
”Enrolled in Higher Education” is a dummy variable that indicates if someone is enrolled in higher education, such as bachelors’
degree or postgraduate education. We first select our sample using the propensity score method. We select the closest neighbor of
each individual by estimating the participation probability using a Probit model, we allow for replacement in the selection process,
i.e., the same individual in the control group can be selected as counterfactual for more than one individual in the treatment group.
To guarantee that we are selecting similar individuals, a caliper of 2p.p. is used – 2p.p. is the maximum level of tolerance for the
distance between the individual’s propensity score in the treatment group and their nearest neighbor in the control group. If the
tolerance level is exceeded, such individual in the treatment group is not considered in the estimation. After the final sample is
defined, we estimate by OLS the difference between the groups in terms of the average outcome variable weighting for how many
times the same individual in the control group was used as counterfactual. We build the confidence intervals using robust standard
errors.

27
Electronic copy available at: https://ssrn.com/abstract=3808661
 Table 6: Effect of each course on employment status variables
(1) (2) (3) (4) (5)
Unemployed Inactive Studying Employed Employed in a
formal job
Short-term Effects
PSM w/ replacement (baseline)
Senai technical 0.0052 -0.0152 -0.0631 0.0942 0.2331∗∗∗
(0.0745) (0.0202) (0.0691) (0.0827) (0.0642)
Sesi high-school 0.0187 -0.0204 0.1673∗∗ -0.1466∗ -0.0232
(0.0751) (0.0204) (0.0711) (0.0839) (0.0644)
Mean dep. var - control group 0.1329 0.0473 0.2357 0.5631 0.2477
Coeff. equality test (T test - p-value) 0.496 0.582 0.000 0.000 0.000
Observations 1516 1516 1516 1516 1516

PSM Epanechnikov kernel

Senai technical 0.0075 -0.0270∗∗ -0.0392 0.0762∗∗∗ 0.1687∗∗∗
(0.0197) (0.0121) (0.0248) (0.0289) (0.0282)
Sesi high-school 0.0210 -0.0322∗∗ 0.1912∗∗∗ -0.1645∗∗∗ -0.0876∗∗∗
(0.0221) (0.0126) (0.0300) (0.0320) (0.0286)
Mean dep. var - control group 0.1306 0.0591 0.2119 0.5810 0.3122
Coeff. equality test (T test - p-value) 0.496 0.582 0.000 0.000 0.000
Observations 2205 2205 2205 2205 2205

OLS
Senai technical 0.0111 -0.0317∗∗∗ 0.0020 0.0551∗∗ 0.1668∗∗∗
(0.0177) (0.0120) (0.0181) (0.0241) (0.0247)
Sesi high-school 0.0020 -0.0371∗∗∗ 0.1225∗∗∗ -0.0642∗∗ 0.0224
(0.0224) (0.0131) (0.0280) (0.0309) (0.0281)
Mean dep. var - control group 0.1203 0.0905 0.1329 0.5968 0.2910
Coeff. equality test (T test - p-value) 0.663 0.615 0.000 0.000 0.000
Observations 2209 2209 2209 2209 2209
Long-term Effects
PSM w/ replacement (baseline)
Senai technical -0.0943 -0.0019 -0.0906 0.1856∗∗ 0.1815∗∗
(0.0609) (0.0109) (0.0638) (0.0757) (0.0728)
Sesi high-school -0.0862 -0.0085 -0.0487 0.1408∗ 0.0917
(0.0604) (0.0102) (0.0636) (0.0750) (0.0714)
Mean dep. var - control group 0.1750 0.0132 0.1713 0.6351 0.3548
Coeff. equality test (T test - p-value) 0.552 0.158 0.003 0.021 0.000
Observations 2016 2016 2016 2016 2016

PSM Epanechnikov kernel

Senai technical -0.0368 -0.0060 -0.1153∗∗∗ 0.1615∗∗∗ 0.1519∗∗∗
(0.0311) (0.0129) (0.0415) (0.0479) (0.0500)
Sesi high-school -0.0288 -0.0126 -0.0734∗ 0.1168∗∗ 0.0621
(0.0302) (0.0123) (0.0411) (0.0468) (0.0478)
Mean dep. var - control group 0.1176 0.0173 0.1961 0.6591 0.3845
Coeff. equality test (T test - p-value) 0.552 0.158 0.003 0.021 0.000
Observations 2057 2057 2057 2057 2057

OLS
Senai technical -0.0364 -0.0052 -0.0534∗∗ 0.1166∗∗∗ 0.1180∗∗∗
(0.0288) (0.0125) (0.0267) (0.0387) (0.0431)
Sesi high-school -0.0353 -0.0139 -0.0619∗∗ 0.1245∗∗∗ 0.0887∗∗
(0.0270) (0.0101) (0.0262) (0.0366) (0.0408)
Mean dep. var - control group 0.1257 0.0180 0.1257 0.7006 0.3952
Coeff. equality test (T test - p-value) 0.946 0.178 0.603 0.719 0.270
Observations 2273 2273 2273 2273 2273
Significance levels: * 10%, ** 5%, ***1%. We use robust standard errors. The table presents the results using employment status variables
in the short and long-term. The columns display the results for different binary variables, and also the shares observed in the control group
consisted of outsiders. We report our baseline results using a nearest neighbor propensity score matching to select our sample, the results using
a Epanechnikov kernel matching to select the sample, and a direct OLS that includes the whole sample (without any selection). When using
the propensity score methods, we first select our sample using the propensity score method with a caliper of 2p.p., and, after the final sample is
defined, we estimate by OLS the difference between the groups in terms of the average outcome variable weighting according to each method.

28
Electronic copy available at: https://ssrn.com/abstract=3808661
 Table 7: Effect of each course on different income brackets
(1) (2) (3)
Less than R$2.000 More than R$2.001 More than R$4.001
Short-term Effects
PSM w/ replacement (baseline)
Senai technical -0.1698∗∗∗ 0.1698∗∗∗ 0.0251∗
(0.0561) (0.0561) (0.0148)
Sesi high-school 0.1169∗∗ -0.1169∗∗ -0.0165
(0.0551) (0.0551) (0.0136)
Mean dep. var - control group 0.8062 0.1938 0.0310
Coeff. equality test (T test - p-value) 0.000 0.000 0.002
Observations 844 844 844

PSM Epanechnikov kernel

Senai technical -0.1647∗∗∗ 0.1647∗∗∗ 0.0181
(0.0307) (0.0307) (0.0132)
Sesi high-school 0.1220∗∗∗ -0.1220∗∗∗ -0.0235∗∗
(0.0289) (0.0289) (0.0118)
Mean dep. var - control group 0.8011 0.1989 0.0380
Coeff. equality test (T test - p-value) 0.000 0.000 0.001
Observations 1219 1219 1219

OLS
Senai technical -0.0621∗∗ 0.0621∗∗ -0.0226
(0.0283) (0.0283) (0.0165)
Sesi high-school 0.0057 -0.0057 -0.0074
(0.0301) (0.0301) (0.0148)
Mean dep. var - control group 0.6498 0.3502 0.1134
Coeff. equality test (T test - p-value) 0.012 0.012 0.247
Observations 1222 1222 1222
Long-term Effects
PSM w/ replacement (baseline)
Senai technical -0.0397 0.0397 -0.0457
(0.0923) (0.0923) (0.0878)
Sesi high-school 0.1832∗∗ -0.1832∗∗ -0.1111
(0.0908) (0.0908) (0.0868)
Mean dep. var - control group 0.4874 0.5126 0.1817
Coeff. equality test (T test - p-value) 0.000 0.000 0.000
Observations 1498 1498 1498

PSM Epanechnikov kernel

Senai technical -0.0929 0.0929 -0.0094
(0.0636) (0.0636) (0.0462)
Sesi high-school 0.1300∗∗ -0.1300∗∗ -0.0748∗
(0.0613) (0.0613) (0.0443)
Mean dep. var - control group 0.5405 0.4595 0.1454
Coeff. equality test (T test - p-value) 0.000 0.000 0.000
Observations 1522 1522 1522

OLS
Senai technical -0.0815 0.0815 -0.0141
(0.0509) (0.0509) (0.0365)
Sesi high-school 0.0037 -0.0037 -0.0153
(0.0478) (0.0478) (0.0333)
Mean dep. var - control group 0.6055 0.3945 0.1193
Coeff. equality test (T test - p-value) 0.004 0.004 0.944
Observations 1681 1681 1681
SSignificance levels: * 10%, ** 5%, ***1%. We use robust standard errors. The table presents the results using income bracket variables in
the short and long-term. The columns display the results for different binary variables, and also the shares observed in the control group
consisted of outsiders. We report our baseline results using a nearest neighbor propensity score matching to select our sample, the results
using a Epanechnikov kernel matching to select the sample, and a direct OLS that includes the whole sample (without any selection).
When using the propensity score methods, we first select our sample using the propensity score method with a caliper of 2p.p., and, after
the final sample is defined, we estimate by OLS the difference between the groups in terms of the average outcome variable weighting
according to each method.

29
Electronic copy available at: https://ssrn.com/abstract=3808661
Table 8: Effects of the courses on the share of satisfaction variables
(1) (2) (3) (4)
Activity Professional Education Enrolled in
Sector Status Background Higher Education
Short-term Effects
PSM w/ replacement (baseline)
Senai technical 0.1180 0.1014 0.1339∗ -0.0491
(0.1026) (0.0851) (0.0739) (0.0817)
Sesi high-school -0.0019 0.0741 0.2008∗∗∗ 0.1129
(0.1059) (0.0860) (0.0740) (0.0828)
Mean dep. var - control group 0.4739 0.6564 0.6935 0.4437
Coeff. equality test (T test - p-value) 0.003 0.271 0.000 0.000
Observations 853 1497 1514 1516

PSM Epanechnikov kernel

Senai technical 0.1646∗∗∗ 0.0941∗∗∗ 0.1413∗∗∗ 0.0086
(0.0376) (0.0273) (0.0257) (0.0298)
Sesi high-school 0.0447 0.0668∗∗ 0.2081∗∗∗ 0.1706∗∗∗
(0.0460) (0.0300) (0.0258) (0.0327)
Mean dep. var - control group 0.4273 0.6637 0.6861 0.3860
Coeff. equality test (T test - p-value) 0.003 0.271 0.000 0.000
Observations 1229 2170 2200 2205

OLS
Senai technical 0.1291∗∗∗ 0.0890∗∗∗ 0.1411∗∗∗ 0.0638∗∗∗
(0.0339) (0.0243) (0.0231) (0.0163)
Sesi high-school 0.0539 0.0826∗∗∗ 0.1559∗∗∗ 0.0651∗∗∗
(0.0470) (0.0298) (0.0250) (0.0190)
Mean dep. var - control group 0.4659 0.6565 0.6847 0.1993
Coeff. equality test (T test - p-value) 0.092 0.812 0.440 0.938
Observations 1231 2173 2204 2209
Long-term Effects
PSM w/ replacement (baseline)
Senai technical 0.0409 0.1520∗∗ 0.0209 -0.0129
(0.0894) (0.0771) (0.0652) (0.0724)
Sesi high-school 0.0262 0.1126 0.0877 0.1470∗∗
(0.0881) (0.0762) (0.0640) (0.0712)
Mean dep. var - control group 0.5595 0.6098 0.8070 0.3538
Coeff. equality test (T test - p-value) 0.586 0.065 0.000 0.000
Observations 1572 2004 2013 2016

PSM Epanechnikov kernel

Senai technical 0.0402 0.0758 -0.0280 -0.1182∗∗
(0.0613) (0.0477) (0.0359) (0.0511)
Sesi high-school 0.0256 0.0364 0.0387 0.0417
(0.0594) (0.0462) (0.0336) (0.0494)
Mean dep. var - control group 0.5601 0.6860 0.8560 0.4591
Coeff. equality test (T test - p-value) 0.586 0.065 0.000 0.000
Observations 1598 2044 2054 2057

OLS
Senai technical 0.0314 0.0176 -0.0191 0.0158
(0.0536) (0.0406) (0.0316) (0.0330)
Sesi high-school 0.0521 0.0277 0.0070 0.0140
(0.0509) (0.0388) (0.0295) (0.0323)
Mean dep. var - control group 0.5478 0.7195 0.8503 0.3473
Coeff. equality test (T test - p-value) 0.501 0.676 0.151 0.929
Observations 1762 2259 2270 2273
Significance levels: * 10%, ** 5%, ***1%. We use robust standard errors. The columns display the results for different binary
variables, and also the shares observed in the control group consisted of outsiders. “Activity Sector” is a dummy variable that
indicates if the individuals are employed in their preferred activity sector. “Professional Status” and “Educational Background” are
dummies that equal 1 if the individuals are satisfied with their professional status and with their education background, respectively.
”Enrolled in Higher Education” is a dummy variable that indicates if someone is enrolled in higher education, such as bachelors’
degree or postgraduate education. We report our baseline results using a nearest neighbor propensity score matching to select our
sample, the results using a Epanechnikov kernel matching to select the sample, and a direct OLS that includes the whole sample
(without any selection). When using the propensity score methods, we first select our sample using the propensity score method
with a caliper of 2p.p., and, after the final sample is defined, we estimate by OLS the difference between the groups in terms of the
average outcome variable weighting according to each method.

30
Electronic copy available at: https://ssrn.com/abstract=3808661
Appendix A - Tables

Table A1: Difference in the average school variables between Sesi and other high
schools - 2017 Census
Public Paraná Private Paraná Sesi Testing the mean differences
(1) (2) (3)
Mean Obs Mean Obs Mean Obs (1) - (3) (2) - (3)
The school has:
School director office 0.9069 1299 0.9888 356 0.9811 53 -0.0743* 0.0076
School professor office 0.9638 1299 0.9944 356 0.9811 53 -0.0173 0.0132
Computer lab 0.8907 1299 0.736 356 0.9623 53 -0.0716* -0.2263***
Science lab 0.7313 1299 0.9242 356 0.9057 53 -0.1743*** 0.0185
Library 0.9292 1299 0.9803 356 1 53 -0.0708** -0.0197
Reading room 0.0508 1299 0.5056 356 0.3019 53 -0.2511*** 0.2037***
Auditorium 0.1339 1299 0.4972 356 0.5283 53 -0.3944*** -0.0311
Number of classrooms 11.659 1299 22.2388 356 10.3019 53 1.3571* 11.9369***
Number of used classrooms 10.4426 1299 20.3062 356 8.1321 53 2.3106*** 12.1741***
Number of computers 28.1647 1299 46.0899 356 38.9811 53 -10.8164*** 7.1088
Number of computers available for students 19.9761 1299 28.0927 356 31.5472 53 -11.5710*** -3.4545
Has internet 0.9931 1299 0.9944 356 0.9811 53 0.0119 0.0132
Has high speed internet 0.8075 1299 0.9719 356 0.9623 53 -0.1547*** 0.0096
Number of students 200.3087 1299 126.3202 356 175.3962 53 24.9125 -49.0760**
Number of professors 22.3272 1299 16.0702 356 15.283 53 7.0442*** 0.7872
Significance levels: * 10%, ** 5%, ***1%.

31
Electronic copy available at: https://ssrn.com/abstract=3808661
Table A2: Difference in the average school variables between Senai and other tech-
nical courses - 2017 Census
Public Paraná Private Paraná Senai Testing the mean differences
(1) (2) (3)
Mean Obs Mean Obs Mean Obs (1) - (3) (2) - (3)
The school has:
School director office 0.9792 48 0.989 91 0.925 40 0.0542 0.0640*
School professor office 0.9792 48 0.978 91 0.95 40 0.0292 0.028
Computer lab 1 48 0.9231 91 0.975 40 0.025 -0.0519
Science lab 0.7708 48 0.2857 91 0.375 40 0.3958*** -0.0893
Library 0.9792 48 0.967 91 0.975 40 0.0042 -0.008
Reading room 0.3542 48 0.5495 91 0.275 40 0.0792 0.2745***
Auditorium 0.5833 48 0.5714 91 0.625 40 -0.0417 -0.0536
Number of classrooms 13.9792 48 16.8132 91 15.9 40 -1.9208 0.9132
Number of used classrooms 10.2917 48 14.7363 91 14.25 40 -3.9583** 0.4863
Number of computers 85.125 48 77.7582 91 72.625 40 12.5 5.1332
Number of computers available for students 61.9792 48 39.7802 91 53.5 40 8.4792 -13.7198
Has internet 1 48 1 91 0.975 40 0.025 0.025
Has high speed internet 0.875 48 0.956 91 0.9 40 -0.025 0.056
Number of students 187.1042 48 151.5934 91 214.475 40 -27.3708 -62.8816
Number of professors 20.2083 48 15.5495 91 17.875 40 2.3333 -2.3255
Significance levels: * 10%, ** 5%, ***1%.

32
Electronic copy available at: https://ssrn.com/abstract=3808661
Appendix B - Questionnaire

QUESTIONÁRIO - BASE
Script inicial do entrevistador:

Bom dia/ Boa Tarde. Meu nome é _________________. Sou entrevistador do INSTITUTO PARANÁ
PESQUISAS, e estamos entrando em contato com você para fazer uma pesquisa sobre questões
ligadas a trabalho e educação. O(a) Sr(a) poderia fazer a gentileza de responder algumas perguntas? Antes
de iniciar, gostaria de deixar claro que as respostas não serão utilizadas para qualquer outro propósito além
de coletar informações sobre Educação e Trabalho e os dados coletados serão levados em consideração no
conjunto das informações coletadas e não de forma individualizada.

Questões Filtro:
F1. Qual a sua escolaridade?
1) Sem escolaridade/ analfabeto (Agradecer e encerrar)
2) Ensino Fundamental Incompleto
3) Ensino Fundamental Completo
4) Ensino Médio Incompleto
5) Ensino Médio Completo
6) Ensino Superior Incompleto
7) Ensino Superior Completo
8) Pós-Graduação ou mais

F2. O(A) Sr(a) estudou ou estuda em alguma unidade do Sistema Fiep, ou seja, alguma Unidade do Senai, Sesi ou IEL?
1) Sim (Exceto perguntas exclusivas do Grupo Controle)
2) Não (Todas as perguntas)

F3. Apenas para os que são egressos dos cursos de Educação do Sistema Fiep: Qual o ano que o(a) Sr(a) concluiu/ terminou os seus
estudos na Unidade do Sistema Fiep?
1) Anterior a 2015 (Agradecer e encerrar)
2) 2015
3) 2016
4) 2017
5) 2018 ou mais (Agradecer e encerrar)

1. Sexo: (Registrar)
1) Masculino 2) Feminino
--------------------------------------------------------------------------------------------------------------------------------------------------
2. Qual a sua idade? (Registrar a faixa correspondente)
1) 16 a 24 anos 2) 25 a 34 anos 3) 35 ou mais
----------------------------------------------------------------------------------------------------------------------------- ---------------------
3. Diga-me, por favor, qual das seguintes situações se aplica melhor ao seu estado civil atual: (Ler as
alternativas)
1) Solteiro(a) 3) Divorciado(a) 5) Outro. Especifique: ______
2) Casado(a) 4) Viúvo(a)
--------------------------------------------------------------------------------------------------------------------------------------------------

33
Electronic copy available at: https://ssrn.com/abstract=3808661
 QUESTIONÁRIO - BASE
4. Atualmente o(a) Sr(a) diria que é: (Ler as alternativas)
1) Pessoa responsável pelo seu domicílio
2) Cônjuge/ companheiro(a) do(a) responsável pelo domicílio
3) Filho(a) do(a) responsável pelo domicílio
4) Neto(a) do(a) responsável pelo domicílio
5) Irmão(ã) do(a) responsável pelo domicílio
6) Outro. Especifique: ____________
------------------------------------------------------------------------------------------------------------------------------------------------
5. Contando com o(a) Sr(a), quantas pessoas, incluindo crianças vivem habitualmente em sua residência?
|_____|
--------------------------------------------------------------------------------------------------------------------------------------------------
6. De uma maneira geral, o Sr(a) diria que está muito satisfeito(a), satisfeito(a), nem satisfeito(a), nem
insatisfeito(a), insatisfeito(a) ou muito insatisfeito(a) com a sua situação profissional?
1) Muito Satisfeito(a) 4) Insatisfeito(a)
2) Satisfeito(a) 5) Muito insatisfeito(a)
3) Nem Satisfeito(a), Nem Insatisfeito(a) 6) Não sabe/ não opinou (não ler, nem estimular)
------------------------------------------------------------------------------------------------------------------------------------------------
7. De uma maneira geral, o Sr(a) diria que está muito satisfeito(a), satisfeito(a), nem satisfeito(a), nem
insatisfeito(a), insatisfeito(a) ou muito insatisfeito(a) com a sua formação escolar?
1) Muito Satisfeito(a) 4) Insatisfeito(a)
2) Satisfeito(a) 5) Muito insatisfeito(a)
3) Nem Satisfeito(a), Nem Insatisfeito(a) 6) Não sabe/ não opinou (não ler, nem estimular)
------------------------------------------------------------------------------------------------------------------------------------------------
8. Qual o Setor que o(a) Sr(a) acredita que traz mais benefícios para o trabalhador: Serviços, Comércio,
Indústria ou Agricultura?
1) Agricultura (ir p/ a 10)
2) Comércio (ir p/ a 10)
3) Indústria
4) Serviços (ir p/ a 10)
5) Não sabe/ não opinou (não ler, nem estimular) (ir p/ a 10)
----------------------------------------------------------------------------------------------------------------------------- ---------------------
9. Qual desses setores industrias o Sr(a) acredita que traga mais benefícios para o trabalhador? (Ler as
alternativas)
1) Não sabe (não ler) 6) Eletro eletrônica
2) Extração mineral (mineração e petróleo) 7) Máquinas e equipamentos
3) Alimentos, Bebidas ou Fumo 8) Veículos
4) Têxtil, vestuário, calçados e couro 9) Construção
5) Química 10) Outro. Especifique_________
----------------------------------------------------------------------------------------------------------------------------- ---------------------

Electronic copy available at: https://ssrn.com/abstract=3808661

 QUESTIONÁRIO - BASE
10. Qual a sua situação profissional principal, ou seja, a situação da atividade profissional que consome mais
tempo de trabalho? (Ler as alternativas)
1) Assalariado registrado (Carteira Assinada) 9) Desempregado (Procura emprego) (ir p/ a 23)
2) Assalariado sem registro 10) Desempregado (Não procura emprego) (ir p/ a 24)
3) Funcionário Público/ Concursado/ Militar 11) Só dona de casa (ir p/ a 24)
4) Autônomo regular (Paga ISS) 12) Só aposentado (ir p/ a 24)
5) Profissional Liberal 13) Só estudante/ não trabalha, nem faz estágio (ir p/ a 24)
6) Empreendedor/ Comerciante/ Empresário 14) Só vive de rendas (ir p/ a 24)
7) Free-lance / Bico
8) Estagiário/aprendiz (Remunerado) (ir p/ a 12) 15) Outros (ANOTE) ________________
--------------------------------------------------------------------------------------------------------------------------------------------------
Exceto para Estagiário/ aprendiz
11. Atualmente o(a) Sr(a) exerce um cargo de Gerente, Coordenador(a), Analista ou Auxiliar/ Assistente?
1) Gerente 3) Analista 5) Outro. Especifique: ______
2) Coordenador(a) 4) Auxiliar/ Assistente
----------------------------------------------------------------------------------------------------------------------------- ---------------------
12. Quanto tempo o Sr(a) passou procurando esse emprego/ trabalho/ essa ocupação profissional? (Ler as
alternativas)
1) Não lembra (não ler) 5) Entre 7 e 9 meses
2) Menos de 1 mês 6) Entre 10 meses e menos de 1 ano
3) Entre 1 e 3 meses 7) 1 ano ou mais
4) Entre 4 e 6 meses
--------------------------------------------------------------------------------------------------------------------------------------------------
13. Quanto tempo o Sr(a) está trabalhando nesse emprego? (Ler as alternativas)
1) Não lembra (não ler) 4) Entre 4 e 6 meses 6) Entre 10 meses e menos de
2) Menos de 1 mês 5) Entre 7 e 9 meses 1 ano
3) Entre 1 e 3 meses 7) 1 ano ou mais
----------------------------------------------------------------------------------------------------------------------------- ---------------------
14. Nos últimos 12 meses o Sr(a) foi promovido no seu trabalho, ou seja, teve uma promoção de cargo ou
aumento de salário no último ano?
1) Não 2) Sim
---------------------------------------------------------------------------------------------------------------------------- ----------------------
15. Qual o seu rendimento mensal com o seu trabalho principal? (Ler as alternativas)
1) Até R$1.000 4) Entre R$ 3.001 e R$4.000 7) Recusou responder (não ler)
2) Entre R$ 1.001 e R$2.000 5) Entre R$ 4.001 e R$5.000
3) Entre R$ 2.001 e R$3.000 6) Mais de R$5.000
----------------------------------------------------------------------------------------------------------------------------- ---------------------
16. Em qual segmento o Sr (a) trabalha: Indústria, Comércio, Serviços ou Agricultura? (RM)
1) Indústria 3) Serviços (ir p/ a 18)
2) Comércio (ir p/ a 18) 4) Agricultura (ir p/ a 18)
----------------------------------------------------------------------------------------------------------------------------- ---------------------

Electronic copy available at: https://ssrn.com/abstract=3808661

 QUESTIONÁRIO - BASE
17. Em que setor Industrial o Sr(a) trabalha? (RM) (Ler as alternativas)
1) Não sabe (não ler) 6) Eletro eletrônica
2) Extração mineral (mineração e petróleo) 7) Máquinas e equipamentos
3) Alimentos, Bebidas ou Fumo 8) Veículos
4) Têxtil, vestuário, calçados e couro 9) Construção
5) Química 10) Outro. Especifique_________
----------------------------------------------------------------------------------------------------------------------------- ---------------------
18. De uma maneira geral, o Sr(a) diria que está muito satisfeito(a), satisfeito(a), nem satisfeito(a), nem
insatisfeito(a), insatisfeito(a) ou muito insatisfeito(a) em trabalhar no Setor (repetir o segmento em que o
entrevistado trabalha: Comércio/ Serviço/ Indústria/ Agricultura)?
1) Muito Satisfeito(a) 4) Insatisfeito(a)
2) Satisfeito(a) 5) Muito insatisfeito(a)
3) Nem Satisfeito(a), Nem Insatisfeito(a) 6) Não sabe/ não opinou (não ler, nem estimular)
------------------------------------------------------------------------------------------------------ ------------------------------------------
19. Se o(a) Sr(a) pudesse ou tivesse oportunidade o(a) Sr(a) trocaria o Setor em que o(a) Sr(a) trabalha para
trabalhar em outro setor, ou seja, o(a) Sr(a) deixaria de trabalhar no (repetir o segmento em que o
entrevistado trabalha: Comércio/ Serviço/ Indústria/ Agricultura) para trabalhar em outro setor? (Caso o
entrevistado diga que trocaria, especificar o setor em que gostaria de trabalhar)
1) Não trocaria de setor 3) Sim, trocaria pelo Setor de Serviços
2) Sim, trocaria pelo o Setor Industrial 4) Sim, trocaria pelo Setor Agrícola
2) Sim, trocaria pelo o Setor de Comércio
----------------------------------------------------------------------------------------------------------------------------- -------------------
20. O(a) Sr(a) exerce tem alguma outra atividade remunerada?
1) Não (ir p/ a 24) 2) Sim
----------------------------------------------------------------------------------------------------------------------------- ---------------------
21. Essa sua outra atividade remunerada, é no Setor da: Indústria, Comércio, Serviços ou Agricultura?
(RM)
1) Indústria 3) Serviços
2) Comércio 4) Agricultura
----------------------------------------------------------------------------------------------------------------------- ---------------------------
22. Qual o seu rendimento mensal com essa sua outra atividade remunerada? (Ler as alternativas)
1) Até R$1.000 (p/ 24) 5) Entre R$ 4.001 e R$5.000 (p/ 24)
2) Entre R$ 1.001 e R$2.000 (p/ 24) 6) Mais de R$5.000 (p/ 24)
3) Entre R$ 2.001 e R$3.000 (p/ 24) 7) Recusou responder (não ler) (p/ 24)
4) Entre R$ 3.001 e R$4.000 (p/ 24)
----------------------------------------------------------------------------------------------------------------------------- ---------------------

Electronic copy available at: https://ssrn.com/abstract=3808661

 QUESTIONÁRIO - BASE
23. Qual o principal motivo que o(a) Sr(a) atribui para o fato de ainda não ter conseguido uma colocação
profissional? (Ler as alternativas)
1) O mercado de trabalho está difícil
2) Não tenho os conhecimentos técnicos necessários para conseguir uma vaga
3) Não tenho as habilidades práticas necessárias para conseguir uma vaga
4) Não tenho o perfil pessoal necessário para conseguir uma vaga
5) Outro. Especifique: __________________
--------------------------------------------------------------------------------------------------------------------------------------------------
24. O Sr(a) estudou ou estuda na modalidade Regular ou na modalidade de Educação de Jovens e Adultos/
Supletivo/ CEBEJA?
1) Estudou na modalidade regular 2) Educação de Jovens e Adultos/ Supletivo/ CEBEJA
----------------------------------------------------------------------------------------------------------------------------- ---------------------
25. A maior parte dos seus estudos foi feita em Instituições Privadas ou Públicas?
1) Instituições Privadas 2) Instituições Públicas
----------------------------------------------------------------------------------------------------------------------------- ---------------------
26. Atualmente o Sr(a) está estudando?
1) Não (Verificar se a escolaridade é superior ou mais, em caso positivo, ir para a 28, em caso negativo, ir para a 30)
2) Sim
----------------------------------------------------------------------------------------------------------------------------- ---------------------
27. Atualmente o Sr(a) está cursando o ensino fundamental, médio regular, médio técnico, graduação, pós-
graduação, ou algum curso de qualificação profissional?
1) Ensino Fundamental (ir p/ 30) 4) Graduação
2) Ensino Médio regular (ir p/ 30) 5) Pós-graduação
3) Ensino Médio técnico (ir p/ 30) 6) Algum curso de qualificação profissional (ir p/ 30)
----------------------------------------------------------------------------------------------------------------------------- ---------------------
Para os que já possuem ensino superior ou estão cursando a graduação ou pós-graduação:
28. Qual a área do curso superior que o(a) Sr(a) está cursando/ cursou? (ESPONTÂNEA)

----------------------------------------------------------------------------------------------------------------------------- ---------------------
Para os que já possuem ensino superior ou estão cursando a graduação ou pós-graduação:
29. Em que ano o(a) Sr(a) concluirá/ concluiu o Ensino Superior? (ESPONTÂNEA)

--------------------------------------------------------------------------------------------------------------------------------------------------
Pensando em cursos de Capacitação/ Qualificação Profissional, gostaria que o(a) Sr(a) dissesse se
considera extremamente importante, muito importante, importante, pouco importante ou sem importância
cada um dos itens que lhe vou ler.
1) Extremamente importante 4) Pouco importante
2) Muito importante 5) Sem importância
3) Importante 6) Não sabe/ não opinou (não ler, nem estimular)

30. Reputação da Instituição que oferta o curso |____|

31. Conhecimento e formação dos Professores |____|

Electronic copy available at: https://ssrn.com/abstract=3808661

 QUESTIONÁRIO - BASE
32. Relacionamento entre Professores e alunos |____|
33. Instalações físicas da Instituição que oferta os cursos |____|
34. Ambiente entre os alunos/ companheirismo |____|
35. Forma de ensinar/ didática das aulas |____|
36. Variedade de cursos ofertados pela Instituição |____|
37. Saídas profissionais/ estágios ofertados |____|
----------------------------------------------------------------------------------------------------------------------------- ---------------------
38. Nos últimos 2 anos, o(a) Sr(a) participou de algum curso/ treinamento profissional?
1) Não (ir p/ a 40) 2) Sim
----------------------------------------------------------------------------------------------------------------------------- ---------------------
39. Qual foi a área desse curso/ treinamento que o(a) Sr(a) participou? (Ler as alternativas)
1) Não se recorda (não ler) 4) Construção Civil 7) Mecatrônica
2) Gestão 5) Alimentos e Bebidas 8) Outro. Especifique _____
3) Tecnologia da Informação 6) Segurança no Trabalho
----------------------------------------------------------------------------------------------------------------------------------- ---------------
40. Quais dos seguintes conhecimentos técnicos o(a) Sr(a) acredita que são muito importantes para o
sucesso na ocupação profissional que o(a) Sr(a) tem /pretende ter? (RM) (Ler as alternativas)
1) Não sabe (não ler) 4) Gestão Financeira 7) Mecânica de automóveis
2) Matemático/ quantitativo 5) Gestão de pessoas
3) Informática 6) Trabalhos manuais 8) Outros, especificar: _____
----------------------------------------------------------------------------------------------------------------------------------------------- ---
41. Quais das seguintes habilidades práticas o(a) Sr(a) acredita que são muito importantes para o sucesso
na ocupação profissional que o(a) Sr(a) tem /pretende ter? (RM) (Ler as alternativas)
1) Não sabe (não ler) 8) Liderança
2) Comunicação escrita 9) Gestão de pessoas e processos
3) Comunicação verbal 10) Negociação
4) Conhecimento do negócio 11) Organização
5) Cumprimento de metas 12) Planejamento
6) Flexibilidade
7) Trabalho em equipe 13) Outros, especificar: _____
----------------------------------------------------------------------------------------------------------------------------- ---------------------
42. Quais das seguintes atitudes o(a) Sr(a) acredita que são muito importantes para o sucesso na ocupação
profissional que o(a) Sr(a) tem /pretende ter? (RM) (Ler as alternativas)
1) Não sabe (não ler) 8) Entusiasmo
2) Agilidade 9) Equilíbrio emocional
3) Criatividade 10) Foco nos resultados
4) Disponibilidade 11) Saber ouvir
5) Empatia 12) Tomar decisões
6) Ética
7) Empreendedorismo 13) Outros, especificar: _____

Electronic copy available at: https://ssrn.com/abstract=3808661

 QUESTIONÁRIO - BASE
----------------------------------------------------------------------------------------------------------------------------- ---------------------
43. De uma maneira geral o(a) Sr(a) diria que lhe faltam mais conhecimentos técnicos, habilidades práticas
ou competências interpessoais para ter mais sucesso na sua vida profissional?
1) Não sabe (não ler, nem estimular) 4) Habilidades práticas (saber fazer)
2) Nada (não ler, nem estimular) 5) Competências interpessoais
3) Conhecimentos técnicos (saber como fazer) 6) Outro. Especifique: ___________
----------------------------------------------------------------------------------------------------------------------------- ---------------------
44. O(A) Sr(a) pretende, nos próximos 2 anos, fazer algum curso/ treinamento para desenvolver essas
competências?
1) Não (ir p/ a 46) 2) Sim
----------------------------------------------------------------------------------------------------------------------------- ---------------------
45. Qual a área desse curso/ treinamento que o(a) Sr(a) pretende/ gostaria de fazer? (RM) (Ler as
alternativas)
1) Não sabe (não ler) 10) Energia 19) Química
2) Alimentos e bebidas 11) Gestão 20) Refrigeração e climatização
3) Automação 12) Gráfica e editorial 21) Segurança no trabalho
4) Automotiva 13) Logística 22) Tecnologia da informação
5) Celulose e papel 14) Madeira e mobiliário 23) Telecomunicações
6) Construção 15) Meio ambiente 24) Têxtil e vestuário
7) Couro e calçados 16) Metalmecânica
8) Educação 17) Metrologia 25) Outro. Especifique _____
9) Eletroeletrônica 18) Polímeros (borracha e plástico)
----------------------------------------------------------------------------------------------------------------------------- ---------------------
Apenas para os Não Egressos:
46. O(A) Sr(a) conhece ou já ouviu falar nos cursos do Sesi/ Senai ou IEL?
1) Não (ir p/ a 49) 2) Sim
----------------------------------------------------------------------------------------------------------------------------- ---------------------
TODOS
47. De uma maneira geral, a imagem que o(a) Sr(a) tem dos cursos do Sesi/ Senai/ IEL é ótima, boa, regular,
ruim ou péssima?
1) Ótima 3) Regular 5) Péssima
2) Boa 4) Ruim 6) Não sabe/ não opinou (não ler)
----------------------------------------------------------------------------------------------------------------------------- ---------------------
48. Nos próximos 2 anos, o(a) Sr(a) diria que há uma possibilidade muito alta, alta, nem alta, nem baixa,
baixa ou muito baixa que o(a) Sr(a) venha a fazer algum dos cursos do Sesi/ Senai/ IEL?
1) Muito alta 3) Nem alta, nem baixa 5) Muito baixa
2) Alta 4) Baixa 6) Não sabe/ não opinou (não ler)
----------------------------------------------------------------------------------------------------------------------------- ---------------------

Electronic copy available at: https://ssrn.com/abstract=3808661

 QUESTIONÁRIO - BASE

49. Dados de contato.

Nome: ______________________________

Bairro: _______________________________

Email: ________________________________

Telefone de contato: _____________________

Electronic copy available at: https://ssrn.com/abstract=3808661