Adult Education as a Stepping-Stone to Better Jobs: An Analysis of the Adult Education Survey in Turkey

Abstract

The objective of this article is to investigate whether adult education (AE) can be used as a tool in facilitating transitions to/in the labor market, using the cross-sectional Adult Education Survey of Turkey (2012). AE is defined as the nonformal education for individuals aged older than 25 years. The outcome of AE is measured by changing jobs for employed and finding a job for the unemployed. Concentrating on employed people, we analyze both the determinants and the outcome of participation in AE for the purpose of changing jobs; and second, concentrating on unemployed people, we analyze both the determinants and the outcome of participation in AE for the purpose of getting employed. We find that once young males who are already working participate in AE for changing work, independent of their education or how AE is financed, they can be successful in doing so. The results of the paper suggest that AE programs offered by the government can serve as a tool in increasing income of the less educated and the unemployed by facilitating their transition to the labor market.

Keywords

adult education employment labor market

Introduction

The quality of human capital is a fundamental component of productivity and sustainable growth (Becker, 1975, 1994). On one hand, an educated and skilled labor force is an important factor to increase the quality of human capital and hence the economic growth.¹ On the other hand, the speed of technological change requires continuous skill updating in the labor market. Individuals with low education are unemployed or work in temporary, precarious jobs with low wages. These individuals will continue to be unemployed or employed at low-wage jobs unless they acquire the required knowledge and skills.

Nevertheless, current formal education system, in general, is insufficient in providing skills and knowledge demanded by today’s employers (Desjardins, Milana, & Rubenson, 2006). Moreover, higher levels of prosperity have increased human longevity, which has prolonged the retirement age; and consequently, the average age of the working population has increased. Therefore, the demand for adult education has been on the rise (see Nilsson & Nyström, 2013, and the references therein). In addition, adult education is essential if the speed of technological change, and, therefore, the need of continuous skill updating in the labor market are taken into account.

In this article, adult education is defined as the nonformal education for individuals older than 25 years. Nonformal education has no formal curriculum requirements and no formal degrees earned after completion. Vocational training and professional certification programs, computer courses, and language courses are examples of nonformal education. By this definition, adult education has the potential to play an essential role in reducing the risk of poverty and social exclusion by increasing the competitiveness and by providing the knowledge and skills required for the labor market (see De Witte et al., 2013; Tilak, 2002; Veen & Preece, 2005).

With regard to the labor market in Turkey, despite improvements such as increase in labor market participation, employment, and formal employment since the early of 2000s, it still faces structural challenges (e.g., high inactivity among youth and women, high unemployment rates, skill mismatch, low productivity, low job creation compared with increase in labor force). Related to this, Turkey has been implementing several incentive programs to increase the labor force participation of women and youth; it introduced incentive programs for firms to decrease their costs in order to increase the job creation and implemented anti-informality action plans over the past decade. In addition to that, Turkish Employment Agency (İŞKUR) and Ministry of National Education work together to increase the skills of labor force. Nevertheless, although education level of labor force has been increasing, a substantial share of firms in Turkey perceive that inadequacy of the education of the workforce limit their ability to grow (and create jobs), operate effectively, and increase competitiveness (own calculations from micro data of Enterprise Survey 2015-2016 of Turkey). According to the 2015 Household Labor Force Survey conducted by Turkish Statistical Institute (TurkStat), nearly 59% of the labor force, 55% of unemployed, and 59% of employed people have secondary education level or lower. In addition, Survey of Adult Skills conducted by the OECD (Organisation for Economic Co-operation and Development) shows that adults (aged 16-65 years) in Turkey perform with below-average proficiency in all three domains—namely, literacy, numeracy, and problem-solving skills in a technology-rich environment.² Considering the literacy scores, 45.7% of adults in Turkey attain only Level 1 or below, which is considerably higher than the OECD average of 18.9%. In numeracy, 50.2% in Turkey attain Level 1 or below for which the OECD average is 22.7% (OECD, 2016). Moreover, if the future workforce skills are gauged by employing the results of Programme for International Student Assessment (PISA), which covers 15-year-old students near the end of their compulsory education and assesses key knowledge and skills, namely, science, reading, and mathematics, Turkey’s performance is not promising either (OECD, 2017). The Ministry of National Education aims to increase the quality in education as it has done for quantity (Republic of Turkey Ministry of National Education, 2015). In this regard, several measures have been undertaken, such as increasing the number of teachers (to decrease the student–teacher ratio), increasing the opportunities for on-the-job training for teachers, and integrating technology to teaching. Turkey has also some initiatives to improve the skills of unemployed people in order to increase their employability. Nevertheless, the studies that aim to evaluate the impact of skill training programs are rare. According to one of them, which was carried out by the World Bank in 2013, although there is negligible impact of ISKUR training programs (which are very widespread programs in Turkey) on employment, the impact on quality of employment is found to be significant with little variation of impact across age, gender, or level of education. Therefore, to have a better paid job in Turkey either today or in the future, well-designed adult education programs become an important tool to provide the necessary skills that are sought for in the labor market.

The objective of this article is to investigate whether adult education can be used as a tool in facilitating transitions to/in the labor market, using the Adult Education Survey (AES) conducted in 2012 by TurkStat. As labor market plays a major role in the lives of the poor in Turkey³ (see Dayioglu & Demir-Seker, 2016), transitions to/in the labor market are expected to have a significant impact on monetary poverty in Turkey as well.

Through its in-depth statistical analysis, this article contributes to the literature in at least two dimensions. (1) There is a paucity of work on the relationship between adult education and labor market outcomes in Turkey; therefore, the article serves to fill this gap in the literature. (2) Existing studies on adult education using micro data⁴ were conducted mostly on developed countries. This article presents a developing country case as a contribution to the international literature.

The article is organized as follows: The next section gives a brief literature review where the motivation and the contribution to the literature are underlined. Then the “Data, Model, and Descriptive Statistics” section presents the data, methodology, and descriptive analysis of adult education in Turkey. Estimation results are presented in the penultimate section. The final section concludes the article.

Literature Review

The factors that may affect the participation in adult education have been investigated in the field of education as well as in other disciplines such as economics, psychology, and sociology. The existing studies mainly focused on personal, financial, and demographic characteristics of the individuals in participation to adult education. For example, Merriam (2008) summarizes the barriers to adult education participation as individual’s general approach to learning and access to education and procedures as well as geographic, demographic, and socioeconomic conditions. Additionally, Roosmaa and Saar (2011) suggest that time constraints, transportation difficulties, nonaccessibility to education related information, and unemployment as well as psychological barriers such as lack of self-confidence are the most important barriers to participation in adult education.

There are other studies that discuss how barriers to participation in adult education would evolve in time and under different conditions. Among these, Rubenson and Desjardins (2009) suggest that family-related and work-related barriers to participation in adult education are eliminated at older ages. Although time limitations prevent individuals from attending education activities in early phases of their work life due to long working hours, in the later phases, time constraints reduce as the working schedule become less restrictive. For people endowed with higher education, however, procedural and individual barriers become less important.

Boeren, Nicaise, and Baert (2010) discuss that individuals make their decisions on participating adult education based on cost–benefit analysis. Older people, as a consequence of lower work-related expected return to education, participate less in adult education programs compared with young individuals. Family education background of the individual has an impact on the education choices. Antikainen (2005) suggests that family structure and experience in education and work create different dynamics in adult education participation.

Another determinant of participation in adult education is gender. A number of studies in the literature suggest that the tendency of women in participating in adult education is higher, while other studies report the opposite claiming that barriers to participation in adult education is higher for women. The studies supporting the former argument are Denton, Pineo, and Spencer (1990) with Canadian data, Lopes and Fernandes (2011) with Portuguese data, Domińczak and Lis (2013) with European data, and Saar, Unt, and Roosmaa (2014) with Estonian data. Kleinert and Matthes (2009) and Boeren (2011), on the other hand, are among the latter group of studies focusing on the barriers for women to participate in adult education. Kleinert and Matthes (2009) provide a work-related explanation to this tendency. They suggest that men tend to participate in adult education activities that are work related, while women prefer education activities that are self-development oriented, supporting that traditional gender roles are still effective.

There are other studies that focus on work-related adult education activities, which is the main focus of the current article. For example, Boateng (2009) studies the relation between participation in adult education and labor market and reports that 80% of the participation of individuals in adult education activities is work related, based on the first AES of EuroStat implemented in 2007. The results of the survey show that while the work schedule is the most important obstacle for men in participating in adult education activities, for women, family responsibilities is the number one obstacle. Using Scottish data, Riddell and Weedon (2009) focus on adult education tendencies of small- and medium-size enterprises. They find that lower skilled workers in manufacturing firms are less likely to receive encouragement and support from their employers. Analyzing the EuroStat AES data, Badescu, Garrouste, and Loi (2011) suggest that the most educated workers receive higher training especially in some new member states (Poland, Cyprus, Bulgaria, and Romania). Moreover, their estimates show that less educated workers are significantly less likely to be trained.

The expectation of a wage increase is another reason of participating in the nonformal adult education. Blanden, Buscha, Sturgis, and Urwin (2012) find a causal effect of adult education on medium-run return for women but no effect for men using British Household Panel Survey for the period 1991-2006. In a cross-country study, Kilpi-Jakonen, De Vilhena, Kosyakova, Stenberg, and Blossfeld (2012) find that the adult education participation leads to greater benefits in career progress, but the results are more pronounced in more developed countries. To provide the causal effect of adult education on labor market outcomes in Switzerland, Schwerdt, Messer, Woessmann, and Wolter (2012) use a voucher program for adult education as a randomized field experiment. They find no significant average effects but find significant heterogeneous effect on earnings, particularly in favor of low-educated individuals. As to the on-the-job training, a particular form of nonformal adult education, Vignoles, Galindo-Rueda, and Feinstein (2004) find a positive impact on earnings. However, the impact decreases if training is received by all workers, since normally firms tend only to train the workers who are more likely to gain more from training.

There is increasing international evidence on the employment impact of skills training programs. The findings are mostly mixed. Some programs yield with positive results, especially in terms of increasing employment probabilities, but some appear to offer little benefit to participants. For instance, while Knipprath and De Rick (2015) find evidence in favor of positive impact of nonformal adult education on the probability of being employed for low-qualified young adults in Flanders, Belgium, Schwerdt et al. (2012) find no effect of voucher-based adult education participation on employment in Switzerland. Card, Kluve, and Weber (2010) evaluate 199 programs and conclude that in the short run (after 1 year) many programs exhibit weak and insignificant effects, but after 2 years, positive returns are evident. Indeed, it is critical how the program is designed (Betcherman, Olivas, & Dar, 2004). For instance, young women are found to benefit from the well-designed training program for youth in Colombia, where female employment is low (Attanasio, Kugler, & Meghir, 2011). Identifying the needs of a target population and correspondingly assigning individuals to suitable programs is the first step (Almeida, Behrman, & Robalino, 2012). In addition, Kluve (2016) finds that the time profiles of “work-first” style job search assistance programs differ from the profiles of “human capital” style training programs. The former tends to have larger short-term effects, whereas the latter has larger impacts in the medium or longer run.

This brief survey of literature shows that skills and educational acquisitions of individuals become insufficient in time due to economic and social change, technological advances, and inefficiencies in the formal education system. Therefore, it is necessary for the individuals to acquire the basic and vocational knowledge in order to catch up the requirements of today’s world (Knowles, Holton, & Swanson, 2014).

In this respect, adult education has the potential to fill a void by providing literacy courses for the socially excluded women and vocational training for low-income individuals and migrants. For example, the share of the illiterate in world population is 18%, and two thirds of them are women. Although the illiterate ratio in Turkey is lower, 82% of the illiterate are women. This fact is explained by economic problems, marriage at young age, inequality of opportunities in education, and child labor. Therefore, adult education has the potential to be used as a tool to reduce poverty and social exclusion in the world as well as in Turkey (Bilir, 2009).

Data, Model, and Descriptive Statistics

Data: Adult Education Survey

The AES, which is a cross-sectional data source, covers adults’ participation in education and training (formal, nonformal, and informal learning) and is one of the main data sources for the European Union lifelong learning statistics that has been implemented in 2007 and 2012.

The survey consolidates various education and training types under three broad headings as follows⁵:

Formal education is defined as education provided by the system of schools, colleges, universities, and other formal educational institutions that normally constitute a continuous “ladder” of full-time education for children and young people, generally beginning at the age of 5 to 7 years and continuing up to age 20 or 25 years.

Nonformal education is defined as any organized and sustained learning activities that do not correspond exactly to the above definition of formal education. Nonformal education may therefore take place both within and outside educational institutions and cater to people of all ages. Depending on national contexts, it may cover educational programs to impart adult literacy, life skills, work skills, and general culture.

Informal learning is defined as intentional learning that is less organized and less structured than the previous types. It may include, for example, learning events (activities) that occur in the family, in the work place, and in the daily life of every person on a self-directed, family-directed, or socially directed basis.

Turkey has also implemented this survey at the national level, and the data of this study are based on the AES conducted by TurkStat in 2012. The survey is representative for Turkey because all settlements within the Republic of Turkey boundaries are covered.

In the AES-2012 of Turkey, an “adult” is defined as an individual aged 18 years and older. However, we use the sample of individuals aged 25 years and older, as the remaining sample has not been made publicly available by TurkStat. The survey is composed of questions on participation in adult education and demographic variables, such as age-group, gender, and education attainment level.

Model

To investigate who is more likely to participate in the various type of adult education, we estimate the following Probit models:

\Pr (a t t e n d_{i} = 1 | X) = β_{0} + {β^{'}}_{1} X_{i} + {β^{'}}_{2} Z_{i} + {β^{'}}_{3} W_{i} + ε_{i},

(1)

where i is an index for the individual. The dependent variable attend_i takes the value 1 if the individual participates in the related adult education program and 0 otherwise. While the vector X_i includes individual characteristics (gender, age, marital status, education, being a student, and father’s education), vector Z_i includes household characteristics (number of individuals in the household, the number of working individuals in the household, the number of children in the household who are aged 4 years or younger, and household income), and vector W_i includes adult education characteristics (sponsor of the adult education).

Table 1 presents the variables that are used in this study and their explanations. The descriptive statistics of these variables are presented in Table A1 in the appendix.

Table 1.

List of Variables on Individual Characteristics, Family Background, Sponsorship of Programs, and Adult Education Participation.

Variable	Explanation	Value
Panel A: List of variables on individual characteristics, family background, and sponsorship of programs
female	Gender of the individual	1 = female; 0 = male
age	Age of the individual
age ²	Age square of individual divided by100
married	Marital status of the individual	1 = married; 0 = others
education1	Education status: no degree from any formal education institutions	1 = no degree; 0= others
education2	Education status: graduated from primary school^a	1 = primary school; 0= others
education3	Education status: graduated from secondary school	1 = secondary school; 0 = others
education4	Education status: graduated from high/vocational school	1 = high/vocational school; 0 = others
education5	Education status: graduated from postsecondary school and higher	1 = postsecondary school and higher; 0 = others
attend-formal	Participation in formal education activities status in the past 12 months	1 = participated; 0 = not participated
father-edu	Education status of father: higher than secondary education	1 = secondary school or higher; 0 = others
hhsize	Number of households
hhchild	Number of children ⩽4 years old among the households
hhlabor	Number of working individuals in the households
hhcons	Household consumption proxied from Household Income Survey
finance-self	The costs of adult education program are paid by the participant or his or her relatives	1 = financed by the participant; 0 = otherwise
finance-gov	The costs of adult education program are paid by the government or related institutions	1 = financed by the government; 0 = otherwise
finance-employer	The costs of adult education program are paid by employer	1 = financed by the employer; 0 = otherwise
Panel B: List of variables on adult education participation
attend	Participation in adult education activities status in the past 12 months	1 = participated; 0 = not participated
attend-w	Reason to participate in adult education: work related (within participating sample)	1 = work related; 0 = others
working-attend	working individuals among those who participate in adult education activities in the past 12 months	1 = participated working; 0 = participated not working
attend-cw	Reason to participate in adult education: changing work (within participating and working sample)	1 = changing work; 0 = others
outcome-cw	Outcome of the adult education participated for the purpose of changing work (within participating and working samples)	1 = successful in finding new job/promotion/higher wage/new duties; 0 = others
attend-fw	Reason to participate in adult education: finding work (within participating and nonworking samples)	1 = finding work; 0 = others
outcome-fw	Outcome of the adult education participated for the purpose of finding work (within participating and nonworking samples)	1 = successful in finding work; 0 = others

Currently, the compulsory primary school education in Turkey is 8 years. However, in our sample, there are individuals who attended school when the compulsory primary school education was 5 years.

Independent variables

The first group of variables in Table 1 (Panel A) is related to characteristics and family information of individuals who are surveyed. Variables female, age, age², and married represent the general characteristics of the individuals. These variables are heavily used in the adult education studies. For instance, Denton et al. (1990), Domińczak and Lis (2013), Kleinert and Matthes (2009), Lopes and Fernandes (2011), and Saar et al. (2014) checked for the effect of gender on adult education participation. Rubenson and Desjardins (2009) and Boeren et al. (2010), on the other hand, investigated the relationship between age and adult education participation.

Family background is another variable that is used in adult education studies (see, e.g., Antikainen, 2005). In this study, we used educational attainments of the parent as an indicator of family background. The variable father-edu represents the education level of the father of the individual. We also considered mother’s education of the individuals in our sample. However, the education level of elder women in Turkey is very low, and as a consequence, we could not get enough variation for mother’s education variable. This observation is consistent with Dincer, Tekin-Koru, and Askar (2016) who use the AES-2007 survey for Turkey.

The other variables related to family of individuals are hhsize, hhchild, hhlabor, and hhcons. The number of individuals in the household is represented by hhsize, whereas the number of working individuals in the household is denoted by hhlabor. The number of children in the household who are ⩽4 years of age is represented by hhchild.

The last variable that is needed about the family is income. However, the decision of participating in adult education is mostly governed by the permanent income of the households, which is generally proxied by household consumption, hhcons. Although income and consumption are widely used to evaluate household living standards and welfare, consumption is more appropriate in the case of developing countries due to inadequate official income records and the greater measurement error for income data (see, e.g., Deaton, 1997; Ravallion, 1992). Besides, the income is a categorical variable with different brackets of household income in the AES. Moreover, consumption data are not readily available in the AES. Therefore, hhcons is imputed by using the Household Budget Survey. Following Elbers, Lanjouw, and Lanjouw (2003), consumption is regressed on a bunch of household characteristics (e.g., household size and location) and individual characteristics (e.g., household head’s age, education, employment status, gender, and marital status) using Household Budget Survey. Estimated parameters from this model are then applied to the regressors in the AES to provide an estimate of the (unobserved) consumption.

Last, there are three variables that represent the sponsor of adult education activities: finance-self, finance-gov, and finance-employer.

For the different models we estimate, we employ different combinations of these explanatory variables.

Dependent variables

We first estimate the determinants of participation in adult education. The dependent variable in this analysis is attend.

As the focus of the article is on labor market implications of the adult education, we investigate work-related participation in the adult education. The dependent variable of this model is attend-w. Work related is used to describe participating in adult education programs for reasons such as finding a job, changing jobs, getting promoted, as a part of compulsory training at work, and so on.

Work-related adult education participation is analyzed in two subsamples: working sample and not working sample.⁶

For the working subsample, we analyze the participation in adult education for finding new job/promotion/higher wage/new duties, which we call as changing job. The dependent variable for this subsample is attend-cw.

However, as we focus only on the working individuals, we encounter a sample selection problem. Therefore, we employ the Heckman (1979) procedure to control for this sample selection bias.⁷ We first run a selection equation with a binary dependent variable that takes the value 1 if the participating individual is working and takes the value 0 otherwise (working-attend). The independent variables of the selection model are female, age, age², married, education, hhsize, hhchild, and hhlabor. We use the number of working individuals in the household (hhlabor) as the exclusion restriction, and so it is not included in the model that investigates participation in adult education for changing work (outcome equation). In other words, it is assumed that the number of working individuals in the household affects the probability of individual’s working (by increasing network channels, etc.), but it does not affect the participation in adult education for changing work. We include this variable in the outcome model and find a statistically insignificant coefficient that validates our assumption. The inverse of Mill’s ratio⁸ obtained from this regression is included in the equation of participation to adult education for changing work. In addition to the variables in the selection equation (other than the exclusion restriction variable), outcome equation has the variables of educational attainment of the father (father-edu) and finance type of the adult education (finance-self, finance-gov, finance-employer) as well.

In the next step, we investigate who are more likely to change their work as a result of adult education participation. The dependent variable is outcome-cw.

The survey asks questions about the attendance to adult education within the past 12 months. Then, if individuals have attended adult education, they are asked about the outcome of that education. Particularly, the survey asks these questions to the individuals who attended adult education: By the help of that education (1) whether they found a (new) job or not, (2) whether they were promoted or not, (3) whether their wages increased or not. We accept that the individual is successful in finding a new job/promotion/higher wage if he or she replied (2) and (3) as “yes,” and for the employed individuals (1) as “yes.” Moreover, if an individual is unemployed and replied (1) as yes, we consider that individual as successful in finding a job. As the probability of an individual who participated in adult education at closer dates to survey date to find/change a job is smaller, our outcome variables are in a way underestimated.

We estimate the model using the Heckman procedure explained above. For the outcome equation, we estimate the following Probit model that has the same set of explanatory variables.

\Pr (o u t c o m e - c w_{i} = 1 | X) = β_{0} + {β^{'}}_{1} X_{i} + {β^{'}}_{2} Z_{i} + {β^{'}}_{3} W_{i} + ε_{i} .

(2)

To comprehensively examine the relationship between adult education and labor market, we investigate the determinants of participation to adult education for finding work.

For the not working subsample, the dependent variable is attend-fw. Finally, we try to see who is more likely to find a job as a result of this education. In this case, we estimate Equation (2) for the dependent variable outcome-fw. The dependent variables used in the estimations and their explanations are presented in Table 1 (Panel B).

Data Overview and Descriptive Statistics

In this section, we present a general overview of the data to understand the adult education decisions made in 2012 in Turkey before presenting and discussing the results of our econometric analysis.

Among 34,558 individuals surveyed who were older than 25 years, 4,486 chose to attend adult education programs, which correspond to roughly 13% of the sample. Among 4,486 individuals who attended adult education programs in Turkey in 2012, 59% attended for work-related reasons and the rest for other reasons.

Next, we examine the composition of participation in and outcome of adult education in the sample of attending individuals as described in Figure 1. Among 4,486 individuals who attended adult education programs in 2012 in Turkey, 51% were working and the rest were not. This balance is interesting considering that the sampling method does not involve any control for employment.

Figure 1.

Composition of participation in and outcome of adult education in the sample of attending individuals.

Figure 1 shows that among 2,285 working individuals, an overwhelming 72% attended for changing work and the rest attended for other reasons. Among 1,640 working individuals who attended adult education programs for the purpose of changing work, 30% were successful in this endeavor and reported that they were able to change their work.

Turning our attention to not working individuals in the sample of attending individuals, we observe that 45% attended to find work and the rest for other reasons. Among 984 not working individuals who attended adult education programs for the purpose of finding work, 27% were successful in this endeavor and reported that they were able to find a job.

Next, we focus on the gender of the participants using different cuts of the data in Figure 2. While female participants account for 54% in the attending sample, their share declines to 35% in the working sample and goes up to 74% in the not working sample, which is no surprise considering the very low participation of women in labor force in Turkey.

Figure 2.

Gender of participants.

Finally, in Figure 3, we present a broad-brush view of the levels of formal education completed by adult education participants using the same cuts of the data as in Figure 2. These levels correspond to education1 to education5 variables in the data description. The observable difference is between the working and not working individuals among attending individuals: While the share of individuals with no education is only about 2% in the working sample, it goes up to 9% in the not working sample. The same kind of rise is true for primary graduates as well as between these two samples. The difference between the shares of middle school graduates among the samples is negligible. However, we observe a significant decline (from 45% to 17%) in the share of university graduates and a noteworthy increase (from 25% to 34%) in the share of high school graduates when we shift from working to not working sample.

Figure 3.

Education level of participants: Panel A—attending sample, Panel B—working individuals in the attending sample, and Panel C—not working individuals in the attending sample.

Estimation Results

Participation in General

Table 2 reports the results of the probit regression and corresponding marginal effects of the sociodemographic determinants of participation in adult education in Turkey in 2012 for the entire sample. The results show that individual and family characteristics are strongly associated with participation in adult education as indicated by the significant Wald χ².

Table 2.

Participation in Adult Education Entire Sample.

Probit estimation (attend = 1)
Variables	Probit	Marginal effects
female	0.058***	0.010***
	(0.021)	(0.004)
age	0.011**	0.002**
	(0.005)	(0.001)
age ²	−0.032***	−0.006***
	(0.006)	(0.001)
married	−0.036	−0.007
	(0.031)	(0.006)
education2	0.262***	0.049***
	(0.043)	(0.008)
education3	0.367***	0.077***
	(0.048)	(0.011)
education4	0.732***	0.172***
	(0.046)	(0.013)
education5	1.107***	0.305***
	(0.051)	(0.018)
attend-formal	0.405***	0.089***
	(0.038)	(0.010)
father-edu	0.067**	0.013*
	(0.033)	(0.006)
hhsize	−0.014*	−0.002*
	(0.008)	(0.001)
hhchild	−0.129***	−0.023***
	(0.025)	(0.005)
hhcons	0.015**	0.003**
	(0.007)	(0.001)
constant	−1.477***
	(0.116)
Observations	34,558	34,558
Wald χ²

Note. The variables female, age, age², and married represent the general characteristics of the individuals. Reference category in education variable is education1, which is for the individuals who do not have any degree. If the individual has still continued to participate in a formal education program in the past 12 months, the dummy variable attend_formal takes the value 1, and it takes the value 0 otherwise. father-edu is the dummy variable that takes the value 1 if the father of the individual has completed ⩾8 years of education and takes the value 0 otherwise. The number of individuals in the household is represented by hhsize. The number of children in the household who are ⩽4 years of age is represented by hhchild, and household consumption is represented by hhcons.

p < .1. **p < .05. ***p < .01.

On average, females are more likely to attend adult education programs compared with males in Turkey in 2012. This result is in line with the results found in Denton et al. (1990) and Dincer et al. (2016) both of which claim that females have a higher likelihood of taking personal-interest courses compared with males. Labor market concerns might challenge this result as shown in the succeeding sections.

Age has a nonlinear effect on the adult education participation likelihood. Older individuals are more likely to attend. The direct positive effect might be related to attending adult education programs for hobby or personal development purposes particularly for retired individuals. However, the positive effect of Age tapers off at older age levels. There may be labor market related reasons. Individuals tend to invest in their education if it will grant them some benefits (Allingham, 2002). Older people approaching the retirement age tend to have weaker prospects and chances on the labor market (Doets, Hake, & Westerhuis, 2001). Therefore, low expected return on education in older age acts as a barrier.

Formal education appears to be very important for participation in adult education in the estimation results presented in Table 2 by statistically significant and positive coefficients of education2 (primary school education) through education5 (tertiary education).⁹ Individuals endowed with successively higher formal education levels are more likely to attend adult education programs as supported by Anderson, Darkenwald, and Future Directions for a Learning Society (1979) who find education as the most powerful predictor of participation in adult education. For example, an individual with tertiary education (education5) is 30.5% more likely to participate in adult education compared with an individual with no formal education, whereas an individual with a primary school degree (education2) is only 5% more likely to attend.

Family background (father-edu) positively and statistically significantly affects the odds for participation in adult education.

Marital status (married) has no impact on the adult education participation, having young child(ren) in the household (hhchild) hinders participation probably due to care responsibilities in the household.

Household consumption (hhcons), which is used as a proxy for household income, has a statistically significant and positive impact on the likelihood of adult education. An individual living in a household with higher living standards is more likely to attend adult education programs. Kim, Hagerdorn, Williamson, and Chapman (2004) also found that adults with the lowest household income (<$20,000) have lower rates of adult education participation compared with adults with higher incomes (>$50,001) for the United States. Indeed, there is a two-sided relationship between income/consumption level and educational attainment.

Work-Related Participation

In this section, we discuss the regression results for work-related participation in adult education programs among the already attending individuals.

The results show that females are less likely to participate in adult education programs for work-related reasons. The difference between this result and the full-sample result reported in Table 2 may be rooted in an argument put forward by Boeren (2011). First, females may attach more personal value to learning than males because of their focus on inter- and intrapersonal aspects. Second, the family-related responsibilities may prevent females from attending work-related education programs because they might have self-selected into not working in the first place.

Age has a profound negative effect on adult education participation for work-related reasons in the sample of people who already attend these programs.¹⁰ In other words, younger individuals are more likely to participate because of higher expected returns to education for these individuals. This result is different from findings of Rubenson and Desjardins (2009) but similar to the findings of Boeren et al. (2010). The former study suggests that since family- and work-related barriers to participation in adult education are eliminated at older ages, adult education participation is higher at older ages. According to the latter study, as people get older their expected return from education decreases.

As in the full-sample results, individuals with more schooling are more likely to attend work-related adult education programs, which is in line with Jenkins, Vignoles, Wolf, and Galindo-Rueda (2003). In other words, formal education again is very important; however, this time the marginal effects are stronger.

When we look at the rest of the results in Table 3, it is obvious that other individual and family characteristics are not associated with participation in work-related adult education.

Table 3.

Participation in Adult Education for Work Attending Sample (attend = 1).

Probit estimation (attend-w = 1)
Variables	Probit	Marginal effects
female	−0.671***	−0.250***
	(0.046)	(0.016)
age	−0.022***	−0.008***
	(0.003)	(0.001)
married	0.045	0.017
	(0.064)	(0.024)
education2	0.356***	0.128***
	(0.132)	(0.045)
education3	0.483***	0.167***
	(0.139)	(0.043)
education4	0.628***	0.222***
	(0.135)	(0.044)
education5	1.076***	0.357***
	(0.136)	(0.038)
hhsize	−0.018	−0.007
	(0.018)	(0.007)
attend-formal	−0.014	−0.005
	(0.067)	(0.025)
father-edu	−0.049	−0.018
	(0.061)	(0.023)
hhchild	−0.059	−0.022
	(0.054)	(0.020)
hhcons	−0.009	−0.004
	(0.012)	(0.005)
constant	0.822***
	(0.191)
Observations	4,486	4,486
Wald χ²

Note. The variables female, age, and married represent the general characteristics of the individuals. Reference category in education variable is education1, which is for the individuals who do not have any degree. If the individual has still continued to participate in a formal education program in the past 12 months, the dummy variable attend_formal takes the value 1, and it takes the value 0 otherwise. father-edu is the dummy variable that takes the value 1 if the father of the individual has completed ⩾8 years of education and takes the value 0 otherwise. The number of individuals in the household is represented by hhsize. The number of children in the household who are ⩽4 years of age is represented by hhchild, and household consumption is represented by hhcons.

p < .1. **p < .05. ***p < .01.

Therefore, we can argue that gender, age, and education appear to be the most important determinants of participation in work-related adult education programs. Moreover, household consumption, thereby household income, has a meaningful positive effect on the participation in adult education for general purposes; however, this effect tapers off when work-related participation is considered in isolation. This may be interpreted as standard of living having no impact on work-related participation.

Participation for and Outcome of Education for Changing Work

Participation

In this subsection, we examine the decision of participation in adult education for changing work among working individuals in the attending sample.

We consider a two-step process in this case: First, selection into working among attending individuals (working-attend = 1), and second, conditional on working, the determinants of attending to adult education programs for the purpose of changing work (attend-cw = 1).

The first two columns of Table 4 report the probit results and the corresponding marginal effects, and the third column presents the results for the selection equation.

Table 4.

Participation in Adult Education for Changing Work Attending Sample (attend = 1).

Heckman selection estimation
	Probit (attend-cw = 1)		Selection (working-attend = 1)
Variables	Probit	Marginal effects	Probit
female	−0.198**	−0.064**	−1.109***
	(0.089)	(0.032)	(0.051)
age	−0.040**	−0.013**	0.150***
	(0.020)	(0.006)	(0.021)
age ²	0.030	0.010	−0.176***
	(0.024)	(0.008)	(0.027)
married	−0.020	−0.006	0.180**
	(0.085)	(0.028)	(0.071)
education2	0.319	0.098	0.067
	(0.237)	(0.069)	(0.132)
education3	0.472*	0.136**	0.469***
	(0.248)	(0.065)	(0.143)
education4	0.465*	0.142*	0.522***
	(0.245)	(0.073)	(0.137)
education5	0.567**	0.172**	0.951***
	(0.256)	(0.078)	(0.132)
father-edu	0.019	0.006
	(0.077)	(0.025)
hhsize	0.005	0.002	−0.188***
	(0.023)	(0.007)	(0.021)
hhchild	−0.074	−0.024	0.284***
	(0.068)	(0.022)	(0.055)
hhlabor		0.000	0.577***
		(0.000)	(0.041)
finance-self	−0.293***	−0.096***
	(0.078)	(0.027)
finance-gov	−0.277***	−0.093***
	(0.077)	(0.027)
constant	1.519***		−2.899***
	(0.505)		(0.397)

Observations	1,910	1,910	4,056

Note. The variables female, age, age², and married represent the general characteristics of the individuals. Reference category in education variable is education1, which is for the individuals who do not have any degree. father-edu is the dummy variable that takes the value 1 if the father of the individual has completed ⩾8 years of education and takes the value 0 otherwise. The number of individuals in the household is represented by hhsize. The number of children in the household who are ⩽4 years of age is represented by hhchild and the number of working individuals in the household is represented by hhlabor. The variables finance-self, finance-gov, finance-employer represent the sponsor of adult education activities. The first of these, finance-self, takes the value 1 if the participant pays all the costs of the adult education activities by himself or herself and takes the value 0 otherwise. The other two variables represent sponsorship of the government and the employer, respectively.

p < .1. **p < .05. ***p < .01.

First, we will start discussing the results of the selection equation. Results related to the characteristics of the individual in Table 4 show that being a male increases the likelihood of working as indicated by the negative sign of Female in the selection equation. Age has a nonlinear positive effect on working. Older individuals are more likely to work; however, this effect dampens as individuals get older. The level of education increases the odds for working. In fact, education3 (middle school education), education4 (high school education), and education5 (tertiary education) have increasingly more positive coefficients, while education2 (primary education) is insignificant. This can be interpreted as having a primary school degree is irrelevant for likelihood of working, while education levels at middle school or above raise the odds for working at successively increasing rates. The results related to the characteristics of the individual’s household show that although hhchild increases the odds for working, hhsize reduces it.

Second, we examine the results of the probit equation in Table 4 where the determinants of participating in adult education programs with the purpose of changing their work are reported. Among the characteristics of the individual, only female, age, and education5 (tertiary education) are statistically significant at 5% level. Males and younger individuals with a university degree or above are more likely to attend adult education programs to change work. Other education variables are either insignificant or barely significant. This may be due to the perception that attending adult education programs will not change much in terms of changing work for the individuals with less than a university diploma.

Two new variables related to the financing of the adult education are introduced in the probit estimations: finance-self and finance-gov. The excluded category is finance-employer. The results show that compared with both self-financed and government-financed education, employer-financed education (the reference category) increases the odds for participation in adult education programs among working individuals to change work.

Outcome

Next, we investigate the outcome of participating in adult education for changing work. Following the methodology in the last subsection, we consider a two-step decision process in this case: First, selection into working among attending individuals (working-attend = 1), and second, conditional on working, the determinants of outcome of attending to adult education programs for the purpose of changing work (outcome-cw = 1).

The probit results and the corresponding marginal effects are reported in the first two columns of Table 5, and the third column presents the results for the selection equation.

Table 5.

Outcome of Adult Education for Changing Work Attending Sample (attend = 1).

Heckman selection estimation
Variables	Probit (outcome-cw = 1)		Selection (working-attend = 1)
	Probit	Marginal effects	Probit
female	−0.285***	−0.085***	−1.109***
	(0.092)	(0.023)	(0.051)
age	−0.049**	−0.015**	0.150***
	(0.022)	(0.007)	(0.021)
age ²	0.038	0.011	−0.176***
	(0.027)	(0.009)	(0.027)
married	−0.096	−0.029	0.184***
	(0.088)	(0.027)	(0.071)
education2	0.017	0.005	0.068
	(0.263)	(0.079)	(0.132)
education3	−0.108	−0.031	0.465***
	(0.271)	(0.076)	(0.143)
education4	0.051	0.015	0.523***
	(0.263)	(0.080)	(0.137)
education5	0.086	0.026	0.951***
	(0.268)	(0.081)	(0.132)
father_educ	0.098	0.030
	(0.074)	(0.023)
hhsize	−0.014	−0.004	−0.187***
	(0.023)	(0.007)	(0.021)
hhchild	0.001	0.000	0.283***
	(0.069)	(0.021)	(0.056)
hhlabor		0.000	0.576***
		(0.000)	(0.041)
finance-self	−0.126	−0.037
	(0.080)	(0.024)
finance-gov	−0.003	−0.001
	(0.076)	(0.023)
constant	0.705		−2.908***
	(0.525)		(0.396)
Observations	1,910	1,910	4,056

Note. The variables female, age, age², and married represent the general characteristics of the individuals. Reference category in education variable is education1, which is for the individuals who do not have any degree. father-edu is the dummy variable that takes the value 1 if the father of the individual has completed ⩾8 years of education and takes the value 0 otherwise. The number of individuals in the household is represented by hhsize. The number of children in the household who are ⩽4 years of age is represented by hhchild and the number of working individuals in the household is represented by hhlabor. The variables finance-self, finance-gov, and finance-employer represent the sponsor of adult education activities. The first of these, finance-self, takes the value 1 if the participant pays all the costs of the adult education activities by himself or herself and takes the value 0 otherwise. The other two variables represent sponsorship of the government and the employer, respectively.

p < .1. **p < .05. ***p < .01.

The selection equation and its estimation results are the same as the one in the previous subsection. We will next discuss the results of the probit estimation in Table 5 where the determinants of outcome of participating in adult education programs with the purpose of changing work are presented.

Males and younger individuals are more likely to change work after attending adult education programs. However, what is more interesting is that none of the other variables have an effect on the outcome of adult education for changing work.

Two groups of potential determinants in the outcome regression are noteworthy: education level and financing of education. The insignificance of education2 (primary school education), education3 (middle school education), education4 (high school education), and education5 (tertiary education) or finance-self and finance-gov show that being successful in changing work after getting adult education has nothing to do with the level of education or who finances it. Put another way, the odds for participating in adult education for changing work is affected by the level of formal education and type of financing but the odds for success in that endeavor is not. This might mean that once an individual participates in adult education for changing work, independent of his or her education or the finance type, he or she can be successful in doing so.

Participation for and Outcome of Education for Finding Work

Participation

In this subsection, we examine the decision of participation in adult education for finding work among not working individuals in the attending sample.

Table 6 reports the results of the probit regression and corresponding marginal effects of the determinants of participation in adult education for finding work among not working. Among the characteristics of the individual, only female, age, education4 (high school education), and education5 (tertiary education) are statistically significant at 5% level. Males and younger individuals with a high school degree or above are more likely to attend adult education programs to find work. The other education variables are insignificant.

Table 6.

Participation in Adult Education for Finding Work Attending and Not Working Sample (working-attend = 0).

Probit estimation (attend-fw = 1)
Variables	Probit	Marginal effects
female	−0.550***	−0.216***
	(0.089)	(0.034)
age	−0.028***	−0.011***
	(0.004)	(0.002)
married	−0.076	−0.030
	(0.098)	(0.039)
education2	0.162	0.065
	(0.172)	(0.068)
education3	0.046	0.018
	(0.183)	(0.073)
education4	0.406**	0.161**
	(0.178)	(0.069)
education5	1.055***	0.382***
	(0.182)	(0.054)
father-edu	−0.192*	−0.076*
	(0.099)	(0.039)
hhsize	−0.022	−0.009
	(0.026)	(0.010)
hhchild	−0.006	−0.003
	(0.087)	(0.035)
finance-self	0.240***	0.095***
	(0.083)	(0.033)
constant	0.881***
	(0.280)
Observations	1,844	1,844
Wald χ²

Note. The variables female, age, and married represent the general characteristics of the individuals. Reference category in education variable is education1, which is for the individuals who do not have any degree. father-edu is the dummy variable that takes the value 1 if the father of the individual has completed ⩾8 years of education and takes the value 0 otherwise. The number of individuals in the household is represented by hhsize. The number of children in the household who are ⩽4 years of age is represented by hhchild, the number of working individuals in the household is represented by hhlabor. The variables finance-self and finance-gov represent the sponsor of adult education activities. The former, finance-self, takes the value 1 if the participant pays all the costs of the adult education activities by himself or herself and takes the value 0 otherwise. The latter variable represents sponsorship of the government. The reference category is finance-gov.

p < .1. **p < .05. ***p < .01.

To control for the financing of adult education, finance-gov is used in the probit regression in Table 6. The excluded category is finance-self this time because the individual is not working. The results show that compared with self-financed education, government-financed education increases the odds for participation in adult education programs among working individuals to find work.

Outcome

Next, we investigate the outcome of participating in adult education for finding work. The probit results and the corresponding marginal effects are reported in Table 7.

Table 7.

Outcome of Adult Education for Finding Work Attending and Not Working Sample (working-attend = 0).

Probit estimation (outcome-fw = 1)
Variables	Probit	Marginal effects
female	−0.217**	−0.043*
	(0.104)	(0.022)
age	0.116***	0.022***
	(0.035)	(0.006)
age ²	−0.184***	−0.035***
	(0.047)	(0.008)
married	−0.267*	−0.050*
	(0.146)	(0.027)
education2	0.078	0.015
	(0.284)	(0.056)
education3	0.275	0.058
	(0.300)	(0.070)
education4	0.301	0.059
	(0.288)	(0.059)
education5	0.630**	0.149*
	(0.291)	(0.083)
father-edu	−0.045	−0.008
	(0.109)	(0.020)
hhsize	−0.047	−0.009
	(0.029)	(0.006)
hhchild	0.129	0.024
	(0.115)	(0.022)
finance-self	0.310***	0.054***
	(0.114)	(0.019)
constant	−2.781***
	(0.649)
Observations	1,844	1,844
Wald χ²

Note. The variables female, age, age², and married represent the general characteristics of the individuals. Reference category in education variable is education1, which is for the individuals who do not have any degree. father-edu is the dummy variable that takes the value 1 if the father of the individual has completed ⩾8 years of education and takes the value 0 otherwise. The number of individuals in the household is represented by hhsize. The number of children in the household who are ⩽4 years of age is represented by hhchild, and the number of working individuals in the household is represented by hhlabor. The variables finance-self and finance-gov represent the sponsor of adult education activities. The former, finance-self, takes the value 1 if the participant pays all the costs of the adult education activities by himself or herself and takes the value 0 otherwise. The latter variable represents sponsorship of the government. The reference category is finance-gov.

p < .1. **p < .05. ***p < .01.

Similar to Schwerdt et al. (2012), being male increases the odds for success in finding work after attending as can be understood from the negative sign in front of female in Table 7. Females attend less as reported in Table 6 and even if they attend, compared with males, they benefit less from this type of adult education programs. Different from the results of determinants of participation for finding work presented in Table 6, older individuals are more likely to find work after attending adult education programs. However, this effect dampens as the individuals get older.

It is observed from Table 7 that only one of the formal education variables (education5—tertiary education) is statistically significant. Level of formal education does not matter much for the odds for success in finding work among not working individuals. In other words, once an unemployed individual participates in adult education for finding a job, his or her education neither reduces nor increases his or her success probability in this endeavor. Therefore, less educated are not necessarily at a disadvantaged position under the circumstances.

The result related to the type of financing is also important as opposed to the results related to outcome of adult education for changing work. Government-financed programs (finance-gov) offered for not working individuals increase the likelihood of finding work after attending these adult education programs. Therefore, government-financed programs increase both the likelihood to attend adult education programs and to find work as a result of adult education via endowing the participants with new skills.

Conclusion

In this article, we analyze the role of adult education as a tool in facilitating transitions to/in the labor market, employing the AES of Turkey in 2012.

The results of our analysis indicate that the determinants of participation in adult education programs without a particular emphasis on labor market concerns show a great degree of similarity to the determinants in the broad literature of adult education (Knowles et al., 2014). Factors such as being female, being older and more educated, having educated parents, and coming from a household with a high-income level and fewer young children increase the odds in favor of attending adult education programs.

Work-related education programs are investigated in two subcategories. First, when adult education programs are attended for the purpose of finding a better job among the already employed adult population, our results show that males and younger individuals are more likely to change work after attending adult education programs. Moreover, the odds for participating in adult education for changing work is affected by the level of formal education and type of financing but the odds for success in that endeavor is not. This might mean that once an individual participates in adult education for changing work, independent of his or her education or the finance type, he or she can be successful in doing so. Therefore, even for people with lower education levels and financing constraints, participation in adult education can be a stepping-stone toward better jobs.

Second, when adult education programs are attended by unemployed individuals for the purpose of finding jobs, our results show that being male increases the odds for success in finding work. Different from the results of determinants of participation for finding work, older individuals are more likely to find work after attending adult education programs. More important, once an unemployed individual participates in adult education for finding a job, his or her education neither reduces nor increases his or her success probability in this endeavor. Therefore, less educated are not necessarily at a disadvantaged position under the circumstances.

As a result, to give a chance to unemployed people to find work, adult education seems to be a useful tool, especially for less educated people. The finding that government-financed programs offered for unemployed individuals increase the likelihood of finding work after attending indicates the importance of adult education programs offered by the government in increasing the chance of unemployed people to be employed and the chance of increasing their income level.

The results of our study have important policy implications. In Turkey, lifelong learning strategy including adult education was prepared in coordination with the Ministry of National Education in 2009 and was approved by the High Planning Council. One of the priorities of this strategy was determined as “the quality of the workforce to reach international competitive levels.” Currently, within the scope of adult education for the labor market, activities such as courses, internship programs are carried out to raise employability of individuals. İŞKUR, which is one of the most important actors in this field, opened 27,351 courses in 2012 for different occupational groups, and a total of 464,645 people have participated in these courses (İŞKUR, 2012). Both in the Republic of Turkey Ministry of Labor and Social Security (2013) and the Republic of Turkey Ministry of Development (2013), the need of extension and update of adult education programs is highlighted.

Since technological and occupational change require labor force that is adaptable and has the flexibility to gain new skills, these programs should be updated in a way to enable individuals to attain the basic skills required by labor market. For this purpose, the regular surveys conducted to collect information on the needs of skills in the labor market demand, such as the one conducted by İŞKUR, should be improved and continued. Besides, awareness about these programs should be increased to reach particularly the less educated people. Because as indicated by Roosmaa and Saar (2011), nonaccessibility to education-related information is one of the barriers to attend these programs.

Education is the most important input of productivity and sustainable development for a developing country. In Turkey, the discussions are always carried on regarding formal education. Although formal education is the first place to which school-aged kids/youth be referred, there are plenty of less educated people in the labor market for whom it is very difficult to reengage in a formal school setup. Considering the current labor force that needs more training as well as the increasing numbers of uneducated migrants in Turkey, adult education would definitely be a strong tool for development that would be adopted by the policymakers. For the younger generation, which is more educated than the older ones, adult education can be employed as a complementary tool for finding better jobs and/or adapting the changes in the labor market.

Footnotes

Appendix

Table A1.

Descriptive Statistics.

	No. of observations
Variable	All	If var = 0	If var = 1	Min	Max	Mean	Standard deviation
female	34,558	15,012	19,546	0	1	0.566	0.496
age	34,558			18	99	44.82	16.71
age ²	34,558			3.2	98	22.88	16.44
married	34,558	8,695	25,863	0	1	0.748	0.434
education1	34,558	27,494	7,064	0	1	0.204	0.403
education2	34,558	21,185	13,373	0	1	0.387	0.487
education3	34,558	30,030	4,528	0	1	0.131	0.337
education4	34,558	28,913	5,645	0	1	0.163	0.370
education5	34,558	30,610	3,948	0	1	0.114	0.318
attend-formal	34,558	32,209	2,349	0	1	0.068	0.252
father-edu	34,558	31,508	3,050	0	1	0.088	0.284
hhsize	34,558			1	20	3.812	1.988
hhchild	34,558			0	6	0.255	0.555
hhlabor	34,558			0	9	1.158	0.987
hhcons	34,558			0.1	23.5	2.096	1.608
attend	34,558	30,072	4,486	0	1	0.13	0.336
attend-w	4,486	1,849	2,637	0	1	0.588	0.492
working-attend	4,486	2,201	2,285	0	1	0.509	0.499
attend-cw	2,285	645	1,640	0	1	0.718	0.450
outcome_cw	2,285	1,737	548	0	1	0.24	0.427
attend-fw	2,201	1,218	983	0	1	0.447	0.497
outcome_fw	2,201	1,926	275	0	1	0.125	0.331
finance-employer	4,486	3,823	663	0	1	0.148	0.355
finance-self	4,486	2,476	2,010	0	1	0.448	0.497
finance-gov	4,486	3,103	1,383	0	1	0.308	0.462

Authors’ Note

The views expressed in this article are those of the authors and should not be attributed to the World Bank.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Seyit Mümin Cilasun received his PhD in economics from the Middle East Technical University in 2009. He worked as a faculty member at Atılım University between 2002-2017. In 2017, he worked as an advisor at the Structural Economic Research Department of the Central Bank of the Republic of Turkey. He is currently an economist at the same department. Associate Professor Cilasun’s research interests are applied microeconometrics, labor economics, and industrial economics.

Sırma Demir-Şeker is an economist in the World Bank at Social Protection and Jobs Global Practice. Prior to joining the Bank, she worked from 2004 to 2016 at the Ministry of Development of the Republic of Turkey. She holds a PhD in economics from the Middle East Technical University, Ankara, Turkey. Her main areas of interest include labor economics, social protection, income distribution, and poverty.

N. Nergiz Dincer received her PhD in economics at Bilkent University in 2005; she continued her postdoctoral studies at the University of California, Berkeley for two years (2005-2006 and 2010-2011) as a visiting scholar. She has been working at TED University since 2012. Currently, she is a professor at the Department of Economics at TED University and director of the TEDU Trade Research Center. Prof. Dincer’s research interests are central banking, international economics, economic growth, and exchange rates.

Ayça Tekin-Koru received her PhD in economics from Purdue University in 2001. She worked as a visiting faculty at Krannert School of Management between 2001-2003 and as a faculty member in the Department of Economics at Oregon State University from 2003-2011. Ayça Tekin-Koru is currently a faculty member of the Economics Department at TED University and the associate director of TED University Trade Research Center. Prof. Tekin-Koru’s research interests are international trade and foreign direct investment.

References

Allingham

(2002). Choice theory: A very short introduction. New York, NY: Oxford University Press.

Almeida

Behrman

Robalino

(2012). The right skills for the job? Rethinking training policies for workers (Human Development Perspectives). Washington, DC: World Bank.

Anderson

R. E.

Darkenwald

G. G.

, Future Directions for a Learning Society. (1979). Participation and persistence in American adult education: Implications for public policy and future research from a multivariate analysis of a national data base. New York, NY: College Entrance Examination Board.

Antikainen

(2005). Transforming a learning society: The case of Finland. Bern, Switzerland: Peter Lang.

Attanasio

Kugler

Meghir

(2011). Subsidizing vocational training for disadvantaged youth in Colombia: Evidence from a randomized trial. American Economic Journal: Applied Economics, 3, 188-220.

Badescu

Garrouste

Loi

(2011). The distribution of adult training in European countries: Evidence from recent surveys (JRC Scientific and Technical Report EUR24895EN-2011). Brussels, Belgium: European Commission.

Becker

G. S.

(1975). Human capital: A theoretical and empirical analysis, with special reference to education (2nd ed.). New York, NY: National Bureau of Economic Research.

Becker

G. S.

(1994). Human capital revisited. In Human capital: A theoretical and empirical analysis with special reference to education (3rd ed., pp. 15-28). Chicago, IL: University of Chicago Press.

Betcherman

Olivas

Dar

(2004). Impacts of active labor market programs: New evidence from evaluations with particular attention to developing and transition countries (World Bank Social Protection Discussion Paper Series 0402). Washington, DC: World Bank.

10.

Bilir

(2009). Yetişkin eğitiminin tarihsel gelişimi [Historical development of adult education]. Istanbul, Turkey: Kalkedon Yayıncılık.

11.

Blanden

Buscha

Sturgis

Urwin

(2012). Measuring the earnings returns to lifelong learning in the UK. Economics of Education Review, 31, 501-514.

12.

Boateng

S. K.

(2009). Significant country differences in adult learning. Eurostat. Statistics in Focus: Populations and Social Conditions, 44, 1-12.

13.

Boeren

(2011). Participation in adult education: A bounded agency approach (Doctoral dissertation, KULeuven, Centre for Research on Professional Learning & Development, Corporate Training and Lifelong Learning, Leuven, Belgium). Retrieved from https://www.researchgate.net/publication/305115262_Participation_in_adult_education_a_bounded_agency_approach

14.

Boeren

(2014). Evidence-based policy-making: The usability of the Eurostat Adult Education Survey. International Journal of Lifelong Education, 33, 275-289.

15.

Boeren

Nicaise

Baert

(2010). Theoretical models of participation in adult education: The need for an integrated model. International Journal of Lifelong Education, 29, 45-61.

16.

Boyadjieva

Ilieva-Trichkova

(2017). Between inclusion and fairness: Social justice perspective to participation in adult education. Adult Education Quarterly, 67, 97-117.

17.

Brown

Lauder

Ashton

(2011). The global auction: The broken promises of education, jobs, and incomes. Oxford, England: Oxford University Press.

18.

Card

Kluve

Weber

(2010). Active labor market policy evaluations: A meta-analysis. Economic Journal, 120(548), F452-F477.

19.

Cincinnato

De Wever

Valcke

(2014). The learning divide in formal adult education: Why do low-qualified adults participate less? In Local change, social actions and adult learning: Challenges and responses (pp. 401-415). Lisbon, Portugal: University of Lisbon.

20.

Dayioglu

Demir-Seker

(2016). Social policy and the dynamics of early childhood poverty in Turkey. Journal of Human Development and Capabilities, 17, 540-557.

21.

De Witte

Nicaise

Lavrijsen

Van Landeghem

Lamote

Van Damme

. (2013). The impact of institutional context, education and labour market policies on early school leaving: A comparative analysis of EU countries. European Journal of Education, 48, 331-345.

22.

Deaton

(1997). The analysis of household surveys: A microeconometric approach to development policy. Baltimore, MD: Johns Hopkins University Press.

23.

Denton

F. T.

Pineo

P. C.

Spencer

B. G.

(1990). The constituencies of adult education programs: Similarities and differences among age groups and other components of the population. Canadian Journal of Education/Revue canadienne de l’education, 15(1), 72-90.

24.

Desjardins

Milana

Rubenson

(2006). Unequal chances to participate in adult learning: International perspectives. Paris, France: United Nations Educational, Scientific and Cultural Organization.

25.

Desjardins

Rubenson

(2013). Participation patterns in adult education: The role of institutions and public policy frameworks in resolving coordination problems. European Journal of Education, 48, 262-280.

26.

Dincer

N. N.

Tekin-Koru

Askar

(2016). Investigation of participation in adult education in Turkey: AES data analysis. Comparative Education Review, 60, 530-548.

27.

Doets

Hake

Westerhuis

(2001). Lifelong learning in the Netherlands. The state of the art in 2000’s. Hertogenbosch, Netherlands: CINOP.

28.

Domińczak

Lis

(2013). Does gender matter for lifelong learning activity? (Neujobs Working Paper No. D16.2A). Warsaw, Poland: Institute for Structural Research.

29.

Elbers

Lanjouw

J. O.

Lanjouw

(2003). Micro-level estimation of poverty and inequality. Econometrica, 71, 355-364.

30.

EuroStat. (2006). Classification of learning activities: Manual. Luxembourg: Office for Official Publications of the European Communities.

31.

Heckman

J. J.

(1979). Statistical models for discrete panel data. Chicago, IL: University of Chicago, Department of Economics and Graduate School of Business.

32.

İŞKUR. (2012). Statistical yearbook, 2012. Ankara, Turkey: Author.

33.

Jenkins

Vignoles

Wolf

Galindo-Rueda

(2003). The determinants and effects of lifelong learning. Applied Economics, 35, 1711-1721.

34.

Kilpi-Jakonen

De Vilhena

D. V.

Kosyakova

Stenberg

Blossfeld

(2012). The impact of formal adult education on the likelihood of being employed: A comparative overview. Studies of Transition States and Societies, 4(1), 48-68.

35.

Kim

Hagedorn

Williamson

Chapman

(2004). Participation in adult education and lifelong learning: 2000-01. National Household Education Surveys of 2001 (NCES 2004-050). Washington, DC: U.S. Department of Education.

36.

Kleinert

Matthes

(2009). Data in the field of adult education and lifelong learning: Present situation, improvements and challenges (RatSWD Working Paper No. 91). Berlin, Germany: RatSWD.

37.

Kluve

(2016). A review of the effectiveness of active labour market programmes with a focus on Latin America and the Caribbean (Research Department Working paper No. 9). Geneva, Switzerland: International Labour Office.

38.

Knipprath

De Rick

(2015). How social and human capital predict participation in lifelong learning: A longitudinal data analysis. Adult Education Quarterly, 65, 50-66.

39.

Knowles

M. S.

Holton

E. F.

III Swanson

R. A.

(2014). The adult learner: The definitive classic in adult education and human resource development. New York, NY: Routledge.

40.

Lopes

Fernandes

M.G.

(2011). Formal, non-formal and informal learning and higher education graduates’ reemployment: Evidence for Portugal (MPRA Paper No. 34445). Munich, Germany: University Library of Munich.

41.

Merriam

S. B.

(2008). Adult learning theory for the twenty-first century. San Francisco, CA: Jossey-Bass.

42.

Nilsson

Nyström

(2013). Adult learning, education, and the labour market in the employability regime. European Journal for Research on the Education and Learning of Adults, 4, 171-187.

43.

OECD. (2016). Turkey–country note–skills matter: Further results from the survey of adult skills. Paris, France: Author.

44.

OECD. (2017). PISA 2015 results (Volume V): Collaborative problem solving. Paris, France: Author.

45.

Ravallion

(1992). Does undernutrition respond to incomes and prices? Dominance tests for Indonesia. World Bank Economic Review, 6, 109-124.

46.

Republic of Turkey Ministry of Development. (2013). Tenth development plan (2014-2018). Ankara, Turkey: Ministry of Development of Turkey.

47.

Republic of Turkey Ministry of Labor and Social Security. (2013). National employment strategy (2014-2023). Ankara, Turkey: Author.

48.

Republic of Turkey Ministry of National Education. (2015). 2015-2019 Strategic plan. Ankara, Turkey: Author.

49.

Riddell

Weedon

(2009). Approaches to dispute resolution in additional support needs in Scotland. European Journal of Special Needs Education, 24, 355-369.

50.

Roosmaa

E. L.

Saar

(2011). Barriers to and in participation in lifelong learning: Country report on Estonia. Retrieved from http://lll2010.tlu.ee/folder.2006-02-22.7534376022/subproject-2/sp2-national-reports/sp2_nr_estonia_b-3.pdf

51.

Rubenson

Desjardins

(2009). The impact of welfare state regimes on barriers to participation in adult education: A bounded agency model. Adult Education Quarterly, 59, 187-207.

52.

Saar

Unt

Roosmaa

(2014). Cumulative inequality effects of adult learning in Estonia. In Elgar

Blossfeld

H.-P.

Kilpi Jakonen

Vono

Vilhena

Buchholz

(Eds.), Adult learning in modern societies: An international comparison from a life course perspective (pp. 162-183). Cheltenham, England: Edward Elgar.

53.

Schwerdt

Messer

Woessmann

Wolter

S. C.

(2012). The impact of an adult education voucher program: Evidence from a randomized field experiment. Journal of Public Economics, 96, 569-583.

54.

Tilak

J. B.

(2002). Education and poverty. Journal of Human Development, 3, 191-207.

55.

TurkStat. (2016). Income and Living Conditions Survey: 2015 (Press Release No. 21584). Ankara, Turkey: Turkish Statistical Institute.

56.

Veen

R. V. D.

Preece

(2005). Poverty reduction and adult education: Beyond basic education. International Journal of Lifelong Education, 24, 381-391.

57.

Vignoles

Galindo-Rueda

Feinstein

(2004). The labour market impact of adult education and training: A cohort analysis. Scottish Journal of Political Economy, 51, 266-280.

58.

Wooldridge

J. M.

(2002). Econometric analysis of cross section and panel data. Cambridge: MIT Press.

59.

World Bank. (2013). Turkey: Evaluating the impact of ISKUR’s vocational training program. (Report No. 82306-TR). Washington, DC: Author.