Entrepreneurship education and entry into self-employment among university graduates

Patrick Premand*, Stefanie Brodmann*, Rita Almeida*, Rebekka Grun*, Mahdi Barouni**

* The World Bank

** IREDU, Université de Bourgogne

Corresponding author:

Patrick Premand

The World Bank

1818 H Street

Washington, DC 20433

[email protected]

Tel.: 001 202 458 7109

1
Highlights

• We examine the effectiveness of entrepreneurship education in Tunisia.

• Participation in an entrepreneurship track was randomized among interested university

students.

• Participation led to a small increase in self-employment one year after graduation.

• Overall employment rates among former participants remained unchanged.

• The program improved business skills, but had mixed impacts on personality and

entrepreneurial traits.

2
Abstract

Entrepreneurship education has the potential to enable youth to gain skills and create their own

jobs. In Tunisia, a recent curricular reform created an entrepreneurship track providing business

training and coaching to help university students prepare a business plan. We rely on randomized

assignment of the entrepreneurship track to identify impacts on students’ labor market outcomes

one year after graduation. The entrepreneurship track led to a small increase in self-employment,

but overall employment rates remained unchanged. Although business skills improved, effects

on personality and entrepreneurial traits were mixed. The program nevertheless increased

graduates’ aspirations towards the future.

Keywords: entrepreneurship education; training; self-employment; skills; program evaluation;

randomized control trial.

3
Acknowledgements

We thank Luca Etter for invaluable support and leadership during the implementation phase of

the program, as well as Maaouia Ben Nasr and Leila Baghadi for excellent support during

follow-up data collection. This study would not have been possible without the support of Ali

Sanaa, Rached Boussema, Fatma Moussa and multiple staff in the Tunisian Ministère de

Formation professionnelle et de l’Emploi (MFPE) and the Ministère de l’Enseignement supérieur

et de la Recherche scientifique (MERS). We thank Hosni Nemsia and his team for effective data

collection. The entrepreneurship track was implemented by the Government of Tunisia with

support from the World Bank. We acknowledge the financial support of the Spanish Impact

Evaluation Trust Fund and of the Gender Action Plan at the World Bank. We are very grateful

for useful comments provided by participants in the Tunis workshop on activation, the Oxford

CSAE conference, as well as seminar participants at the World Bank and IZA. We are also

grateful to the editor, two anonymous referees, Harold Alderman, Najy Benhassine, Roberta

Gatti, David Margolis, Daniela Marotta, Eileen Murray, Berk Ozler, Bob Rijkers, Thomas

Sohnesen and Insan Tunali for helpful comments. The views expressed in this paper are those of

the authors and do not necessarily reflect those of the World Bank or any of its affiliated

organizations. All errors and omissions are our own.

4
1. Introduction

Entrepreneurship has long been considered a key element of the growth process (Schumpeter,

1912; Baumol, 1968). Some theories of entrepreneurship model individuals’ decisions between

entry into wage and self-employment. The theoretical literature highlights the role of wealth in

shaping this decision in the presence of capital-market imperfections (Banerjee and Newman,

1993; Ghatak and Jiang, 2002). Heterogeneity in individual preferences (Kihlstrom and Laffont,

1979) as well as in ability or entrepreneurial skills (Jiang et al., 2010) can also affect

occupational choices. Since entrepreneurial ability is not necessarily innate, education and

training programs that seek to shape these entrepreneurship skills are multiplying around the

world. Still, the evidence that these programs can effectively facilitate entry into self-

employment remains thin (Valerio et al., 2014).

The role of entrepreneurship in the development process is eliciting increasing attention from

policymakers and scholars (Naudé, 2014). In developing countries, only a small share of the

labor-force is employed in wage jobs (Gindling and Newhouse, 2012). In economies with limited

creation of private-sector wage jobs, entrepreneurship-support interventions are promising policy

options for the creation of more attractive skilled jobs. In this context, many policymakers

consider that entrepreneurship education has a strong potential to enable youth to gain skills and

generate their own skilled jobs.

The Middle East and North Africa is one of the regions with the highest youth unemployment

rates among university graduates (Gatti et al., 2013; Groh et al., 2015). In Tunisia, 46 percent of

graduates of the 2004 class were still unemployed eighteen months after graduation (MFPE and

World Bank, 2009). Unemployment among youths holding a university degree increased from 34

percent in 2005 to 62 percent in 2012. In this context, Tunisia has attempted various reforms

5
aiming to promote employability or self-employment among university graduates. Among them,

a new entrepreneurship track was introduced into the undergraduate (licence appliquée)

curriculum in 2009. Students enrolled in the last year of their undergraduate degree were invited

to apply to the entrepreneurship track, which entailed business training as well as personalized

coaching sessions. Students could then graduate by writing and defending a business plan instead

of a traditional undergraduate thesis.

In this paper, we analyze the impact of the entrepreneurship track on labor market outcomes by

relying on randomized assignment of the program among applicants. The paper makes several

contributions to the empirical literature on entrepreneurship education and training. First, we

provide unique experimental evidence on the effectiveness of entrepreneurship education

delivered in school in shaping employment outcomes among university graduates. Moreover, it

is the first study providing such evidence outside OECD countries, and for the Middle East and

North Africa in particular. Second, whereas most studies on entrepreneurship training have

focused on its impacts on productivity of established entrepreneurs, our results complement the

more limited literature analyzing the impacts on entry into self-employment. Third, the paper

contributes to the broader literature on active labor market policies, which tends to focus on

programs targeting low-skilled youths or unemployed individuals. In contrast, our work looks at

the effectiveness of a training program for higher education graduates, before they enter the

labor-market. Lastly, we analyze the impacts of entrepreneurship training on a range of skills

such as business skills, personality dimensions or entrepreneurial traits. As such, the paper

provides a link between the economic literature on the effectiveness of training programs on

labor outcomes, and the broader psychology and entrepreneurship literature studying the specific

skills or traits associated with successful entry into self-employment.

6
Results show that entrepreneurship education significantly increased the rate of self-employment

among university graduates approximately one year after graduation. However, the effects are

small in absolute terms, ranging from 1 to 4 percentage points. Given the low prevalence of self-

employment in the population, these small absolute effects imply that program participants were

on average 46 to 87 percent more likely to be self-employed compared with graduates from the

control group. However, the employment rate among applicants remained unchanged, suggesting

a substitution from wage employment and into self-employment. Findings on intermediary

outcomes are consistent with the limited employment results: the program improved business

skills, but had mixed impacts on personality and little effects on entrepreneurial traits.

Nevertheless, participation in the entrepreneurship track heightened graduates’ aspirations

towards the future shortly after the Tunisian revolution.

The rest of the paper is organized as follows. Section 2 discusses the placement of the paper in

the literature. Section 3 briefly sets the country context and describes the entrepreneurship track.

Section 4 describes the randomized assignment and take-up of the entrepreneurship track.

Section 5 presents the empirical strategy. Section 6 discusses the main effects of the program on

labor market outcomes. Section 7 analyzes a range of skills as intermediary outcomes that can

contribute to explain the observed employment impacts. Section 8 concludes.

2. Related Literature

This paper relates with different strands of the literature. First, we relate directly to the empirical

evidence on the effectiveness of entrepreneurship education programs in shaping individual

skills and facilitating entry into self-employment. Several OECD countries provide

7
entrepreneurship education in school. Despite the popularity of these programs, the evidence on

their effectiveness remains thin. Peterman and Kennedy (2003) and Souitaris et al. (2007) find

some impacts of entrepreneurship training on entrepreneurial intentions among secondary school

and high-school students, respectively. In contrast, Oosterbeek et al. (2010) show that an

entrepreneurship education program had no effect on university students’ entrepreneurial skills

and had a negative effect on the intention of becoming an entrepreneur. A limitation of these

studies, however, is that they measure impacts on students’ intentions while in school, not on

actual project creation or employment outcomes after students have graduated and joined the

labor-market 1. Given this limited evidence-base, the effectiveness of entrepreneurship education

remains a topic of active debate. We provide unique evidence on the impacts of an

entrepreneurship track introduced in Tunisian universities on the labor-market outcomes of

participants one year after their graduation.

Second, we relate to a growing literature analyzing the effectiveness of entrepreneurship-support

interventions, including programs providing a mix of capital and skills (for a review, see Cho

and Honorati, 2014). Most studies on business training analyze whether the skills of existing

entrepreneurs can be strengthened to improve their productivity (for a review, see McKenzie and

Woodruff, 2014). Recent contributions show that business training can affect enterprise owners’

practices, although effects on employment or productivity are more limited (Karlan and Valdivia,

2011; Drexler et al., 2014; Bruhn and Zia, 2013; Klinger and Schündeln, 2011). In contrast,

fewer studies focus on whether business training can equip individuals with the skills required to

enter into self-employment. De Mel et al. (2014) show that business training targeted to women

in urban Sri Lanka affected business practices but not productivity among existing business

owners, and that the same training accelerated entry into self-employment in the short-run.

8
Fairlie et al. (2015) find limited overall treatment effects of entrepreneurship training in the

United States, although they find relatively larger short-term effects on business ownership

among individuals previously unemployed. We complement this limited literature analyzing the

effectiveness of entrepreneurship training in facilitating entry into self-employment.

Third, we also complement the broader literature on active labor market policies by documenting

the effectiveness of a training program for a high-skilled group of university students before they

enter the labor-force. Active labor market policies tend to focus on programs to foster

employability and productivity among low-skilled youths or unemployed individuals (for

reviews, see Kluve et al., 2010, or Almeida et al., 2012). Most of the existing evidence on

training programs in developing countries comes from Latin American programs and tends to

focus on the effect of providing technical and vocational training to low-skilled, at-risk youth on

their probability to enter wage employment (e.g. Attanasio et al., 2011; Card et al., 2011). The

active labor market literature in developing countries is comprehensive and casts doubts on the

cost-effectiveness of training programs (Almeida et al., 2012). The findings generally show that

trainees of more comprehensive programs are more likely to find a job and tend to have better

quality jobs than non-trainees, although differences in labor earnings are mixed. In contrast, this

paper isolates the impact of a training program for high-skilled youths before they enter the

labor-market, focusing on youths’ transition from university to work and the decision to enter

into self-employment. It is unclear a priori whether training programs should have relatively

larger or smaller impacts among the high-skilled. On the one hand, low-skilled youths have

lower human capital than university students, and as such the marginal returns to additional

training may be higher among them. On the other hand, high-skilled youths may face fewer

9
constraints to enter self-employment in the first place, so that the impact of entrepreneurship

training may be larger among them.

Finally and importantly, we relate also to the broader psychology and entrepreneurship literature

studying the skills or personality traits needed for successful entry into self-employment. A range

of attributes, including business skills, personality and entrepreneurial traits have been shown to

be associated with productivity or employment outcomes, including occupational choice and

entrepreneurship (Almlund et al., 2012). A line of research analyzes whether entrepreneurs have

significantly different personality or entrepreneurial traits (for reviews, see Brandstätter, 2011;

Zhao and Seibert, 2006; Rauch and Frese, 2007; Zhao et al., 2010; or Frese and Gielnik, 2014),

for instance by comparing entrepreneurs to managers or wage workers (e.g. de Mel et al., 2010;

Cobb-Clark and Tan; 2010). Other studies have analyzed whether the personal attributes needed

to enter self-employment are different than those needed to remain in self-employment

(Ciaverella et al. 2004; Caliendo et al., 2014). Studies that compare entrepreneurs to other groups

of individuals tend to find that entrepreneurs have some different personality or entrepreneurial

traits. However, these observed differences often document associations rather than causality,

and are not able to identify which skills or traits are more malleable. Another strand of the

literature shows that behavioral skills and personality remain malleable, particularly among

young adults (Almlund et al., 2011; Robins et al., 2001; Roberts and Mroczek, 2008). However,

the economic literature assessing the effectiveness of entrepreneurship training or

entrepreneurship education in improving labor-market outcomes does not usually document

impacts along these domains. Only a few papers provide evidence of training impacts on

behavioral skills, aspirations or attitudes, whilst also documenting impacts on labor-market or

business creation outcomes (Groh et al., 2012; Macours et al., 2013; Solomon et al., 2013). In

10
this paper, we also study the impacts on entrepreneurship training on a range of skills, including

business skills as well as personality and entrepreneurial traits. These are intermediary outcomes

that can contribute to explaining observed changes in the main employment outcomes. By doing

so, we are able to relate our findings with this broader psychology and business literature.

3. The entrepreneurship education track: business training and coaching

In Tunisia, both the graduation rate from university and the unemployment rate among tertiary-

educated youth have been increasing steadily. Access to post-secondary education has soared

over the past twenty years. Gross enrollment rates in tertiary education reached 34 percent in

2009, up from 12 percent in 1995 2. At the same time, unemployment among youth holding a

university degree increased from 34 percent in 2005 to 62 percent in 2012. 3 While tertiary-

educated youth made up less than 16 percent of those employed in the Tunisian labor market in

2010, they accounted for over 34 percent of the unemployed. In this context, the graduates’

employment challenge has become one of the main concerns for policymakers in Tunisia.

An innovative entrepreneurship track was introduced into the tertiary curriculum in the academic

year 2009/10. Until that point, during the last semester of the applied undergraduate curriculum,

students took an internship and wrote an academic thesis to graduate. In June 2009, the Ministry

of Education and Higher Education passed a reform creating an entrepreneurship education track

where students would receive business training and coaching to develop a business plan. In

August 2009, the Ministries of Education and Higher Education and of Vocational Training and

Labor jointly signed an order to allow students to graduate by submitting their business plan

instead of the traditional thesis. The newly established entrepreneurship track primarily aimed to

11
increase self-employment and foster an entrepreneurship culture among university graduates;

more broadly, the program also aspired to improve participants’ employment outcomes.

The program was launched in all Tunisian universities delivering licences appliquées in 2009. 4

Communication campaigns took place on campus and in the media to inform students about the

newly introduced alternative to the standard curriculum. Once in the entrepreneurship track,

students were offered support for developing a business plan through business courses and

personalized coaching. The entrepreneurship track provided students with: (i) business training

organized by the public employment office; (ii) external private sector coaches, mainly

entrepreneurs or professionals in an industry relevant to the student’s business idea; and (iii)

supervision from university professors in development and finalization of the business plan. For

each student, the final product of the program was a comprehensive business plan that served as

an undergraduate thesis.

Students received entrepreneurship education between February and June 2010, starting with

intensive business training to develop, modify, or refine an initial business idea. Students took

twenty days of full-time intensive training at local employment offices (Agence Nationale

d’Emploi et de Travail Indépendent, ANETI) between February and March 2010 5. The training

was called Formation Création d’Entreprise et Formation des Entrepreneurs (CEFE) and was

part of the active labor market menu offered by ANETI. The training was conducted in small

groups and included practical research on the ground, aimed at fostering participants’ behavioral

skills, business skills and networking skills.

The first part of the training consisted of four modules: (a) for the person, aimed at developing an

entrepreneurship culture and behavioral skills; (b) for the project, aimed at developing business

12
ideas through brainstorming and followed by SWOT (strength, weaknesses, opportunities, and

threat) analysis to isolate the best project idea for each participant; (c) for management, aimed at

general management principles (including leadership, partnership choice, organization, time

management, and planning tools); and (d) for marketing, aimed at identification of the relevant

market and market research (competition, clients, technology standards, etc.) as input into cost

analysis.

In a second phase, participants had the opportunity to present their ideas and get feedback from

bankers and business experts. Students participated in three training modules: (a) information

research, when participants focused on the challenges of implementing the projects; (b) business

plan education, when participants estimated key project parameters, such as investments,

revenues, and business expenses; and (c) building networks, when at least five outside experts or

business professionals were invited to give talks.

After the completion of the courses, students were assigned a personalized coach for support in

finalizing their business plan. Coaches were private sector entrepreneurs or specialized coaches

from ANETI or the Ministry of Industry’s network of start-up offices (Agence de Promotion de

l’Industrie, API). Students were expected to participate in eight coaching sessions, either

individually or in small groups. Coaching took place from April to June 2010. In parallel,

students also received supervision from one of their university professors. In June 2010, the

business plans were completed and defended by students at their university as part of the

graduation requirements.

Upon graduation, participants in the entrepreneurship track were invited to submit their business

plans to a business plan competition (concours des meilleurs plan d’affaires “entreprendre et

13
gagner”). A jury selected fifty winners who were eligible to receive seed capital for establishing

the business outlined in their business plans. The first five winners were eligible for seed capital

of 15,000 Dinars each (approximately US$10,000), the next twenty winners, 7,000 Dinars; and

the last twenty-five winners, 3,000 Dinars. Prizes were only paid if students had been able to

secure all the complementary funding needed to set-up their project. Fewer than 15 winners

fulfilled that requirements and actually cashed the prize.

4. Data, randomized assignment and compliance

4.1 Baseline Data

In 2009/10, 18,682 students were enrolled in the third year of licence appliquée in Tunisian

universities. All these students were invited to submit an application form for the

entrepreneurship track in November or December 2009. In total, 1,702 students (or 9.1 percent of

all eligible students nationwide) applied to receive entrepreneurship education. Of those, 1,310

students applied individually and 392 applied in pairs, so that in total, 1,506 projects were

registered.

Table 1 shows the number of enrolled students and applicants by gender and university. The

third column shows the distribution of the application rate. The last two columns present the

distribution of all students enrolled in the third year of licence appliquée in 2009/10 and of

applicants, by gender and by university respectively. Two-thirds of the applicants were women.

While this is a high participation rate for women, it is not higher than the proportion of women in

the overall population of enrolled students in Tunisia. Demand for the program varied across

universities. Differences in application rates are likely explained by variations in the

14
implementation of the information campaigns and intensity of advertisement about the program, 6

as well as regional variations in perceived employment opportunities. 7

[Table 1 here]

The baseline data comes from two sources. An application form was collected in November and

December 2009. The application form contained information about students’ socio-economic

background and employment experience, as well as their parents’. Additional information was

collected through a phone survey in January and February 2010. It included proxies for risk and

time preference. It also included measures of entrepreneurial traits similar to those selected by de

Mel et al. (2010) based on the entrepreneurship psychology literature. These measures capture a

range of traits that have been documented to characterize entrepreneurs: passion for work;

tenacity; polychronicity; locus of control; achievement; power motivation; centrality of work;

impulsiveness and personal organization.

The baseline survey suggests that the intervention responded to a strong demand from students

and that applicants had high expectations for their participation in the program: 88 percent of

applicants expected that the intervention would facilitate their insertion in the labor market, and

89 percent expected to have higher earnings thanks to the intervention.

4.2 Randomized assignment and compliance

15
Causal impacts of the program are identified based on randomized assignment to the

entrepreneurship track among applicants. The program was oversubscribed, so that half of the

applicants were randomly assigned to the entrepreneurship track and the other half were assigned

to continue with the standard curriculum. Randomized assignment was conducted at the project

level, stratified by gender and study subject. 8 757 projects were assigned to the treatment (658

individual projects, and 99 projects in pair) and 742 projects were assigned to the control group

(652 individual projects; 97 projects in pairs).

Table 2 presents the average baseline characteristics of the treatment group (assigned to the

entrepreneurship track) and the control group (assigned to the standard curriculum), as well as

differences between the two at baseline. 9

[Table 2 here]

Overall, randomization achieved good balance. Still, in any randomization procedure, a small

number of variables are expected to be statistically different across the treatment and control

groups. In this case, the difference in past experience in self-employment is statistically different.

We will return to this point below, as past experience may determine future occupational choice,

given the documented hysteresis associated with occupational choice among individuals over

time. Overall, there were few systematic differences between participants and non-participants

and the differences were quantitatively small. The empirical analysis will control for the

characteristics in Table 2 that are statistically different between the two groups at baseline.

Administrative records from the implementing agency reveal imperfect compliance with

assignment to the treatment group, which is mostly driven by drop-out of the entrepreneurship

track. Of the 856 students who applied and were randomly assigned to the entrepreneurship

16
track, 67 percent completed the business training, and 59 percent completed both business

training and coaching. Overall, 41 percent dropped out of entrepreneurship education prior to

completing both training and coaching. Table A2 in annex presents marginal effects from the

estimation of a logit model to describe the profile of students who complied with their

assignment to the entrepreneurship track and completed it.

Administrative data reveal high compliance for students assigned to the control group. The

twenty days of CEFE training were provided by employment offices so that some control

students may also have been able to take the training after graduating, although personalized

coaching would not have been available to them. Administrative and survey data show that take-

up of the CEFE training was low in the control group, with only twenty-nine students (or 3.4

percent of the control group) completing the CEFE training after graduation.

4.3 Follow-up data and post-revolution context

After the baseline data was collected and the randomization performed, students participated in

the entrepreneurship track between February and June 2010. Graduation took place in June 2010.

In October 2010, qualitative data was collected to gather students’, coaches’, and professors’

perceptions on the implementation of the intervention.

The follow-up data was collected through face-to-face interviews between April and June 2011,

approximately nine to twelve months after the end of the academic year. The instrument included

similar questions than the baseline, with the same measures of preferences and entrepreneurial

traits. Additional modules were introduced, including a detailed labor module similar to the

Tunisian labor force survey, a module on business skills related to the content of the training, a

module on networks and a module on access to credit. It also contained additional measures of

17
personality and aspirations. The measures of personality capture the commonly used Big Five

dimensions (John et al., 2008; Almlund et al., 2011): extraversion; agreeableness;

conscientiousness; emotional stability; and openness to experience. Specifically, the TIPI scale

developed by Gosling et al. (2003) was used to obtain brief measures of the “Big Five”

personality dimensions. Aspirations were measured through positive items as in the CESD

depression scale, capturing positive attitudes towards the future (Radloff, 1977).

Despite the high mobility of the population of graduates, thorough tracking procedures led to

high response rates at follow-up 10: 92.8 percent of the 1,702 applicants were tracked. 11 This low

level of attrition is noteworthy since many studies on entrepreneurship education suffer from

high attrition. Attrition was balanced and uncorrelated with treatment status. 12

The follow-up survey was conducted 3-6 months after the Tunisian revolution, which occurred in

January 2011. Therefore, the results discussed in this paper are obtained in the context of the

Arab Spring. In the follow-up survey, individuals were asked to report their perceptions on how

the revolution affected their employment opportunities. Graduates revealed positive outlooks,

including a stronger desire to look for employment, and less interest in migrating abroad than

before the revolution. They also stated that the revolution increased their prospects for wage and

self-employment. In addition, there was no difference in the relative intensity at which the

revolution affected graduates’ perception of opportunities in wage and self-employment. In other

words, students did not believe that the revolution disproportionately affected their chances to

obtain wage jobs or enter self-employment. As such, the post-revolution context in which the

results are obtained does not affect the internal validity of the findings.

18
5. Empirical identification strategy

5.1 Specifications

Identification of program impacts relies on the randomized assignment of applicants to the

entrepreneurship track. We first present intent-to-treat (ITT) estimates, measuring the impact of

offering business training and coaching independently of actual take-up. We estimate the

following individual-level intent-to-treat regression:

Yi = βTi+γXi+πis+εi (1)

where 𝑌𝑌𝑖𝑖 is the outcome of interest for student i at follow-up, Ti is a binary variable for being

randomly assigned to the treatment group, Xi is a set of control variables, πis are fixed effects for

each randomization strata (by gender and subject) and εi is a mean-zero error term. 13

We present results for three alternative specifications. In specification I, Xi includes a set of

control variables from the baseline application form. 14 In specification II, Xi contains a constant

and an expanded set of controls including those from the baseline application form as well as

additional variables from the baseline survey measuring entrepreneurial traits. 15 This expanded

set of controls reduces the effective sample size to 1,432 due to combined attrition in the baseline

phone survey and the follow-up survey. In specification I and II, standard errors are clustered at

the level of the randomization strata (by gender and study subject). Specification III includes the

same set of variables as in specification I but standard errors clustered by the governorate where

students live at baseline.

19
In addition to ITT estimates, we also present and briefly discuss “treatment on the treated”

(TOT) estimates for each of the three specifications. The last section documented that not all

students assigned to the treatment group remained in the program (and a few control students

took up the business training component of the entrepreneurship track). TOT estimates are

obtained by 2SLS by instrumenting actual completion of the entrepreneurship track with the

randomized assignment to treatment. We define program completion or compliance for students

in the treatment group as completing the business training and receiving coaching. 16

TOT estimates are local average treatment effects that measure the impact of the

entrepreneurship track for the students who complied with their assignment to the treatment or

control group. Very few students in the control group took the business training (CEFE provided

by the public employment agency) after graduation. In this sense, TOT estimates essentially

produce the average impact of the program for students who did not drop-out from the

entrepreneurship track.

Importantly, almost all the results below are robust across ITT and TOT estimates, with TOT

estimates of a larger magnitude as would be expected. Given the consistency of the results across

both sets of estimates, we mainly focus on discussing ITT estimates since they are our preferred

set of estimates and more directly policy-relevant 17.

5.2 Hypothesis and outcomes

The main question is whether entrepreneurship education (including business training and

coaching) promoted self-employment among tertiary graduates. To answer this question, we use

three alternative outcome indicators. The first captures whether the respondent owned a project

at any point over the twelve months prior to the survey. The second indicator captures self-

20
employment based on a seven day recall consistent with official ILO definitions used in

Tunisia 18. The third, and more conservative, indicator is based on a seven day recall, excluding

self-employed individuals in seasonal activities 19. None of the self-employment indicators

includes family workers.

As a second hypothesis, we test whether the entrepreneurship track increased overall

employment among beneficiaries. On the one hand, skills acquired through entrepreneurship

education may be transferable across occupations. The entrepreneurship track can potentially

equip students with skills valued by employers and as such increase graduates’ probability of

finding wage jobs. On the other hand, the assignment to the entrepreneurship track may induce a

substitution away from wage employment. For instance, the program may negatively affect the

probability that participants find wage jobs in the private sector, either because it equips students

with sub-optimal skills for wage employment or because the standard curriculum may be more

valuable to finding wage jobs since it includes writing an academic thesis and undertaking an

internship. To shed light on these potential mechanisms, we estimate the impact of the

entrepreneurship track on overall employment as well as its two main components, self-

employment (as above) and wage employment. We also measure the impact of the interventions

on other variables of employment status (unemployed, studying, inactive). 20 Finally, we estimate

the impact of the intervention on some employment characteristics, including hours worked,

earnings, self-reported reservation wage for public and private sector wage employment, whether

the individual has a contract or is covered by social security.

Third, we analyze potential mechanisms through which the intervention can affect employment

outcomes. These different channels relate to the content and objectives of the intervention

described in section 2. The entrepreneurship track aimed to provide participants with business

21
skills, technical knowledge and experience directly useful to produce a business plan. In parallel,

a component of the business training aimed to shape students’ behavioral skills and

entrepreneurship culture. However, the program was not grounded in psychological theory to

outline a clear set of behavioral skills or domains of personality it aimed to affect. In this context,

we test whether the intervention affected (i) business skills, (ii) personality dimensions, (iii)

entrepreneurial traits, and (iv) aspirations towards the future. Changes in these intermediary

outcomes can contribute to explain the observed employment impacts. We also test (and report

in the annex) alternative mechanisms that may affect graduates’ employment outcomes including

through changes in preferences, networks and access to credit.

6. Main Results: Labor market outcomes

This section discusses program impacts on labor market outcomes. The main findings are

reported in Table 3 including self-employment (Panel A), employment status (Panel B), and

employment characteristics (Panel C). Column one reports the number of observations; the

second and third columns report the sample means for the dependent variable in the control and

treatment groups. The next 4 columns present results from specification I followed by

specifications II and III. ITT estimates are in columns (1), (3) and (5), TOT estimates in columns

(2), (4) and (6).

[Table 3 here]

6.1 Self-employment

22
Estimates show that entrepreneurship education increased self-employment among participants

approximately one year after graduation. The positive impact of the entrepreneurship track on

graduates’ self-employment holds across a range of indicators, such as whether the individual

reported owning a project any point over the twelve months prior to the survey, whether he/she

was self-employed in any activity last week, or whether he/she was self-employed in permanent

activities last week. All indicators exclude family workers. Focusing on self-employment in any

activity in the last seven days (the official definition of self-employment in Tunisia), the ITT

estimate shows a 3 percentage point increase in the probability of being self-employed. For those

students who actually completed the entrepreneurship track (education and coaching), the TOT

estimate reveals a 5 percentage point increase in the likelihood of being self-employed in any

activity in the last week.

While increases in self-employment are robust across specifications and indicators, the estimated

effects are small in absolute terms, ranging for 1 to 4 percentage points for ITT estimates. Since

the rate of self-employment is low in the control group, these small absolute impacts lead to

relatively large effect sizes. Indeed, the average self-employment rate in the control group is 4.4

percent. 21 Therefore, a 3 percentage point increase in self-employment in any activity over the

last week is equivalent to a 68 percent increase over the self-employment rate in the control

group. Average effect sizes for intent-to-treat estimates range from 46 to 87 percent depending

on the specification and self-employment indicator.

As mentioned in section 4.2, a significant difference in past experience in self-employment was

observed between the treatment and control group at baseline. As table 3 shows, estimated

program impacts are robust across specifications, including when controlling for baseline

differences (specification I and III), as well as when controlling for a broad range of preferences

23
and entrepreneurial traits that are typically associated with the propensity to become self-

employment (specification II). It may still be possible that part of effects are driven by

unobserved individual-specific preferences for self-employment that we do not fully capture by

including all the baseline control variables (including past experience in self-employment).

6.2 Employment status

Table 3 (Panel B) shows that only 28 percent of graduates in the control group were employed

one year after graduation, contrasting with 48 percent being unemployed. 22 This highlights the

slow school-to-work transition among young university graduates.

While the program increased self-employment, we find no evidence that the program

significantly affected overall employment as captured by the likelihood of being employed in the

last seven days. In fact, estimates suggest a reduction in the probability that program

beneficiaries hold salaried employment. Even though the effect is not significant, the decrease in

wage employment is of the same magnitude as the significant increase in self-employment.

Similar to findings in Fairlie et al. (2015) in the US, these results suggest that the program

changed the composition of employment by inducing a partial substitution from wage

employment to self-employment for participants in the entrepreneurship track.

It is worth noting, however, that this shift from wage employment into self-employment may free

up job opportunities for students that do not participate in the entrepreneurship track. This may

lead to higher overall employment. Unfortunately, the design of this study does not allow us to

quantify this type of general equilibrium effects.

24
Overall, while the program increased graduates’ self-employment in a context where the

availability of wage jobs is limited, the results show that the entrepreneurship track did not

promote graduates’ chances of finding a salaried job nor did it have an impact on the probability

of being employed in any activity one year after graduation. This is partly explained by the fact

that the entrepreneurship track is only effective in increasing self-employment for a limited

(although significant) number of students. At the same time, the evidence does not support the

hypothesis that the entrepreneurship track would also better align students’ skills with

employers’ needs and improve their prospect of finding wage jobs. On the contrary, the results

suggest trade-offs between policies that aim to promote self-employment and policies that aim to

facilitate the transition from school to wage jobs.

6.3 Employment characteristics

Table 3 (Panel C) presents the impacts of the entrepreneurship track on employment

characteristics such as hours worked, earnings, having a contract, being covered by social

security, working in a large firm, and reservation wages. The variables capturing the

characteristics of employment (including earnings) contain zeros for those individuals not

working. Two other outcomes include whether the worker is employed in a job with social

security coverage and whether he/she has a written contract. These variables are binary; i.e., they

take a value of one if an individual is employed with social security coverage or has a written

contract and zero if he/she is not working at all or works without coverage or without a written

contract. This distinction allows us to shed some light on the program’s potential effect on entry

into higher-end, formal sector jobs.

25
The results show no significant impact of the entrepreneurship track on earnings or hours

worked, even if the estimates are positive for both variables. 23 The entrepreneurship track did not

promote entry into higher quality jobs among participants either. In particular, there were no

significant program impacts on employment in the formal sector or in the size of the firm where

graduates worked. These results are consistent with the findings that overall employment

remained unchanged.

The results also suggest that the program increased students’ reservation wage for private sector

jobs, i.e. the minimum wage at which an individuals would accept a job offer. Higher reservation

wage for private sector jobs is consistent with the program leading to greater valuation of self-

employment or entrepreneurial activities in general. This result can contribute to explain the

partial substitution from wage to self-employment documented above. In contrast, the program

did not affect the reservation wage for public sector jobs. This suggests that self-employment is a

substitute for private sector jobs, but not for public sector jobs. 24

7. Skills as Intermediary Outcomes

The previous section showed that the program increased self-employment among participants

without affecting their overall employment rates. Here we tease out the channels and

intermediary outcomes through which the program affected employment outcomes. This is done

by assessing program impacts on (i) business skills, (ii) personality dimensions, (iii)

entrepreneurial traits, as well as (iv) aspirations towards the future.

26
7.1 Business skills

Table 4 displays estimated program impacts on a range of skills. It uses the same specifications

as in Table 3. Panel A shows strong impacts on participants’ self-reported business skills. Results

show that beneficiaries are more likely to report having practical experience in undertaking

projects or in producing a business plan. They also report better knowledge about topics taught in

the entrepreneurship track 25. Students in the treatment group have knowledge of about 52% of

the content of business plans, 25 percentage points higher than the control group. When the

business skills score is standardized, it is clear that the impact on business skills is of large

magnitude (0.7 standard deviation for the ITT estimates). These impacts are closely related to the

core content of the business courses, which were relatively effective in imparting business

knowledge to participants. Still, not all students assigned to the entrepreneurship track fully

acquired the technical knowledge. This is fully consistent with the dropout patterns shown

above.

[Table 4 here]

7.2 Personality Dimensions and Entrepreneurial Traits

As discussed in section 3, the entrepreneurship track contained a module designed “for the

person”, aimed at developing entrepreneurship culture and behavioral skills. During qualitative

interviews undertaken prior to the follow-up survey, some facilitators stressed that one of their

main objectives was to change students’ personality and “turn them into entrepreneurs”. Despite

this general intention, the program was not grounded in a psychological theory, outlining a clear

set of personality dimensions or entrepreneurial traits that it desired to affect 26. Still, we test

27
whether the program impacted a range of personality dimensions and entrepreneurial traits that

are often associated with entrepreneurship in the literature.

Panel B and C of Table 4 document impacts on personality dimensions and entrepreneurial traits.

Measuring personality dimensions and entrepreneurial traits in large-scale surveys is

challenging. We are unable to use extensive measures, but rely instead on brief measures taken

from the literature. The five indicators in panel B capture the five dimensions of the most

commonly used theory of personality (Almlund et al., 2011; John et al., 2008): extraversion;

agreeableness; conscientiousness; emotional stability, and openness to experience. These are

measured based on the TIPI scale developed by Gosling et al. (2003). The scale is known to have

somewhat diminished psychometric properties, but was designed to study personality dimensions

in situations when the use of more extensive measures is not feasible 27. The nine indicators in

panel C capture a range of more specific entrepreneurial traits: impulsiveness; passion for work;

tenacity; polychronicity; locus of control; achievement; power motivation; centrality of work,

and personal organization. These entrepreneurial traits are measured as in de Mel et al. (2010),

who discuss how they stem from the psychology literature 28. All measures are internally

standardized so that they have a mean of 0 and a standard deviation of 1 in the control group.

Therefore, all coefficients can be interpreted in terms of standard deviations from the average

“level” in the control group.

The results show that the intervention led to measurable and significant changes in several “Big

Five” dimensions. These observed changes are consistent with the fact that personality has been

shown to be particularly malleable among young adults (Roberts and Mroczek, 2008). The

results suggest that participants in the entrepreneurship track increase their level of extraversion,

and decrease agreeableness, conscientiousness and emotional stability. The observed changes in

28
personality dimensions are of relatively small magnitude (0.1 to 0.2 standard deviations for ITT

estimates). In particular, these changes are of smaller magnitude than the changes in business

skills discussed above. When interpreting these findings in the context of the broader psychology

and entrepreneurship literature, the observed changes in personality are qualitatively mixed. In

fact, not all changes would appear to make individuals more inclined to be entrepreneurs. Indeed,

meta-reviews in the psychology and business literature have suggested that entrepreneurs tend to

score higher on conscientiousness, openness to experience, extraversion, and emotional stability,

and lower in agreeableness (for instance, see Brandstätter, 2011; Zhao and Seibert, 2006; Zhao et

al., 2010). As such, the observed increase in extraversion and decrease in agreeableness are

changes likely to be conducive to entry into entrepreneurship. However, the decrease in

emotional stability and conscientiousness are not. These mixed results are consistent with the

limited employment impacts documented above.

We interpret the observed changes in personality as broadly consistent with the treatment

group’s experience in the entrepreneurship track, and in particular how that experience differed

from the control group’s experience with the traditional curriculum. The significant increase in

extraversion means that former participants to the entrepreneurship track tend to have a stronger

outwards orientation and be more sociable. It is in line with some of the elements of the

entrepreneurship track seeking to make students more outspoken, inviting them to reach out to a

range of business professionals and more generally to expose them to experiences outside the

university circles. The decrease in agreeableness means that former participants to the

entrepreneurship track have a lower tendency to act in a cooperative, unselfish manner. This is

also broadly consistent with aspects of the entrepreneurship track such as defending business

29
ideas in front of professionals, applying negotiation skills, or getting immersed in a competitive

business environment.

In contrast, the observed decreases in conscientiousness and emotional stability are not changes

that are likely to be conducive to entrepreneurship. When analyzing changes in entrepreneurship

traits following an entrepreneurship education program, Oosterbeek et al. (2010) find faster

increases in entrepreneurial skills among the control group than among entrepreneurship

education students. A similar effect may be at play in our context. While it cannot be formally

confirmed since personality is not observed at baseline, the experience of the control group of

writing an academic thesis may be relatively more effective in shaping conscientiousness or

emotional stability than entrepreneurship education.

To summarize, the observed changes in personality are of smaller magnitude than the results on

business skills. They are also more mixed in the sense that not all the observed changes are likely

to be more conducive to entrepreneurship. We interpret these results to be consistent with the

overall limited employment impacts documented above.

Panel C of Table 4 documents limited program impacts on a range of entrepreneurial traits

beyond the “Big Five”. Results show significantly lower impulsivity, and higher centrality of

work among past participants to the entrepreneurship track. Both of these changes are

qualitatively more conducive to entrepreneurship (as per the discussion in de Mel et al. (2010)).

However, the magnitude of changes in these entrepreneurial traits is limited (0.1 standard

deviation for ITT estimates), and the results are not fully robust across specifications. All other

entrepreneurial traits, including power motivation and tenacity, remain unchanged.

30
The identification of program impacts on personality dimensions but not on entrepreneurial traits

deserves a discussion. The entrepreneurial traits included in the study do not necessarily map

into the broader personality dimensions: the observed changes in personality are not expected to

be the sum of some observed impacts on entrepreneurial traits. This is a broader issue in the

entrepreneurship literature, where many potentially relevant entrepreneurial traits are cited (e.g.

Rauch and Frese, 2007). It is not always evident how various entrepreneurial traits map to each

other, how to prioritize them, and how they contribute to personality dimensions. In addition, it

is a subject of debate whether analysis should focus on personality or on more specific

entrepreneurial traits. Some argue that analyzing higher-order dimensions is preferable. Zhao and

Seibert (2006) note that it is an empirical question whether lower-order traits provide useful

information beyond the personality dimensions. In contrast, Rauch and Frese (2007) or Frese and

Gielnik (2014) have shown that a range of specific entrepreneurial traits correlate with

entrepreneurship outcomes, and favor analyzing changes in specific traits. Caliendo et al. (2014)

find that both personality dimensions and specific entrepreneurial traits are associated with entry

into self-employment.

Overall, our results are qualitatively similar to Oosterbeek et al. (2010), who also do not find

impacts of entrepreneurship education on entrepreneurial traits. The fact that entrepreneurial

traits are unaffected may suggest that the program was not precise enough in targeting specific

changes in behavioral skills that are most relevant for entrepreneurship. Entrepreneurship

education programs may need to be better grounded in psychological theory to outline more

clearly the specific personality dimensions or entrepreneurial traits they seek to target, and how

they intend to do so.

31
7.3 Aspirations towards the future

Although the results on employment outcomes and skills are mixed, Table 4 (Panel D) shows

some consistent positive impacts on graduates’ aspirations towards the future. Participants report

being much more optimistic, more likely to feel like they are moving forward in life, or thinking

about how to move forward in life. Students assigned to the entrepreneurship track also reveal

having relatively more faith in the future compared to graduates from the control group. These

results are robust and are consistent across a range of different indicators measured

independently, including an “optimism” sub-scale composed of six different questions (as in de

Mel et al., 2010). Overall, these results offer strong evidence that program participants perceived

a heightened sense of opportunities for the future. These results contrast with some previous

evaluations of entrepreneurship education have not found similar effects (for instance

Oosterbeek et al., 2010).

7.4 Other potential mechanisms

Finally, Table A3 in annex contains a range of ancillary results documenting impacts on other

intermediary outcomes. There are several interesting findings. First, no impacts are observed on

preference parameters such as risk or time-preference. In contrast to skills, preference parameters

are stable and are not affected by the intervention. Second, while the entrepreneurship track

contributed to expand networks, participants’ business networks are not very active. Third, there

is no direct evidence that the intervention alleviated credit constraints.

32
8. Conclusion

This paper relies on randomized assignment to measure impacts of the introduction of a track

providing entrepreneurship education in Tunisian universities. This new track offered business

training and personalized coaching for students to develop a business plan for a project of their

choice. Students in this track had the option to graduate with a business plan instead of a

traditional thesis. We evaluate the impact of randomized assignment to the entrepreneurship

track on employment outcomes. We also assess changes in intermediary outcomes such as

business skills, personality dimensions, entrepreneurial traits, as well as aspirations towards the

future.

We find that assignment of university students to the entrepreneurship track was effective in

increasing self-employment among graduates approximately one year after graduation, but that

the effects are small in absolute terms, ranging from 1 to 4 percentage points in the probability of

being self-employed. Given the low prevalence of self-employment in the control group, these

small absolute effects imply that beneficiaries of the program were on average 46 to 87 percent

more likely to be self-employed compared with graduates from the control group. However, the

intervention did not increase the overall employment rate among beneficiaries. These results

suggest a substitution from wage employment to self-employment, similar to findings in Fairlie

et al. (2015) in the U.S. They are also consistent with findings that private sector reservation

wages are higher among participants in the entrepreneurship track.

We also shed light on the changes in skills that underlie the employment results. The program

improved business skills, but had mixed impacts on personality and little effects on

entrepreneurial traits. Overall, these contribute to explain the limited employment impacts found.

33
Nevertheless, the program did lead to positive impacts on graduates’ aspirations towards the

future. Given this, additional research on potential long-term effects of the entrepreneurship track

could explore the possibility that some employment impacts may take longer to materialize.

While the evaluation design does not allow us to formally disentangle the effects of the

entrepreneurship track (business training and personalized coaching) from the start-up capital

offered to winners of the business plan competition, we interpret the results as being mainly

driven by participation in the entrepreneurship track (training and coaching) and to the business

skills developed. Indeed, few winners cashed the monetary prize and most participants still

report lack of access to credit as the main remaining constraint to entry into self-employment.

The findings in this paper have thus several important policy implications. First, our results

suggest limited effectiveness of entrepreneurship education and training offered to university

students with relatively little screening or targeting. Second, the results highlight potential trade-

offs in designing programs aiming to foster self-employment and those geared towards

facilitating access to wage employment. Finally, the mixed results on personality and

entrepreneurial traits are consistent with the overall limited employment impacts.

Entrepreneurship education programs may benefit from a clearer definition of which specific

skills or entrepreneurial traits they seek to improve, along with a more comprehensive

articulation of how changes in skills are expected to lead to employment outcomes.

34
References

Almeida, R., J. Behrman, & Robalino D. (Eds.). (2012). The right skills for the job?: Rethinking

training policies for workers. Human Development Perspectives, The World Bank: Washington,

DC.

Almlund, M., A. L.Duckworth, J. J. Heckman and T. Kautz. (2011). “Chapter 1 - Personality

Psychology and Economics”. In S. M. Eric A. Hanushek and W. Ludger (Eds.), Handbook of the

Economics of Education. Elsevier.

Attanasio, O. P., A. D Kugler, and C. Meghir. (2011) “Subsidizing Vocational Training for

Disadvantaged Youth in Colombia: Evidence from a Randomized Trial.” American Economic

Journal: Applied Economics, 3(3).

Banerjee, A., and A. Newman (1993). "Occupational Choice and the Process of Development."

Journal of Political Economy 101(2).

Baumol, W. (1968). “Entrepreneurship in Economic Theory.” American Economic Review,

Papers and Proceedings, 58(2).

Brandstätter, H. (2011). "Personality aspects of entrepreneurship: A look at five meta-analyses."

Personality and Individual Differences 51(3).

Bruhn, M. and D. McKenzie. (2009). "In Pursuit of Balance: Randomization in Practice in

Development Field Experiments." American Economic Journal: Applied Economics 1(4).

Bruhn, M. and B. Zia. (2013). “Stimulating Managerial Capital in Emerging Markets: The

Impact of Business Training for Young Entrepreneurs”. Journal of Development Effectiveness,

5(2).

Caliendo, M., F. Fossen, and A. Kritikos. (2014). "Personality characteristics and the decisions to

become and stay self-employed." Small Business Economics 42(4).

35
Card, D., P. Ibarraran, F. Regalia, D. Roasas-Shady, and Y. Soares. (2011). “The Labor Market

Impacts of Youth Training in the Dominican Republic.” Journal of Labor Economics, 29(2).

Cho, Y., and Honorati, M. (2014). Entrepreneurship programs in developing countries: A meta

regression analysis. Labour Economics 28.

Ciavarella, M. A., A. K. Buchholtz, C. M. Riordan, R. D. Gatewood, and G. S. Stokes. (2004).

"The Big Five and venture survival: Is there a linkage?" Journal of Business Venturing 19(4).

Cobb-Clark, D. A. and M. Tan. (2010). “Noncognitive skills, occupational attainment, and

relative wages.” Labour Economics 18(1).

Credé M, Harms P, Niehorster S, & Gaye-Valentine, A. (2012). An evaluation of the

consequences of using short measures of the Big Five personality traits. Journal of Personality

and Social Psychology 102(4).

De Mel, S., D. McKenzie and C. Woodruff. (2010). “Who are the Microenterprise Owners?

Evidence from Sri Lanka in Tokman v. de Soto.” In: J. Lerner and A. Schoar (eds.) International

Differences in Entrepreneurship, NBER.

De Mel, S., D. McKenzie, and C. Woodruff. (2014). “Business training and female enterprise

start-up, growth, and dynamics: Experimental evidence from Sri Lanka." Journal of

Development Economics 106.

Drexler, A., G. Fischer and A. Schoar. (2014). “Keeping it simple: Financial Literacy and Rule

of Thumbs.” American Economic Journal: Applied Economics 6(2).

Fairlie, R., D. Karlan and J. Zinman. (2015). “Behind the GATE Experiment: Evidence on

Effects of and Rationales for Subsidized Entrepreneurship Training." American Economic

Journal: Economic Policy 7(2).

36
Frese, M. and M. M. Gielnik. (2014). "The Psychology of Entrepreneurship." Annual Review of

Organizational Psychology and Organizational Behavior 1(1).

Gatti, R., M. Morgandi, R. Grun, S. Brodmann, D. Angel-Urdinola, J. M. Moreno; D. Marotta,

M. Schiffbauer, E. Mata Lorenzo. 2013. Jobs for Shared Prosperity : Time for Action in the

Middle East and North Africa. Washington, DC: World Bank

Ghatak, M., and N. Jiang. (2002). "A Simple Model of Inequality, Occupational Choice, and

Development" Journal of Development Economics 69(1).

Gindling, T. H. and D. Newhouse. (2014). "Self-Employment in the Developing World." World

Development 56.

Gosling, S. (2003). "A very brief measure of the Big-Five personality domains." Journal of

Research in Personality 37(6).

Groh, M., N. Krishnan, D. McKenzie, and T. Vishwanath. (2012). Soft skills or hard cash? The

impact of training and wage subsidy programs on female youth employment in Jordan. World

Bank Policy Research Working Paper No. 6141, World Bank, Washington DC.

Groh, M., D. McKenzie, N. Shammout, and T. Vishwanath. (2015) "Testing the importance of

search frictions and matching through a randomized experiment in Jordan." IZA Journal of

Labor Economics 4(1).

Jiang, N., P. Wang, and H. Wu. (2010). "Ability-Heterogeneity, Entrepreneurship, and Economic

Growth." Journal of Economic Dynamics and Control, 34(3).

John, O. P., Naumann, L. P., and Soto, C. J. (2008). Paradigm Shift to the Integrative Big-Five

Trait Taxonomy: History, Measurement, and Conceptual Issues. In O. P. John, R. W. Robins, &

L. A. Pervin (Eds.), Handbook of personality: Theory and research (pp. 114-158). New York,

NY: Guilford Press.

37
Karlan, D., and M. Valdivia. (2011). “Teaching Entrepreneurship: Impact of Business Training

on Microfinance Institutions and Clients.” Review of Economics and Statistics, 93(2).

Kihlstrom, R., and J. Laffont. (1979). "A General Equilibrium Entrepreneurial Theory of Firm

Formation Based on Risk Aversion." Journal of Political Economy 87(4).

Klinger, B., and M. Schündeln. (2011). “Can Entrepreneurial Activity be Taught? Quasi-

Experimental Evidence from Central America.” World Development 39(9).

Kluve, J., F. Rother, and M. L. Sánchez-Puerta. (2010). “Training Programs for the Unemployed,

Low Income and Low Skilled Workers.” In Almeida, R., Behrman J. and D. Robalino (Eds),

Skills Development Strategies to Improve Employability and Productivity: Taking Stock and

Looking Ahead. World Bank, Washington DC.

Macours, K., P. Premand, and R. Vakis. (2013). "Demand Versus Returns ? Pro-Poor Targeting

of Business Grants and Vocational Skills Training." Policy Research Working Paper Series No.

6389, World Bank, Washington DC.

McKenzie, D. and C. Woodruff. (2014). “What are We Learning from Business Training and

Entrepreneurship Evaluations around the Developing World?" World Bank Research Observer

29(1).

MFPE and World Bank. (2009). “Dynamique de l’emploi et adéquation de la formation parmi les

diplômés universitaires. Analyse comparative des résultats de deux enquêtes (2005 et 2007).”

World Bank, Washington DC.

Naudé, W. (2014). “Entrepreneurship and Economic Development: Theory, Evidence and

Policy.’ In Currie-Alder, B., R. Kanbur, D. Malone and R. Medhora (eds). International

Development: Ideas, Experience, and Prospects. Oxford: Oxford University Press.

38
Oosterbeek, H., M. van Praag, and A. Ijsselstein. (2010). “The Impact of Entrepreneurship

Education on Entrepreneurship skills and motivation.” European Economic Review 54(3).

Peterman, N. E. and J. Kennedy. (2003). "Enterprise Education: Influencing Students’

Perceptions of Entrepreneurship." Entrepreneurship Theory and Practice 28(2).

Radloff, L.S. (1977). “The CES-D scale: A self-report depression scale for research in the

general population.” Applied Psychological Measurement 1.

Rauch, A. and M. Frese. (2007). "Let's put the person back into entrepreneurship research: A

meta-analysis on the relationship between business owners' personality traits, business creation,

and success." European Journal of Work and Organizational Psychology 16(4).

Roberts, B. W. and D. Mroczek. (2008). "Personality Trait Change in Adulthood." Current

Directions in Psychological Science 17(1).

Robins, R. W., R. C. Fraley, B. W. Roberts, and K. H. Trzesniewski. (2001). "A Longitudinal

Study of Personality Change in Young Adulthood." Journal of Personality 69(4).

Schumpeter, J. A. (1912). The Theory of Economic Development (R. Opie, Trans.). Cambridge,

MA: Harvard University Press.

Solomon, G., M. Frese, C. Friedrich and M. Glaub. (2013). "Can personal initiative training

improve small business success?: A longitudinal South African evaluation study." International

Journal of Entrepreneurship and Innovation 14(4).

Souitaris, V., Zerbinati, S. and A. Al-Laham. (2007). "Do entrepreneurship programmes raise

entrepreneurial intention of science and engineering students? The effect of learning, inspiration

and resources." Journal of Business Venturing 22(4).

Valerio, A., B. Parton, and A. Robb. (2014). Entrepreneurship Education and Training

Programs around the World: Dimensions for Success. Washington, DC: World Bank.

39
Woods, S. A., and Hampson, S. E. (2005). Measuring the Big Five with single items using a

bipolar response scale. European Journal of Personality, 19(5).

Zhao, H. and S. Seibert. (2006). "The Big Five personality dimensions and entrepreneurial

status: A meta-analytical review." Journal of Applied Psychology 91(2).

Zhao, H ; S. E. Seibert, and G. T. Lumpkin (2010). "The Relationship of Personality to

Entrepreneurial Intentions and Performance: A Meta-Analytic Review." Journal of Management

36(2).

40
 Table 1: Characteristics of Students in Licence Appliquée and Entrepreneurship Track

Number of Number of Distribution

students enrolled applicants to among students in Distribution
Application rate
in "Licence entrepreneurship "licence among applicants
Appliquée" track appliquée"
Female 12,539 1,129 9.0% 67.1% 66.3%
Gender
Male 6,143 573 9.3% 32.9% 33.7%
Gafsa 1,060 304 28.7% 5.7% 17.9%
Monastir 1,935 316 16.3% 10.4% 18.6%
Sfax 2,005 284 14.2% 10.7% 16.7%
Jendouba 1,550 216 13.9% 8.3% 12.7%
Ez-Zitouna 93 10 10.8% 0.5% 0.6%
University

Kairouan 1,237 109 8.8% 6.6% 6.4%

Carthage 2,012 120 6.0% 10.8% 7.0%
Gabès 1,798 108 6.0% 9.6% 6.3%
Sousse 2,351 141 6.0% 12.6% 8.3%
Tunis 1,010 61 6.0% 5.4% 3.6%
Tunis El-Manar 1,787 22 1.2% 9.6% 1.3%
Manouba 1,659 11 0.7% 8.9% 0.6%
Total 18,682 1,702 9.1% 100% 100%
Source : Observatoire National de l’Emploi et des Qualifications.

41
 Table 2: Baseline Balance for Effective Sample

N Mean Mean Difference St. Err. for

Variables from Application Form
Control Treatment (T-C) difference
Male 1580 0.33 0.32 -0.01 0.01
Applied in pair 1580 0.23 0.23 -0.00 0.01
Had a project idea when applying 1580 0.84 0.86 0.02 0.01
Age 1578 23.00 23.07 0.07 0.06
Single 1580 0.99 0.98 -0.01 0.01
Average grade in 2nd year of university (0-20) 1560 11.43 11.52 0.09 0.06
Lowest grade in 2nd year of university (0-20) 1443 6.16 6.22 0.05 0.14
Highest grade in 2nd year of university (0-20) 1539 17.05 17.09 0.03 0.10
Took entrepreneurship course 1580 0.74 0.76 0.02 0.01
Grade at entrepreneurship course 1184 13.61 13.53 -0.09 0.13
Knowledge of Arabic (1-5) 1580 3.71 3.67 -0.04 0.05
Knowledge of French (1-5) 1580 3.52 3.50 -0.03 0.04
Knowledge of English (1-5) 1580 3.09 3.08 -0.01 0.06
Ever worked 1580 0.70 0.72 0.02 0.02
Age at first job 1112 17.48 17.15 -0.32* 0.16
Duration of first job (months) 1105 5.93 6.35 0.42 0.76
First job: seasonal worker 1580 0.36 0.35 -0.00 0.03
First job: wage worker 1580 0.19 0.19 -0.01 0.02
First job: family aid 1580 0.06 0.05 -0.01 0.01
First job: self-employed 1580 0.08 0.11 0.03*** 0.01
Has experience related to project 1580 0.62 0.63 0.02 0.02
Knows an entrepreneur 1580 0.59 0.63 0.04 0.02
Has helped an entrepreneur 1580 0.27 0.30 0.03 0.02
Is willing to take risk 1580 0.96 0.93 -0.02* 0.01
Prefers 1000 TND for sure to a salary between 500 TND and 1500
1580 0.26 0.25 -0.01 0.02
TND based on performance
Household size 1579 6.49 6.51 0.02 0.10
HH earnings between 0 and 300 TND 1580 0.25 0.25 0.00 0.02
HH earnings between 301 and 500 TND 1580 0.30 0.30 0.00 0.02
HH earnings between 501 and 800 TND 1580 0.21 0.20 -0.02 0.02
HH earnings above 801 TND 1580 0.24 0.25 0.01 0.02
Family can provide financial support 1580 0.64 0.63 -0.02 0.02
Variables from Phone Survey
Years since baccalaureate 1432 3.38 3.37 -0.01 0.04
Grade at baccalaureate (0-20) 1432 10.64 10.60 -0.04 0.05
Prefers 1000 TND for sure in 6 months to 800 TND now 1432 0.51 0.55 0.05** 0.02
Willingness to take risk (1-10) 1432 7.41 7.46 0.05 0.07
Certainty equivalent for a lottery with a 50 percent chance of
1427 1,003.99 1,084.85 80.85** 33.57
winning 2000 TND and a 50 percent chance of winning 0 TND
Impulsiveness (normalized score) 1,432 0.00 -0.10 -0.10** 0.04
Passion for work (nomalized score) 1,432 -0.00 0.02 0.02 0.06
Tenacity (normalized score) 1,432 -0.00 0.11 0.11* 0.06
Polychronicity (normalized score) 1,432 0.00 -0.01 -0.01 0.04
Locus of control (normalized score) 1,432 -0.00 0.08 0.08 0.06
Achievement (normalized score) 1,432 -0.00 0.19 0.19*** 0.05
Power motivation (normalized score) 1,432 -0.00 0.01 0.01 0.07
Centrality of work (normalized score) 1,432 0.00 -0.04 -0.04 0.06
Personal organization (normalized score) 1,432 0.00 0.10 0.10* 0.06
Optimism (normalized score) 1,432 -0.00 0.05 0.05 0.05
Results reported: number of observations in survey; mean of treatment and control groups at baseline;
difference between the two; and standard errors for difference between treatment and control group.
Results for effective sample for estimation (excluding attritors at follow-up)
Effective sample is 1,580 for variables in the baseline application form, and 1,432 for the baseline phone
survey (due to the combined attrition in the baseline phone survey and the follow-up survey)
* significant at 10 percent. ** significant at 5 percent. *** significant at 1 percent

42
 Table 3: Main Results, Program Impacts on Employment Outcomes

Specification I Specification II Specification III

(1) (2) (3) (4) (5) (6)
N C T ITT St. Err TOT St. Err ITT St. Err TOT St. Err ITT St. Err TOT St. Err
A. Self-Employment
Self-employed in last 12 months 1,580 0.05 0.09 0.04*** 0.01 0.07*** 0.02 0.04*** 0.01 0.07*** 0.02 0.04*** 0.01 0.07*** 0.02
Self-employed (any activity in last 7 days) 1,580 0.04 0.08 0.03** 0.01 0.05** 0.02 0.03** 0.01 0.05** 0.02 0.03** 0.01 0.05** 0.02
Self-employed (in last 7 days, excluding seasonal activities) 1,580 0.03 0.04 0.01* 0.01 0.02* 0.01 0.01* 0.01 0.02* 0.01 0.01 0.01 0.02 0.02
B. Employment Status
Employed in last 7 days 1,580 0.28 0.29 -0.00 0.02 -0.00 0.04 -0.00 0.02 -0.00 0.04 -0.00 0.03 -0.00 0.05
Salaried worker in last 7 days 1,580 0.21 0.18 -0.03 0.02 -0.05 0.03 -0.03 0.02 -0.05* 0.03 -0.03* 0.02 -0.05* 0.03
Unemployed in last 7 days 1,580 0.48 0.49 0.01 0.03 0.01 0.05 0.02 0.03 0.03 0.05 0.00 0.03 0.00 0.05
Studying in last 7 days 1,580 0.19 0.18 -0.00 0.02 -0.01 0.03 -0.01 0.02 -0.02 0.03 0.00 0.02 0.00 0.03
Inactive in last 7 days 1,580 0.03 0.03 0.01 0.01 0.01 0.01 -0.00 0.01 -0.00 0.01 0.01 0.01 0.01 0.02
C. Characteristics of Employment
Hours worked in last week 1,570 8.55 9.35 0.66 0.98 1.12 1.64 0.48 0.99 0.76 1.54 0.69 0.93 1.17 1.48
Total labor earnings (monthly) 1,502 1.22 1.14 -0.06 0.13 -0.11 0.20 -0.08 0.13 -0.13 0.20 -0.06 0.12 -0.11 0.20
Total labor earnings (monthly, log) 1,502 74.79 88.97 17.51 33.86 29.80 56.38 17.50 33.23 27.97 51.90 10.70 14.06 18.30 22.68
Has contract 1,580 0.12 0.10 -0.02 0.02 -0.03 0.03 -0.02 0.02 -0.03 0.03 -0.02 0.02 -0.03 0.03
Covered by Social Security 1,580 0.05 0.06 0.01 0.01 0.01 0.02 0.01 0.01 0.02 0.02 0.01 0.01 0.02 0.02
Works in large firm 1,485 0.07 0.07 0.00 0.01 0.00 0.02 -0.00 0.01 -0.01 0.02 0.00 0.01 0.00 0.02
Reservation wage for private sector job (monthly) 1,579 473.50 491.20 17.13* 8.73 28.85** 14.68 12.03 9.56 19.09 14.91 18.76* 9.96 31.69* 16.18
Reservation wage for private sector job (log, monthly) 1,579 6.10 6.13 0.03* 0.02 0.06** 0.03 0.02 0.02 0.03 0.03 0.03* 0.02 0.06** 0.03
Reservation wage for public sector job (monthly) 1,577 487.86 491.45 4.15 7.30 6.99 12.00 -1.25 8.44 -1.99 12.96 5.18 8.66 8.75 13.85
Reservation wage for public sector job (log, monthly) 1,577 6.14 6.15 0.01 0.02 0.02 0.03 -0.00 0.02 -0.00 0.03 0.01 0.02 0.02 0.03
Note: number of observation, average for control group, average for treatment group, intent-to-treat (ITT) estimates, standard errors for ITT estimates, treatment-on-the-treated (TOT) estimates for completing

entrepreneurship training and attending coaching sessions, standard errors for TOT estimates. Standard errors clustered by strata in specification I and II, by governorate in specification III.
Estimates are obtained separately for each outcome and each specification.
In all specifications controls include strata fixed-effects (by gender and subject), as well as a set of control variables from the baseline application form, including age at first job,
previous experience in self-employment, prior experience with projects, prior experience in helping an entrepreneur and mother’s employment status.
Controls in specification II also include baseling behavioral skills of the respondents at baseline (patience, willingness to take risk, impulsiveness, tenacity and sense of achievement).
Sample size for Specification I and II: N=1,580. Sample Size for Specification III: N=1,432 (due to attrition in baseline phone survey)
* significant at 10%. ** significant at 5%. *** significant at 1%.

43
 Table 4: Intermediary Impacts on Business skills, Personality Dimensions, Entrepreneurial Traits and Aspirations Towards the Future

Specification I Specification II Specification III

(1) (2) (3) (4) (5) (6)
N C T ITT St. Err TOT St. Err ITT St. Err TOT St. Err ITT St. Err TOT St. Err
A. Business skills
Has practical experience in projects 1,577 0.37 0.48 0.10*** 0.02 0.17*** 0.04 0.11*** 0.02 0.17*** 0.03 0.10*** 0.02 0.17*** 0.03
Knows how to produce a business plan 1,579 0.45 0.77 0.31*** 0.03 0.53*** 0.05 0.32*** 0.03 0.52*** 0.05 0.31*** 0.03 0.52*** 0.05
Business Plan Knowledge (composite score) 1,579 0.27 0.52 0.25*** 0.03 0.42*** 0.04 0.26*** 0.03 0.41*** 0.04 0.24*** 0.02 0.41*** 0.03
Business Plan Knowledge (normalized score) 1,579 0.00 0.71 0.70*** 0.08 1.17*** 0.11 0.72*** 0.08 1.15*** 0.10 0.68*** 0.06 1.15*** 0.08
B. Personality Dimensions
Big 5: Extraversion (normalized score) 1,580 -0.00 0.11 0.10** 0.05 0.18** 0.07 0.05 0.05 0.08 0.08 0.11** 0.05 0.18** 0.08
Big 5: Agreeable (normalized score) 1,578 -0.00 -0.23 -0.24*** 0.05 -0.40*** 0.08 -0.23*** 0.05 -0.37*** 0.08 -0.25*** 0.04 -0.42*** 0.06
Big 5: Conscientiousness (normalized score) 1,577 -0.00 -0.14 -0.14** 0.05 -0.24*** 0.08 -0.13** 0.06 -0.21** 0.09 -0.14*** 0.04 -0.24*** 0.06
Big 5: Emotionnally Stable (normalized score) 1,579 0.00 -0.11 -0.11** 0.04 -0.18*** 0.07 -0.07* 0.04 -0.12** 0.06 -0.12** 0.05 -0.20** 0.08
Big 5: Openness (normalized score) 1,577 0.00 -0.02 -0.03 0.04 -0.05 0.06 -0.03 0.04 -0.05 0.06 -0.04 0.04 -0.06 0.06
C. Entrepreneurial Traits
Impulsiveness (normalized score) 1,573 0.00 -0.12 -0.12** 0.05 -0.21** 0.09 -0.11 0.07 -0.18* 0.11 -0.13** 0.06 -0.22** 0.09
Passion for work (normalized score) 1,579 0.00 0.03 0.03 0.05 0.06 0.09 -0.01 0.06 -0.02 0.09 0.04 0.05 0.06 0.09
Tenacity (normalized score) 1,576 -0.00 0.03 0.04 0.05 0.07 0.08 0.01 0.05 0.02 0.08 0.05 0.04 0.09 0.07
Polychronicity (normalized score) 1,577 -0.00 -0.05 -0.05 0.05 -0.08 0.08 -0.03 0.05 -0.05 0.08 -0.04 0.06 -0.07 0.10
Locus of control (normalized score) 1,579 -0.00 0.02 0.02 0.06 0.03 0.10 0.01 0.06 0.02 0.10 0.03 0.05 0.04 0.08
Achievement (normalized score) 1,576 -0.00 0.02 0.04 0.06 0.06 0.10 -0.02 0.05 -0.03 0.08 0.04 0.04 0.07 0.06
Power Motivation (normalized score) 1,574 0.00 -0.05 -0.04 0.05 -0.07 0.09 -0.10* 0.05 -0.15* 0.08 -0.04 0.04 -0.06 0.07
Centrality of work (normalized score) 1,578 -0.00 0.09 0.10* 0.05 0.16* 0.09 0.04 0.05 0.07 0.08 0.11** 0.04 0.19*** 0.07
Personal organization (normalized score) 1,580 -0.00 0.08 0.08 0.07 0.14 0.11 0.05 0.07 0.07 0.11 0.09 0.06 0.15 0.10
D. Aspirations towards the future
Optimism (normalized score) 1,578 -0.00 0.12 0.13*** 0.04 0.21*** 0.07 0.12** 0.05 0.18*** 0.07 0.13*** 0.04 0.22*** 0.06
Days feels moving forward 1,578 3.79 4.09 0.28** 0.11 0.47*** 0.17 0.23* 0.13 0.37* 0.19 0.25* 0.14 0.43* 0.25
Days thinking about how to move forward 1,578 5.62 5.87 0.23** 0.11 0.39** 0.19 0.25** 0.12 0.39** 0.19 0.21* 0.12 0.36* 0.21
Has more faith in future now than last year 1,574 0.52 0.57 0.04* 0.02 0.08* 0.04 0.05* 0.02 0.07* 0.04 0.05 0.03 0.08 0.05
Note: Same specification as table 3, see note therein.
* significant at 10%. ** significant at 5%. *** significant at 1%.

44
 Annexes

Table A1: Baseline Balance for Effective Sample, Additional Variables

N Mean Mean Difference St. Err. for

Variables from Application Form
Control Treatment (T-C) difference
Subject: Food/Agriculture 1580 0.02 0.02 0.00 0.00
Subject: Humanities 1580 0.16 0.17 0.01 0.01
Subject: Sciences 1580 0.09 0.09 -0.00 0.01
Subject: Accounting 1580 0.09 0.09 -0.01 0.01
Subject: Economics and Finance 1580 0.08 0.08 -0.00 0.00
Subject: Civil Engineering 1580 0.02 0.02 0.00 0.01
Subject: IT 1580 0.10 0.10 -0.00 0.00
Subject: Telecommunication 1580 0.06 0.05 -0.00 0.00
Subject: Languages 1580 0.09 0.09 0.00 0.00
Subject: Business Administration 1580 0.03 0.03 0.00 0.00
Subject: Marketing 1580 0.04 0.04 0.01 0.01
Subject: Sports and Tourism 1580 0.03 0.03 -0.00 0.00
Subject: Technical 1580 0.15 0.15 0.00 0.00
Subject: Others 1580 0.03 0.03 -0.00 0.00
University Ez-Zitouna 1580 0.00 0.01 0.00 0.00
University of Tunis 1580 0.03 0.04 0.00 0.01
University of Sousse 1580 0.08 0.08 -0.00 0.02
University of Monastir 1580 0.19 0.18 -0.01 0.02
University of Kairouan 1580 0.07 0.07 0.00 0.01
University of Sfax 1580 0.16 0.16 -0.00 0.02
University of Gafsa 1580 0.18 0.19 0.02 0.01
University of Gabes 1580 0.07 0.06 -0.01 0.01
University of Manouba 1580 0.01 0.00 -0.01 0.01
University of Tunis El-Manar 1580 0.01 0.01 -0.00 0.01
University of Carthage 1580 0.07 0.07 0.00 0.01
University of Jendouba 1580 0.13 0.13 0.00 0.02
Father has primary education 1580 0.41 0.45 0.03* 0.02
Father has secondary education 1580 0.43 0.39 -0.03 0.02
Father has tertiary education 1580 0.16 0.16 -0.00 0.02
Mother has primary education 1580 0.66 0.67 0.01 0.03
Mother has secondary education 1580 0.28 0.27 -0.01 0.03
Mother has tertiary education 1580 0.06 0.06 -0.00 0.01
Father is salaried worker 1580 0.36 0.34 -0.02 0.02
Father is self-employed 1580 0.27 0.27 0.00 0.02
Father is retired 1580 0.25 0.26 0.01 0.01
Father is unemployed 1580 0.02 0.02 0.00 0.01
Mother is salaried worker 1580 0.09 0.09 -0.00 0.02
Mother is self-employed 1580 0.07 0.08 0.02 0.01
Mother is retired 1580 0.02 0.03 0.02 0.01
Mother is unemployed 1580 0.03 0.04 0.01 0.01
Baccalaureate: Humanities 1432 0.24 0.23 -0.01 0.01
Baccalaureate: Economics 1432 0.19 0.18 -0.01 0.02
Baccalaureate: Sciences 1432 0.23 0.25 0.02 0.02
Baccalaureate: Math 1432 0.19 0.20 0.01 0.02
Baccalaureate: Technical 1432 0.15 0.13 -0.02 0.02
Results reported: number of observations in survey; mean of treatment and control groups at baseline;
difference between the two; and standard errors for difference between treatment and control group.
Results for effective sample for estimation (excluding attritors at follow-up)
Effective sample is 1,580 for variables in the baseline application form, and 1,432 for the baseline phone
survey (due to the combined attrition in the baseline phone survey and the follow-up survey)
* significant at 10 percent. ** significant at 5 percent. *** significant at 1 percent

45
Table A2: Compliance with Assignment to Entrepreneurship Track (Marginal Effects)

Training & coaching

Training completion
completion
(1) (2) (3) (4)
Unemployment rate in governorate 0.01* 0.00 0.01* -0.00
(0.00) (0.00) (0.00) (0.00)
Subject: Economics, Finance, Business -0.16* -0.20* -0.18* -0.18*
(0.04) (0.05) (0.04) (0.05)
Male -0.09** -0.08** -0.11* -0.10**
(0.04) (0.04) (0.04) (0.04)
Applied in pair 0.12* 0.12* 0.06 0.06
(0.04) (0.04) (0.04) (0.04)
Had a project idea 0.09*** 0.12** 0.05 0.09***
(0.05) (0.05) (0.05) (0.05)
Family can provide financial support 0.13* 0.06 0.17* 0.01
(0.04) (0.04) (0.04) (0.05)
Is willing to take risk 0.01 0.03 0.01 0.03
(0.07) (0.07) (0.07) (0.08)
Preference for stable salary -0.01 -0.00 -0.01 0.02
(0.04) (0.04) (0.04) (0.05)
Household income 301-500 TND (ref = <300 0.06 0.07 0.03 0.04
(0.04) (0.04) (0.05) (0.05)
Household income 501-800 TND 0.07 0.09*** 0.05 0.06
(0.05) (0.05) (0.05) (0.06)
Household income> 800 TND -0.02 0.02 -0.06 -0.02
(0.05) (0.05) (0.05) (0.05)
University of Sousse (ref = Tunis) 0.02 0.20*
(0.07) (0.06)
University of Monastir -0.01 0.20*
(0.06) (0.05)
University of Kairouan 0.11*** 0.25*
(0.06) (0.05)
University of Sfax -0.22* -0.19*
(0.07) (0.07)
University of Gafsa 0.12*** 0.29*
(0.07) (0.06)
University of Gabes -0.09 -0.07
(0.09) (0.09)
University of Jendouba 0.15** 0.15**
(0.06) (0.07)
Number of observations 856 856 856 856
R2 0.063 0.093 0.0611 0.1184
Notes: Results shown are marginal effects from a probit model on determinants of compliance
(defined as "training completion" in specification (1) and (2), and "training and coaching
completion" in specifications (3) and (4)) among sutdents assigned to the entrepreneurship track.
67% of selected students completed the training, 59% the training and coaching. * p<0.01, **
p<0.05, *** p<0.1

46
 Table A3: Ancillary Results (Intermediary Impacts on Preferences, Networks and Access to Credit)

Specification I Specification II Specification III

(1) (2) (3) (4) (5) (6)
N C T ITT St. Err TOT St. Err ITT St. Err TOT St. Err ITT St. Err TOT St. Err
A. Preferences
Willingness to take risk (0-10) 1,575 6.06 6.10 -0.02 0.14 -0.03 0.24 -0.04 0.14 -0.06 0.21 -0.02 0.10 -0.03 0.16
Certainty equivalent for lottery with 50% chance of winning 0 and
1,556 674.44 694.33 16.21 19.53 27.43 31.83 -2.89 17.97 -4.62 27.78 14.32 18.95 24.34 31.71
50% chance of winning 2000DT
Risk taker 1,556 0.18 0.18 -0.01 0.02 -0.01 0.03 -0.02 0.02 -0.03 0.03 -0.01 0.02 -0.02 0.03
Patience 1,577 0.27 0.29 0.02 0.02 0.03 0.04 0.00 0.02 0.00 0.03 0.02 0.02 0.03 0.03
B. Networks
Registered at Employment Office 1,702 0.78 0.82 0.04 0.02 0.07* 0.04 0.04 0.02 0.06* 0.03 0.03* 0.02 0.06* 0.03
Knows an employment agent 1,580 0.14 0.28 0.13*** 0.02 0.22*** 0.03 0.15*** 0.02 0.23*** 0.03 0.13*** 0.02 0.22*** 0.03
Number of times spoke to employment agent in last month 329 2.26 1.83 -0.31 0.39 -0.42 0.47 -0.14 0.30 -0.18 0.36 -0.32 0.49 -0.43 0.60
Knows an entrepreneur 1,580 0.44 0.49 0.05* 0.02 0.08* 0.04 0.04 0.03 0.06 0.04 0.05** 0.02 0.08*** 0.03
Number of times spoke to entrepreneur in last month 726 5.05 5.11 -0.01 0.65 -0.01 0.98 0.08 0.66 0.12 0.97 0.04 0.77 0.07 1.17
Knows a banker 1,580 0.25 0.31 0.06*** 0.02 0.09*** 0.03 0.06** 0.02 0.09*** 0.03 0.06** 0.03 0.10*** 0.04
Number of times spoke to a banker in last month 440 2.44 3.67 1.16** 0.53 2.00** 0.88 0.77 0.56 1.25 0.83 1.29* 0.74 2.27* 1.29
C. Access to credit
Knows how to apply for credit 1,580 0.20 0.22 0.02 0.02 0.03 0.03 0.02 0.02 0.04 0.03 0.02 0.02 0.03 0.03
Expect to be able to obtain credit 1,568 0.30 0.39 0.08** 0.04 0.14** 0.06 0.09** 0.04 0.14** 0.06 0.09*** 0.02 0.15*** 0.03
Has applied for credit 1,580 0.04 0.08 0.04** 0.02 0.06** 0.02 0.05*** 0.02 0.07*** 0.02 0.04* 0.02 0.06* 0.03
Has obtained credit 1,580 0.003 0.004 0.001 0.001 0.002 0.002 0.001 0.001 0.002 0.002 0.001 0.002 0.002 0.003
Note: number of observation, average for control group, average for treatment group, intent-to-treat (ITT) estimates, standard error for ITT estimates, treatment-on-the-treated (TOT) estimates for completing

entrepreneurship training and attending coaching sessions, standard errors for TOT estimates. Standard errors clustered by strata in specification I and II, by governorate in specification III.
Estimates are obtained separately for each outcome and each specification.
In all specifications controls include strata fixed-effects (by gender and subject), as well as a set of control variables from the baseline application form, including age at first job,
previous experience in self-employment, prior experience with projects, prior experience in helping an entrepreneur and mother’s employment status.
Controls in specification II also include baseling behavioral skills of the respondents at baseline (patience, willingness to take risk, impulsiveness, tenacity and sense of achievement).
Sample size for Specification I and II: N=1,580. Sample Size for Specification III: N=1,432 (due to attrition in baseline phone survey)
* significant at 10%. ** significant at 5%. *** significant at 1%.

47
Endnotes

1
Most existing studies on entrepreneurship education also face methodological issues since they are based on quasi-experimental data
and have limited external validity since they tend to use data from just a few schools.
2
Source: EduStats.
3
Source: Tunisia Labor Force Surveys.
4
This includes the following universities: Ez-Zitouna, Jendouba, Gabès, Gafsa, Tunis, Kairouan, Mannouba, Monastir, Carthage,
Sfax, Sousse, Tunis, and Tunis El-Manar.
5
The training lasted on average 6 hours a day, for a total of approximately 120 hours.
6
University professors played an instrumental role in informing students about the program. 84 percent of all applicants heard about
the program through professors, 39 percent from posters, and 17 percent from other students, friends, and relatives.
7
Application to the entrepreneurship track was particularly high in regions with the highest unemployment.
8
The fourteen groups of subjects were: Economics and Finance; Accounting; Business Administration; Marketing; Humanities;
Languages; Sciences; Technical; Telecommunication; Civil Engineering; IT; Sports and Tourism; Food/Agriculture, and Other.
9
Table 2 is presented for the effective sample used for estimations and composed of the 1,580 students that could be tracked at follow-
up. Results are almost identical when using the full baseline sample of 1,702 students. Table A1 in the annex presents differences in
treatment and control groups for several other characteristics.
10
Detailed contact information was collected in the baseline surveys. Most students register at employment offices upon graduation,
and contact information (phone numbers and address) from the employment office database was also collected and merged with the
data.
11
The attrition rate is lower than in other similar surveys. For instance, the attrition rate for the 2005 Tunisia graduate tracer survey
was 11 percent. Oosterbeek et al. (2008) have an attrition rate of 56 percent in their study of entrepreneurship education for university
students in the Netherlands.
12
Attrition in the full baseline sample was 7.2 percent at follow-up. At baseline, 10.1 percent of applicants could not be reached for
the complementary phone survey conducted in January and February 2010. Combined attrition in either this baseline phone survey or
the follow-up survey collected in 2011 is 15.9 percent. Attrition in both surveys was 1.4 percent. All of these attrition indicators were
balanced across treatment and comparison groups.
13
We include a binary variable for each randomization strata to increase power (Bruhn and McKenzie, 2009).
14
The controls include unbalanced variables from the baseline application form, such as age at first job, previous experience in self-
employment, prior experience with projects, prior experience in helping an entrepreneur, and mother’s employment status.
15
In particular, these include patience, willingness to take risk, impulsiveness, tenacity, and sense of achievement.
16
Results are very similar when compliance is defined as completing the business training only (not presented here).

48
17
As mentioned in section 4, while the baseline survey included a mix of administrative data and phone interviews, the follow-up
survey was conducted face to face. The empirical strategy also relies on randomized assignment and identifies program impacts
through first-difference in outcomes at follow-up (measured face to face). Specification I and III includes control variables from the
baseline administrative forms. Specification II includes some controls from the complementary baseline survey done by phone.
Results from these various specifications are robust. In addition, results are also robust by taking first-difference only without any
baseline control variables (this specification is not presented given the preferred specification includes baseline controls as mentioned
above). Taken together, the identification strategy based on randomized assignment along with the robustness of the results suggests
that having various sources of baseline data is not a cause of concern.
18
The indicators on project ownership and employment status do not fully match, and as such it is useful to consider the robustness of
results across these two indicators.
19
Project owners that report seasonal employment have activities in sectors such as agriculture, construction, craft productions or
other services where businesses may not operate year-round.
20
All these indicators are based on a 7 day recall period, consistent with official ILO definitions used in Tunisia.
21
This is in line with the low rate of self-employment among university-educated in Tunisia in general. Among 25 – 34 year old with
a university degree, 4.6 percent were classified as independent workers and 5.6 percent as employers according to the Labor Force
Survey of 2010.
22
These results are comparable with data from a tracer survey of university graduates from the class of 2004, which found that 46
percent of graduates were still unemployed 18 months after graduation (MFPE and World Bank, 2009).
23
Conditional on being wage-employed, i.e., only looking at employed individuals, the results suggest that program beneficiaries hold
slightly better quality jobs, as they were more likely to have full-time contracts, and less likely to be supported by a wage-subsidy
(stages d'Initiation à la Vie Professionnelle “SIVP”), but more likely to hold term contracts (contrats à durée déterminée “CDD”).
24
Consistent with a higher reservation wage for private sector jobs, individuals in the treatment group are more likely to state having
rejected a job offer because the salary is too low.
25
Respondents were asked questions about the components of a business plan (such as a supply assessment or a marketing plan), based
on which a composite knowledge score was created.
26
In contrast, some business skills training program have much more well-defined and specific entrepreneurial trait. For instance,
Glaub et al. (2014) focus on ‘personal initiative’.
27
The TIPI scale was designed to measure broad dimensions of personality. In large-scale surveys, it is typically not possible to use
extensive measures of the “Big Five” personality dimensions – including extensive instruments that also provide measurement of
specific facets within each big five dimension. The TIPI scale is based on polar factors within each dimension. Given the way the
scale was constructed, correlation between items within a dimension is not sufficiently informative to establish its statistical properties

49
(Gosling et al. 2003; Wood & Hampson, 2005). Gosling et al. (2003) argues that test-retest analysis is the most appropriate approach
to establish validity of the scale. They show that, while the scale has somewhat diminished psychometric properties compared to
longer scales, it displays substantial convergence, substantial test-retest reliability and expected patterns of external correlations. Credé
et al. (2012) discuss the psychometric limitations of short measures in more details. Cultural variations in personality dimensions also
remain an active field of research. The TIPI scale was used in the study in light of the large-scale coverage of the survey, as well as its
scope.
28
As for the TIPI scale of personality dimensions, entrepreneurial traits are observed through brief measures based on a few items.
Similar limitations as those highlighted in the previous note apply. The measures of entrepreneurial traits were based on de Mel et al.
(2010) since it was considered the most closely related to our study at the time of designing the baseline instrument.

50