Impact of individual and household characteristics on the employment probability among youth from Bosnia and Herzegovina

Abstract Employment has been identified at the top of the list of young people’s concerns across Europe. Given the fact that in Bosnia and Herzegovina youth is one of the most vulnerable group, mainly due to the high unemployment rate, the main goal of this paper to determine the key individual and household characteristics of young people that influence their employment probability in order to support further development of decision-making policies in the labour market of Bosnia and Herzegovina. By using the USAID MEASURE-BiH National Youth Survey data set we analyse the effects of various individual and household characteristics on the probability of youth employment in Bosnia and Herzegovina. The analysis has revealed that education, age, gender and certain household characteristics have an impact on the probability of youth employment. The paper is expected to produce useful pieces of information that might be helpful for government decision-makers in Bosnia and Herzegovina in the process of creating employment policies to support young people.


Introduction
As well-known, employment has been identified at the top of the list of young people's concerns across Europe. As pointed out by Kluve et al. (2019) bringing young people into productive work is a key labour market challenge in both developing and developed economies. Therefore, to try to solve the problem of youth unemployment in 2009 the Member States adopted the EU youth strategy that acknowledges young people as both a most vulnerable group and a precious resource in an ageing Europe (European Commission, 2012). The strategy has following two main objectives (Commission of the European Communities, 2009): to provide more and equal opportunities for young people in education and the job market and to encourage young people to actively participate in society.
Youth unemployment isn't just a temporary problem in one's life. It may have long term negative effects, such as reduction of opportunity for lifetime earnings, higher likelihood of precarious employment, etc. Also, the negative consequences of youth unemployment affect not only young individuals but the entire society. According to Strasser (1997), increasing youth unemployment is connected with negative impacts on the personal perspectives of life, political opposition and integration problems and, as Isengard (2003) has already pointed out, may lead to social problems like a lack of orientation and hostility towards foreigners, which in turn lead to increased social expenditures. Also, as pointed out by O'Higgins (2009), it is now firmly established that what happens in the youth labour market depends on what occurs in the economy as a whole.
To date, a considerable body of research has sought to understand the probability of youth employment, or, to be more precise to determine the factors that determine the most the likelihood of being employed, young person. Although in Bosnia and Herzegovina, youth is one of the most vulnerable group, mainly due to high youth unemployment rate, there is still no extensive research into the empirical and theoretical aspects of the probability of youth employment in this country, due to which young people are constantly leaving searching for better future abroad.
Furthermore, the effect of individual and household characteristics on the probability of youth employment in Bosnia and Herzegovina has not yet been clarified, so it is not obvious which one is dominant. In order to address the identified research gap, the main goal of this paper is to determine the key individual and household characteristics of young people from Bosnia and Herzegovina that influence their employment probability. Therefore, the starting point of this research study is related to addressing the following question: Do individual and household factors of young people have an impact on their employment?
To support further development of decision-making policies in the labour market of Bosnia and Herzegovina it is important to identify individual and household characteristics that determine youth employment probability. In that respect, this research study is the first comprehensive research dealing with an in-depth analysis of individual and household factors that determine youth employment probability in Bosnia and Herzegovina. Due to many specific features of the labour market of Bosnia and Herzegovina, as well as the fact that the relationship between labour supply and demand should be enhanced by reforming programmes, the results of this research might, therefore, be recognised as useful guidelines for evidence-informed policy decision making, such as tailoring specific and adequate labour market programmes.
The paper is organised as follows. After the introduction, the following section gives a brief outline of hypothesis development. The paper moves on describing methodology, after which follows the discussion of the results. In the end, a summary of the main conclusions is given following the analysis findings.

Hypothesis development
This research builds on existing knowledge relevant to examining the factors that contribute to the probability of youth employment. Many of the recent research studies have focussed primarily on the various individual and household characteristic as key factors that discriminate between being employed and unemployed young person. For example, Karamessini et al. (2019) have examined individual-level factors gender, educational attainment, nationality, age, the respective region's degree of urbanisation and parental education) influencing youth unemployment and inactivity in nine European countries. Similar, D an acic a (2015) analysed the effect of various individual characteristics (gender, age, education) on re(employment) probability for young people in Romania. Ahmad and Azim (2010) have found evidence that age, sex, marital status, migration, training, location, education level and characteristics of the household have a significant impact on employment probabilities of youth in Pakistan. Viitanen (1999) used three sets of factors as the predictors of employment probability, i.e.,: personal characteristics (such as low level of education or no formal qualifications, low language ability, and attendance to large schools with lower academic achievements), family characteristics (such as presence of unemployed parents or unemployed siblings, low family income and single-parent families), and regional characteristics (such as level of unemployment). Ahmad et al. (2015) have concluded that being a female reduces the chances of fulltime work and full-time students in Pakistan. In their research, Kirchner Sala et al. (2015) give an overview of variables explaining young people's school-to-work transitions, such as individual and family characteristics, education, aspirations, expectations and attitudes of young people and parents, family background, household income level and macro-economic circumstances. Gardecki (2001) grouped factors of youth employment probability into four areas: individual characteristics, family determinants, neighbourhood and geographic factors, and spatial mismatch measures. Riddell and Song (2011) have primarily focussed on the causal effects of formal education on re-employment outcomes of unemployed job seekers. Results indicate that education significantly increases re-employment rates of the unemployed. Julkunen (2001) analyzed how different factors, gender, household composition, education, work and unemployment experiences, different coping strategies, work involvement, family support and unemployment insurance, simultaneously predicted the probability of re-employment. Marks and Fleming (1998) examined unemployment among Australian young people between by using social and demographic background factors, the national unemployment rate, school factors including school achievements, postsecondary qualifications and unemployment history. Eide and Showalter (2005) investigated the relationship between high school quality and the probability of extended unemployment among non-collegebound men. Hussain et al. (2016) explored the demographic factors that directly or indirectly influence the labour force participation in Pakistan. The research concluded that the level of education, training, age, location, residential period and being male has a positive and significant impact on labour force participation. Harris (1996) explained the impact of particular personal characteristics on the probability of Australian youth unemployment. Results indicated that age, education and financial commitments exert have a positive impact on employment prospects. Hammer (1999) examined the impact of receiving unemployment benefits upon job chances of the previously unemployed. When controlling for personal characteristics, such as age, education, work commitment, job seeking and duration of employment, the results indicated a decrease in the probability of reemployment for unemployed youth receiving unemployment benefits. Adamchik (1999) also used personal characteristics to analyse the influence of unemployment benefits on the probability of exiting from unemployment to employment.
Based on the previous research, and concerning the research question the following hypotheses have been proposed H 1 : Education is positively related to the probability of youth employment.
H 2 : Demographic characteristics determine the probability of youth employment.
H 3 : Household characteristics may be considered as antecedents of the probability of youth employment.

The limitations of the study
The concept of youth employment is multidimensional. Hence, there may be some possible limitations in this study. The first limitation refers to the omitted variables problem. There are many other variables (i.e., macroeconomic policies, trade and foreign direct investment, technology and innovation, etc.), not only individual and household characteristics, that effect youth employment, particularly in the case of labour markets in the transitional countries such as Bosnia and Herzegovina. However, this limitation does not diminish the importance of investigating the impact of individual and household characteristics on the probability of youth employment. As Isengard (2003) has already pointed out, the individual risk of long-term unemployment is not equally high for all young people, but rather depends on various socio-economic and structural factors like gender, education, nationality and region of residence. Impact of certain individual and household characteristics on the probability of youth employment has already been confirmed in many recent papers that address this issue (D an acic a (2015), Ahmad and Azim (2010) Viitanen (1999); Riddell andSong (2011), Julkunen (2001), Marks and Fleming (1998); Ahmad et al. (2015), Msigwa and Kipesha (2013), Harris (1996), Oancea et al. (2016), etc.). Another limitation of the study, and similar to Rodokanakis and Vlachos (2013), is that the data available are cross-sectional rather than longitudinal and therefore we cannot study any population changes across time.
Taking into account the above-mentioned, the authors decided to select the topic of this paper because, according to the authors' best knowledge, very few publications, dealing with individual and household factors that determine the probability of youth employment in Bosnia and Herzegovina, are found in the existing literature.

Data source and sample
In 2017, the United States Agency for International Development Bosnia and Herzegovina Mission (USAID/BiH) commissioned IMPAQ International (IMPAQ), under the Monitoring and Evaluation Support Activity (MEASURE-BiH), to conduct the National Youth Survey in Bosnia and Herzegovina (NYS-BiH). NYS-BiH provides insights into the state of the youth from Bosnia and Herzegovina, examining their perceptions, attitudes, and experiences on relevant topics including education, employment, inter-ethnic relations, political and civic participation, and migration intentions (Monitoring and Evaluation Support Activity (Measure-BiH), 2018).
According to National Youth Survey in Bosnia and Herzegovina 2018: Findings Report (2018) the NYS-BiH was conducted in January and February of 2018 using a nationally representative sample of 4,500 BiH youth ranging from 15 to 44 years of age. The sample design for NYS-BiH was based on BiH Census 2013 data and was designed to ensure nationwide representation and to be large enough to allow for a variety of empirical analyses. Overall, the sample consists of 900 randomly selected sampling points and 4,500 interviews with respondents ageing from 15 to 44 years of age. The sample was constructed using a multi-stage stratified probability sampling approach. To ensure representative coverage, the sample was stratified by entities (Federation of Bosnia and Herzegovina and Republika Srpska) and District Br cko, ethnic majority areas, and geographic regions. Within each region, the sample was further stratified to include municipalities of all sizes. A sample from each municipality was then stratified to include urban and rural areas proportionally. Table 1 gives a brief overview of the basic characteristics of the entire sample.
In this research, the focus is on youth, therefore, we first have to define what does the term 'youth' stand for. In the UNDP youth strategy 2014-2017, term 'youth' refers to young women and men, in all their diversity of experiences and contexts, taking into consideration the existing definitions of youth used at the country and/or regional level(s). UNDP (2014) proposes to focus principally on young women and men ages 15 À 24, but also to extend that youth group to include young men and women ranging from ages 25 À 30 (and even beyond through age 35), based on contextual realities and regional and national youth policy directives. In this research, we will include in the sample young women and men ranging from ages 15 -35.

Variables
Although the concept of the research is limited by the USAID MEASURE-BiH National Youth Survey (NYS) data set, in the above-presented literature similar attempts aimed at identifying individual and household factors that may have an impact on the probability of employment of youth may be identified. Therefore, the main premise behind our theoretical concept is that various individual and household characteristics determine youth employment where we focus primarily on the following characteristics: education ( In our study, education is measured by the highest level of formal education (secondary or tertiary) and by participation in non-formal education, such as vocational training programme at a company; job search assistance (training on CV preparation, completing job applications, job interviewing and similar); short courses (e.g., languages, ICT skills, communication skills, etc.) and vocational classroom training programmes. Demographic characteristics are measured by age, gender, ethnicity and residence. Finally, household characteristics are operationalised by the household size, household income, remittances and contribution to the household budget.

Research design
The research is organised in three phases. The first phase brings an analysis of the basic parameters of descriptive statistics of the selected variables. These results have been considered of immense importance in terms of proper understanding of specificities of the sample. The second phase of the analysis is based on the probit model, relating youth employment probabilities to a vector of individual and household characteristics. In the last phase, the empirical results of the research have been presented and discussed. Table 2 contains a short overview of the selected variables.

Results
The youngest respondent from our sample is 15 years old, and the oldest is 35 years old. The average age is 24.02 years with a standard deviation of 5.71. The average number of household members is 2.41 with a standard deviation of 1.30.

Model performance analysis
To evaluate the impact of selected factors on the probability of youth employment in Bosnia and Herzegovina probit model was used. Pearson chi-square statistics results confirmed the entire model (with all predictors included) as statistically significant (p ¼ 0.000). In other words, the model as a whole fits significantly better than a model with no predictors. This was also confirmed by the Hosmer and Lemeshow goodness-of-fit test. According to the classification tables, the model correctly classifies 90.31% of cases. When it comes to the area under the ROC curve, its value is 0.9401. Table 3 displays the results of the estimated model.
According to the results presented in Table 3, twelve independent variables were statistically significant in the model.
Results of marginal effects at the mean (MEMs), average marginal effects (AMEs) and odds ratio (OR) for the estimated probit regression model are presented in Table  4. We only report these results for those variables whose parameters were statistically significant.

Discussion
It is argued by the OECD (2011) that people with higher levels of education have employment prospects; the difference is particularly marked between those who have attained upper secondary education and those who have not. That is why our first proposed hypothesis stated that education is positively related to the probability of youth employment in Bosnia and Herzegovina.
Looking at the results presented in Table 3, and when it comes to formal education, young women and men, from Bosnia and Herzegovina, with completed secondary school (as the highest level of formal education obtained) comparing to others levels of formal education, would have about 7.68 times higher likelihood of being employed comparing to those with other lower level of formal education. The predicted probability of being employed is 16.00% greater for an individual with completed secondary school. When all covariates are at their means, the predicted probability of being employed is 32.40% greater for an individual with completed secondary school. The expected change is statistically significant. Speaking of tertiary education, those young women and men who would complete tertiary education (as the highest level of formal education obtained) would have 10.94 times higher likelihood of being employed compared to those with a lower level of formal education. The predicted probability of being employed is 18.60% greater for an individual with completed tertiary education. When all covariates are at their means, the predicted probability of being employed is 37.60% greater for an individual with completed tertiary education. The expected change is statistically significant. Presented results are following the findings of Oancea et al. (2016) who, by estimated several logit models, found out that education influences the odds of being unemployed and that increasing levels of education are correlated with decreasing odds of unemployment. Yusuf et al. (2019), Msigwa and Kipesha (2013), Ahmad and Azim (2010), Hussain et al. (2016), Isengard (2003, etc. came to similar conclusions when it comes to the impact of education on employment status. Speaking of training, as a type of nonformal education, results are indicating that young women and men from Bosnia and Herzegovina who have participated in training programmes supporting self-employment, or vocational classroom training programmes, would have 1.57 times higher likelihood to be currently employed compared to those who have not. The expected change is statistically significant. These results are also visible in Abdullai et al. (2012), Cairo and Cajner (2013) and Otero (2016) who have also shown a positive link between non-formal education and youth employment. Based on the presented results we can accept our first hypothesis that education has a positive impact on youth employment probability in Bosnia and Herzegovina.
Our second research hypothesis stated that demographic characteristics determine the probability of youth employment in Bosnia and Herzegovina.
Our results are pointing to a possible age employment gap. In the literature, age has already been determined as an important employment determinant (D an acic a (2015), Ahmad and Azim (2010), Hussain et al. (2016), Harris (1996), Hammer (1999, etc.).
Besides the age gap, the results of our research are also pointing to a possible gender gap in terms of youth employment, where men would have 1.99 times higher likelihood to be employed when comparing to women. The predicted probability of being employed is 5.40% greater for young men from Bosnia and Herzegovina. When all covariates are at their means, the predicted probability of being employed is 11.0% greater for young men from Bosnia and Herzegovina. The expected change is statistically significant. These results aren't surprising. In that respect, O'Reilly et al. (2019) have even stated that despite national differences in youth employment, many countries share striking similarities in the uneven sectoral distribution of job opportunities for young women and men in Europe. These results are following the results reported by Adeniran et al. (2020), Dibeh et al. (2019), Nunez and Livanos (2010), Garrouste et al. (2010), Lynch (1986), Ahmad and Azim (2010), Tasci and Tansel (2005), Hussain et al. (2016), Rodokanakis and Vlachos (2013), Msigwa and Kipesha (2013), Karamessini et al. (2019) and many others. Based on the presented results we can accept our second hypothesis, i.e., demographic characteristics determine the probability of youth employment in Bosnia and Herzegovina. Our third hypothesis dealt with the impact of household characteristics on the probability of youth employment in Bosnia and Herzegovina. The results are reporting that, besides household income, regular contribution to the household's budget seems to impact the probability of employment. This is following the findings of Harris (1996) who stated that financial commitments exert have a positive impact on employment prospects among youth. Similar, Ahmad et al. (2015) found out that responsibilities within household increase the economic participation of youth. Based on the presented results we can accept our third hypothesis, i.e., household characteristics may be considered as antecedents of the probability of youth employment.
To date, a considerable body of research has sought to understand the impact of certain individual and household characteristics on the probability of youth employment. However, this study is the first comprehensive research dealing with an indepth analysis of youth employment probability in Bosnia and Herzegovina. Therefore, presented results make a significant contribution to the contemporary literature, especially when it comes to understanding individual and household antecedents of the probability of youth employment in the case of the transitional countries such as Bosnia and Herzegovina.
These results also provide useful practical implications. The study is one of the first research studies dealing with this topic in Bosnia and Herzegovina which was conducted following the principles of representative sampling methodology framework. As such, the findings obtained during this research can be viewed as reliable and can serve as a solid base for the development of evidence-informed policy decision making, such as tailoring specific and adequate active labour market programmes for young people.

Conclusion
To date, a considerable body of research has sought to understand the impact of certain individual and household characteristics on the probability of youth employment. However, our research is the first comprehensive research dealing with an in-depth analysis of youth employment probability in Bosnia and Herzegovina. The present findings demonstrate that education, age, gender and certain household characteristics, have an impact on the probability of youth employment.
Here we will emphasise once again those results that might have certain policy implications. First, in this research, we have found evidence of a statistically significant relationship between the current employment status of young women and men in Bosnia and Herzegovina and their involvement in training programmes. Therefore, the real policy implications of this research can be seen in the evidence-informed policy decision making, such as, tailoring specific and adequate active labour market programmes for young people that primarily need be focussed on: vocational training programme at a company, job search assistance (training on CV preparation, completing job applications, job interviewing and similar), training programmes that support self-employment and vocational classroom training programmes.
Second, to boost employment of youth in Bosnia and Herzegovina, policymakers need to consider curricula adjustments in formal education, that would include a recommendation for extra-curricular activities or practical work experience organised by the high school and/or university, internships, volunteering and/or other types of work experience not organised by the high school and/or university and some form of education/training outside of the school/university. Third, the results of this research are pointing to a possible gender gap in terms of youth employment, where men have a higher likelihood to get a job comparing to women. Being a male has a positive and significant impact on the employment status of youth in Bosnia and Herzegovina. Therefore, policymakers need to make sure also to tailor specific women's training and employment programme and to motive them for inclusion.
Although the concept of the research is limited by the USAID MEASURE-BiH National Youth Survey (NYS) data set, these results may produce useful pieces of information which might be helpful for government decision-makers in Bosnia and Herzegovina in the process of creating employment policies to support young people. Another possible limitation is that the data available are cross-sectional rather than longitudinal and therefore we cannot study any population changes across time.
Taking into account that corruption is one of the key factors influencing an overall quality of life in Bosnia and Herzegovina, future research studies could tackle this issue in more detail in terms of to what extent corruption prevents a good or excellent candidate from getting a job. Besides, an analysis could be made in terms of analysing the motivation of graduate students in terms of their personal beliefs towards the influence of corruption on their employment in Bosnia and Herzegovina and their readiness to seek for a job opportunity abroad.