Intergenerational effects of further vocational training in Germany

ABSTRACT

This article analyses the influence of unemployed parents’ further vocational training on their children’s future apprenticeship and employment opportunities. We focus on households receiving means-tested benefits in Germany and investigate whether parents’ further vocational training helps avoid an intergenerational transmission of unemployment. Parents’ further vocational training can affect children’s future apprenticeship and employment opportunities directly, for instance by raising the value attributed to education and training within the family. Indirect effects via parents’ improved employment opportunities are possible as well. The sample consists of adolescents who were 14–17 years old when their parents participated in a further vocational training programme. We evaluate apprenticeship and employment outcomes for the young adults at ages 17–23 using entropy balancing methods on the basis of rich large-scale register data. Our findings indicate significantly positive effects of parents’ further vocational training on children’s apprenticeship enrolment but non-significant or negative effects on children’s regular employment at young ages. Moreover, mediation analysis gives evidence that these effects are partly mediated via an increase in parents’ own employment opportunities. Overall, our analyses indicate that parents’ further vocational training improves the qualifications of the next generation as well.

RÉSUMÉ

Cet article analyse l’influence des formations continues suivies par des parents sans emploi sur les opportunités d’apprentissage et d’emploi futures de leurs enfants. Nous nous concentrons sur des ménages percevant des prestations liées aux ressources en Allemagne et nous penchons sur la question de savoir si la formation continue des parents contribue à éviter une transmission intergénérationnelle du chômage. La formation continue des parents peut avoir une influence directe sur les opportunités d’apprentissage et d’emploi futures des enfants : elle peut notamment augmenter l’importance accordée à l’éducation et à la formation au sein de la famille. Une incidence indirecte par une amélioration des opportunités d’emploi des parents est aussi possible. L’échantillonnage se compose d’adolescents âgés de 14 à 17 ans, dont les parents ont participé à un programme de formation continue. Nous évaluons les débouchés d’apprentissage et d’emploi pour les jeunes adultes âgés de 17 à 23 ans à l’aide de méthodes de pondération entropique sur la base de données de registres à grande échelle complètes. Nos conclusions mettent en évidence des effets positifs significatifs des formations continues suivies par les parents sur le programme d’apprentissage des enfants, mais des effets non significatifs ou négatifs sur l’emploi régulier des enfants à un jeune âge. De plus, l’analyse de médiation révèle que ces effets sont partiellement liés à une augmentation des opportunités d’emploi des parents. De manière générale, les analyses réalisées montrent que la formation continue des parents améliore également les qualifications de la génération suivante.

KEYWORDS:

• Intergenerational mobility
• unemployment
• training
• welfare benefits
• labour market policy
• Germany

MOTS-CLÉS:

• Mobilité intergénérationnelle
• chômage
• formation
• prestations sociales
• politique de marché du travail
• Allemagne

Introduction

Parents’ employment and working conditions affect their children’s educational opportunities in the long term. Parental working conditions and work-family balance have been found to affect the family environment and the well-being and educational success of their children (Kaiser, Li, & Pollmann-Schult, 2017; Tulk, Montreuil, Pierce, & Pépin, 2016). Parents’ unemployment lowers children’s optimism about their educational opportunities (Lindemann & Gangl, 2018), negatively affects children’s educational outcomes (Coelli, 2011; Lindemann & Gangl, 2017) and is associated with children’s higher risks of unemployment as adults (Ekhaugen, 2009; Mäder, Riphahn, Schwientek, & Müller, 2015). Moreover, parents have been found to subjectively view breadwinning as a form of care (Schmidt, 2018). Thus not being able to provide such care may be detrimental to the family climate.

Lacking a vocational qualification greatly diminishes an individual’s employment opportunities in Germany (Söhnlein, Weber, & Weber, 2016). Further vocational training, administered by jobcentres at the local level, provides specific vocational qualifications, and substantially improves long-term unemployed people’s employment opportunities (Bernhard, 2016). Our research asks whether the next generation benefits from parents’ further vocational training as well, thus potentially contributing to avoiding an intergenerational transmission of unemployment. Underlying causal mechanisms may be that parents’ improved employment opportunities – especially for better quality work – via further vocational training enable them to invest in their children’s education, or that their employment raises their children’s self-efficacy.

Our sample consists of young people aged 14–17 with at least one non-employed parent in households receiving means-tested unemployment benefits in June 2007. Using entropy balancing methods, we study effects of parents’ further vocational training in the second half of 2007 on their children’s apprenticeship, pre-vocational training and regular employment outcomes in the years 2008–2014. Using mediation analysis, we determine the proportion of the intergenerational effect that is mediated via effects of further vocational training on parents’ employment opportunities.

Unemployment benefit II

This article focuses on households receiving Unemployment Benefit II (UB II), a flat-rate means-tested welfare benefit for households with at least one member who is capable of employment. UB II recipients include people who are not eligible for or have run out of their unemployment insurance benefit. The base UB II benefit as of January 2018 is 416 € for singles, 374 € for each partner of a couple, and 240–316 € for children, depending on their age. Costs of housing and heating as well as health insurance are covered additionally. Reduced UB II rates can be obtained to supplement earnings or unemployment insurance benefits up to the means-testing threshold. UB II recipients are expected to accept any type of employment and to participate in any active labour market programme (ALMP) proposed by their case manager in the jobcentre to improve their employment opportunities.

UB II recipients make up the majority of the stock of the unemployed in Germany (66% in 2017), while inflows of unemployed people into UB II (51%) and unemployment insurance (49%) are nearly equal. Nearly half (48%) of all unemployed UB II recipients are long-term unemployed, i.e. have been unemployed for over 12 months (Statistik der Bundesagentur für Arbeit, 2018a, 2018b), which implies lower opportunities of employment entry than for the short-term unemployed (Klinger & Rothe, 2012). Moreover, children in households receiving UB II are likely to be subject to poverty for extended periods of time (Tophoven, Lietzmann, Reiter, & Wenzig, 2017). Thus it is crucial for UB II recipients to have access to measures that improve their employment opportunities.

Further vocational training programmes for UB II recipients in Germany

We focus on further vocational training programmes, as they have been found to have substantial positive employment effects for UB II recipients (Bernhard, 2016). Further vocational training programmes have a long tradition in Germany, and can be classified into medium-term programmes running for several months and longer-term (at least two-year) comprehensive vocational retraining programmes providing a vocational degree. We focus on medium-term further vocational training because case numbers for longer-term comprehensive retraining courses are too small. Medium-term courses include vocationally oriented programmes, such as accountancy training, business English, elderly care, data processing and programming, safety officer training, or train conductor qualifications. Medium-term courses also include partial vocational qualification programmes that accumulatively lead to a vocational degree as well as supplementary courses qualifying for an advanced vocational position (e.g. master craftsman) (Bundesagentur für Arbeit, 2017).

To participate in further vocational training programmes, benefit recipients receive vouchers for courses offered by certified external for-profit or non-profit providers. Case managers in jobcentres may award such vouchers given that they consider further vocational training necessary for a benefit recipient to find employment. A voucher may also be awarded if the benefit recipient does not yet have a vocational degree, but has already been employed for at least three years, or already has a vocational degree, but has been employed in unqualified occupations, has been unemployed or has done unpaid care work for more than four years (Social Codes II and III).

In 2005, only about 65,000 UB II recipients started a further vocational training programme, rising to somewhat over 160,000 in 2007 and more than 200,000 each year in 2008 to 2010, but declining again after 2010 to approximately 150,000 in 2013 (Bernhard, 2016; Statistik der Bundesagentur für Arbeit, 2014).

On the individual level, Bernhard (2016) finds stable positive employment effects of 8–12 percentage points of medium-term further vocational training throughout the duration of three to eight years after programme start for UB II recipients in Germany. Kruppe and Lang (2014) evaluate the effects of longer comprehensive retraining courses providing a degree, and find effects of 12–19 percentage points after four years.

Kluve (2010) provides a meta analysis indicating that more than half of international studies evaluating training programmes find significant positive employment effects. Moreover, Creed, Bloxsome, and Johnston (2001) provide evidence for Australia that training courses improve unemployed participants’ self-esteem and job-search self-efficacy. So far, however, there is no German or international study that has analysed intergenerational employment, education or psychological effects of ALMPs.

The German education system

Secondary school education

Secondary school education in Germany begins at the fifth grade, at which time children are directed into different educational pathways. The early age at which this takes place is often named as an important cause of low intergenerational educational mobility in Germany (e.g. Piopiunik, 2014). A degree qualifying for vocational training can be obtained after grade nine or 10, and qualification for tertiary education after grade 12 or 13. At age 16–17, 72% of the population was still enrolled in any type of secondary education (Statistisches Bundesamt, 2018).

Apprenticeships

Apprenticeships are the most frequent type of post-secondary education in Germany (Bundesinstitut für Berufsbildung, 2017). The majority of students begin vocational training at ages 17–20, while 11% are younger and 29% are older (Statistisches Bundesamt, 2017). The German apprenticeship system, known as the dual education system, is a combination of firm-based training and classes taken at vocational schools.

In 2014, a vocational degree was the highest degree attained by 57% of the population aged 30–35, while only 26% held a tertiary degree, and 16% had no vocational or tertiary degree (Autorengruppe Bildungsberichterstattung, 2016). Being without a vocational degree strongly lowers employment opportunities. Only 4.6% of the labour force who held a vocational degree and 2.4% of those with a tertiary degree were unemployed in 2015, compared to 20.3% of those without any vocational or tertiary degree (Söhnlein et al., 2016). Moreover, Riphahn and Zibrowius (2016) find that vocational training leads to lower risks of non-employment, higher opportunities of permanent full-time employment and higher wages at age 25 compared to young people without vocational training.

As apprenticeships are the most common form of post-secondary education in Germany, and as they greatly improve graduates’ employment perspectives, raising apprenticeship opportunities for children from households receiving means-tested benefits is an important goal. Thus this article investigates effects of unemployed parents’ further vocational training on their children’s opportunities of apprenticeship enrolment.

We do not investigate effects on children’s tertiary education. The time elapsed since the introduction of UB II is not yet sufficient to conclusively study intergenerational effects on university education. Moreover, as described in the next section, enrolment rates in tertiary education can be expected to be very low for our sample, such that intergenerational effects of parents’ further vocational training on their children’s tertiary education would most likely be very small.

Intergenerational transmissions of educational opportunities

Parental education remains an important determinant of children’s educational opportunities in Germany (Minello & Blossfeld, 2017). For instance, in birth cohorts 1970–1986, a tertiary degree was attained by only 10%–12% of children whose parents had a low level of education (at most a lower secondary degree with a vocational degree), compared to 42%–43% of those whose parents had a tertiary degree (Boockmann, Kleimann, Meythaler, Nielen, & Späth, 2015). Moreover, parents’ and children’s probabilities of being without a vocational degree are strongly correlated (Bundesinstitut für Berufsbildung, 2017).

In the current article, the sample consists of families receiving UB II. In this sample, 44% of the mothers (in families with a mother: 42%/ 96%) and 33% of the fathers (in families with a father: 21%/ 63%) had no vocational degree (Table 1). Given an intergenerational transmission of education, children in our sample can thus be expected to have an above average risk of remaining without any vocational education, and to have very low opportunities of attaining a tertiary degree.

Table 1. Descriptive statistics for selected covariates.

		Balancing descriptives^a
		Covariate means			Standardized difference in means
	Total sample distribution	Treated	Controls pre-balancing	Controls post- balancing	Pre-balancing	Post- balancing
Family status
Two parents, married	50%	reference category
Two parents cohabiting	9%	10%	9%	10%	0.037	0
Lone mothers	37%	35%	37%	35%	−0.062	0.001
Lone fathers	4%	4%	4%	4%	0.017	0
Mother’s vocational degree
No degree	42%	reference category
Vocational degree	44%	55%	44%	55%	0.225	0.001
University/ college degree	4%	7%	4%	7%	0.119	0
Missing	6%	4%	6%	4%	−0.114	0
Not applicable because lone father	4%	4%	4%	4%	0.017	0
Father’s vocational degree
No degree	21%	reference category
Vocational degree	37%	43%	36%	43%	0.14	0
University/ college degree	3%	6%	3%	6%	0.11	0
Missing	2%	2%	2%	2%	−0.012	0
Not applicable because lone mother	37%	35%	37%	35%	−0.062	0.001
Mother’s nationality
Not German	26%	reference category
German	70%	74%	70%	74%	0.104	0.001
Not applicable because lone father	4%	4%	4%	4%	0.017	0
Father’s nationality
Not German	21%	reference category
German	42%	48%	42%	48%	0.126	0
Not applicable because lone mother	37%	35%	37%	35%	−0.062	0.001
Mother: duration since last job
<6 months	18%	reference category
6–11 months	5%	7%	5%	7%	0.065	0
1 year	8%	9%	8%	9%	0.03	0
2–4 years	21%	21%	21%	21%	0.004	0
5–7 years	11%	9%	11%	9%	−0.059	0
8–15.5 years	8%	7%	8%	7%	−0.037	0
>15.5 years or never	25%	19%	25%	19%	−0.152	0
Not applicable because lone father	4%	4%	4%	4%	0.017	0
Father: duration since last job
<6 months	19%	reference category
6–11 months	4%	7%	4%	7%	0.115	0
1 year	6%	8%	6%	8%	0.094	0
2–4 years	15%	16%	15%	16%	0.025	0
5–15.5 years	13%	9%	13%	9%	−0.13	0
>15.5 years or never	6%	7%	6%	7%	0.042	0
Not applicable because lone mother	37%	35%	37%	35%	−0.062	0.001
N	190,173	3,514	186,659

^aComplete tables of means and variances for all covariates pre- and post-balancing are provided in the supplementary appendix. All covariates are listed in the Data and Methods section.

Pathways for intergenerational transmissions of education and labour market opportunities

Parents’ education can directly affect children’s education, and can also have indirect effects via parents’ labour market opportunities. Direct intergenerational transmissions of education can occur along several different pathways. More highly educated parents have higher educational aspirations for their children (Piopiunik, 2014), are better able to help them with their studies and provide knowledge on the functioning of the education system (Pfeffer, 2008). In addition, Harding, Morris, and Hughes (2015) summarize evidence of direct positive causal effects of longer maternal education on children’s cognitive outcomes and educational attainment. Moreover, greater value attributed to education in the family can positively influence children’s educational achievements. For instance, Boll and Hoffmann (2015) find that mothers’ ‘soft factors’ (interest in further education, interest and participation in politics, or interest in exercise) positively affect children’s school achievements.

Moreover, in a qualitative study for the United States, Shiffman (2011) provides direct insight into mechanisms by which parents’ adult education led to greater involvement in their children’s education. Parents’ education participation led to joint study situations in which parents helped children with their homework, and in some cases children helped parents prepare for classes. On the basis of their shared learning experiences, parents were able to convey their educational ambitions to their children. Adult education also raised parents’ self-efficacy and fostered social networks that supported children’s education.

In addition to these direct effects, parents’ education can also indirectly affect children’s education by first affecting parents’ employment status. Parents’ employment status can then affect children’s education via various pathways. First, unemployed parents generally have lower economic resources than employed parents and are thus less well able to financially support their children’s education (Kalil & Wightman, 2011). Second, parental unemployment can lead to lower well-being or optimism about educational opportunities (Lindemann & Gangl, 2018) or to the perception that efforts may not pay off (Andersen, 2013; Gustafsson, Katz, & Österberg, 2017). Similarly, Tulk et al. (2016) find that parents’ working conditions affect their adolescent children’s psychological well-being and educational success.

Several studies have found evidence of negative causal effects of parents’ unemployment or welfare benefit receipt on children’s educational achievements (e.g. Coelli, 2011; Kallio, Kauppinen, & Erola, 2016). Lindemann and Gangl (2017) find negative effects of parents’ unemployment specifically on children’s transition to vocational training. Müller, Riphahn, and Schwientek (2017) is an exception in finding no or partly positive effects.

Beyond affecting children’s education, parent’s unemployment, welfare benefit receipt or low income could also affect children’s employment outcomes. International studies agree on the finding that children whose parents were unemployed or received welfare benefits have higher risks of unemployment or welfare benefit receipt as adults themselves (e.g. Kauppinen et al., 2014; Mäder et al., 2015). Such an association may be spurious, reflecting differences in parents’ levels of education, ambitions or income that existed prior to their unemployment or welfare benefit receipt. Causal effects are also possible, given that parents’ unemployment or welfare benefit receipt induce changes in their situation. Such possible changes include reductions in parents’ financial resources, changes in their ambitions, self-esteem or perceptions of self-efficacy. Several studies have attempted to determine causal effects using various methodological approaches, such as sibling difference, fixed effects, exogenous variation, instrumental variable, or Gottschalk methods. While Mäder et al. (2015) and Ekhaugen (2009) find an association between parents’ and children’s unemployment, they find no evidence of causal effects. Edmark and Hanspers (2015) similarly find no evidence of a causal effect of parents’ welfare benefit receipt. Müller et al. (2017) on the other hand find evidence of a causal effect of fathers’ unemployment on daughters’ but not sons’ worklessness and Siedler (2004) and Bratberg, Nilsen, and Vaage (2015) find causal evidence of an intergenerational transmission of welfare benefit receipt.

Thus evidence of causal effects of parents’ on children’s unemployment or welfare benefit receipt is mixed. In contrast, as discussed earlier, it seems clear that parents’ lower levels of education lead to a disadvantaged labour market situation for themselves and to lower educational outcomes for their children. This article examines whether improvements in welfare-recipient parents’ level of qualification via further vocational training affects their children’s apprenticeship and employment outcomes. The article studies both direct effects of parents’ further vocational training and indirect effects via parents’ employment opportunities. The next section develops hypotheses on such effects.

Hypotheses

This article studies further vocational training by parents during a decisive time in their children’s educational career, the final years of secondary school and the transition to apprenticeship training. As discussed in the previous section, parents’ training can have direct effects on their children’s education outcomes, for instance by fostering joint study situations in which parents can convey their heightened training aspirations. Thus parents may be able to help their children achieve better grades on their school leaving certificate and to write more successful apprenticeship applications. Parents’ newly gained social network via their training course may also be helpful in searching for apprenticeship openings. Therefore, our first two hypotheses are that unemployed parents’ further vocational training positively affects their adolescent children’s later apprenticeship opportunities as young adults (hypothesis 1), and that part of this effect is a direct effect of parents’ further vocational training (hypothesis 2).

Beyond the direct effect of parents’ further vocational training, we also expect an indirect effect via an increase in parents’ employment opportunities. As discussed earlier, medium-term further vocational training has been found to substantially increase participants’ employment opportunities (Bernhard, 2016). In turn, as discussed previously, parents’ employment can improve their children’s education outcomes, for instance via greater optimism about their educational opportunities or greater financial resources. Therefore, hypothesis 3 is that part of the effect of parents’ further vocational training on children’s apprenticeship opportunities is an indirect effect via parents’ employment.

In addition, we expect parents’ further vocational training to positively affect their children’s later employment opportunities as young adults. Possible mechanisms for this effect are that parents’ increased qualifications and levels of employment raise their children’s perceptions of self-efficacy and ambitions, resulting in children’s greater employment opportunities. Moreover, as a result of their parents’ further vocational training, children might first increase their apprenticeship rates, which can then later translate into better employment opportunities. Thus hypothesis 4 is that parents’ further vocational training increases adolescent children’s employment opportunities as young adults. Similar to the effect on children’s apprenticeship opportunities, we expect part of the effect on children’s employment opportunities to be a direct effect of parents’ further vocational training and part of the effect to be an indirect effect via parents’ increased employment opportunities.

Finally, pre-vocational training programmes are programmes offered by jobcentres to improve young people’s opportunities of successfully applying for an apprenticeship. These programmes are directed especially towards young people with difficulties making the transition to an apprenticeship. Parents who participated in further vocational training might inquire about training options offered by the jobcentre for their children, leading to higher participation in pre-vocational training for children whose parents participated in further vocational training (hypothesis 5).

Data and methods

We use the Integrated Employment Biographies (IEB) and the Unemployment Benefit II History data set (LHG) which provide longitudinal employment, unemployment, and programme participation information as well as household information. These data sets were prepared for scientific analysis on the basis of Federal Employment agency register data stemming from notifications sent by employers to health and pension insurance funds and data from employment offices (vom Berge, König, & Seth, 2013).

The sample consists of adolescents aged 14–17 in households receiving UB II on 1 June 2007, whose mother and/or father was not employed but capable of employment and not already participating in an ALMP at this point in time. Young people either of whose parents participated in further vocational training programmes in the second half of 2007 are compared to young people whose parents did not participate in the programme during that time. Following Sianesi (2004), the comparison group includes people who participated in the programme in later time periods, to avoid conditioning on the future. We do not expect future participation to play a major role, as only 4.5% of comparison group members began further vocational training before their child’s 18th birthday.

In the first two years after the introduction of UB II in 2005, participation in further vocational training was still very low (Bernhard, 2016). Thus we chose the second half of 2007 as the earliest possible time window that allows sufficient sample sizes, while providing as long an observation period as possible after parents’ programme participation.

We study children’s outcomes several years later, in 2008–2014. First, we study outcomes for the full sample at ages 19–21. We then divide the sample into two subgroups, as this allows observing outcomes for a longer time-span altogether. For children aged 14–15 at their parents’ programme participation, we measure outcomes at ages 17–21, and for children aged 16–17 at their parents’ programme participation, we measure outcomes at ages 19–23. We do not make conclusive predictions on the relative strength of the effects for the two groups. Intergenerational effects have more time to develop for the younger group, but the older group is at a pivotal time in their educational career, which might also lead to stronger effects.

In the case of siblings within the same age group, we randomly chose one of the siblings for the analysis. This is why the sample size for the complete group is slightly lower than the sum of the two subgroups. Children’s apprenticeship and employment outcomes are measured on the exact days of their 17th–23rd birthdays. Children’s pre-vocational training participation is measured as any pre-vocational training in the year following each birthday.

We used entropy balancing methods to calculate Average Treatment Effects on the Treated (ATTs) of parents’ programme participation on children’s outcomes, applying the Stata utility ebalance (Hainmueller & Xu, 2013). Entropy balancing is a method for weighting cases in the non-participant group (control group) in order to adjust their covariate distribution to that in the participant group (treatment group) with respect to specified covariate moments. Thus entropy balancing makes it possible to reweight data for causal inference in observational studies (Hainmueller, 2012).

In this article, we specified the first two moment constraints (means and variances) to adjust the covariate distributions in the control group to those in the treatment group via entropy balancing. The covariates used to calculate weights using entropy balancing include adolescents’ individual characteristics, such as gender, age group, activity status at the sampling time point, number of siblings, age of the youngest sibling, and whether they are living in a one-parent or two-parent family. We also consider characteristics of all parents living in the household, such as education, disability, age, whether they are lone parents, marital status, nationality, employability, cumulated employment experience in contributory employment and minijobs, cumulated past UB II receipt, number of previous job spells, past labour market programme participations, duration since parents’ last job and characteristics of that job, including occupation, earnings and employment status. Moreover, we take regional labour market classifications into account.

To identify causal treatment effects, the stable unit treatment assumption as well as the strong ignorability assumption must hold (Rosenbaum & Rubin, 1983). The stable unit treatment assumption postulates that an individual’s outcome is independent of the treatment of other individuals in the population. Further vocational training is a small programme, less than 2% of the sample participated during the observed participation period. Thus it is unlikely that comparative employment advantages of participants should affect the employment opportunities of non-participants.

The strong ignorability assumption is that the outcome is independent of the treatment decision given balancing on the included variables. To meet this assumption, it is important to take the comprehensive set of variables listed above into account. As described earlier, assignment to further vocational training is not random. There are clear qualification criteria for the programme. On the basis of these criteria, participants and non-participants can be expected to differ regarding prior vocational degrees, employment experience, prior occupation and household structure. We can account for these differences on the basis of the variables included in our estimates.

Beyond such observed differences, unobserved characteristics may influence programme participation. Particularly motivated benefit recipients can be reasonably assumed to more often address training options with their case manager. In a recent methodological study using linked survey and administrative data, Caliendo, Mahlstedt, and Mitnik (2017) show that attitudes and personality traits significantly affect programme participation probabilities. However, including variables on attitudes, personality traits or motivation does not make a difference for estimated programme effects, as long as detailed labour market histories and household information are controlled for (Biewen, Fitzenberger, Osikominu, & Paul, 2014; Caliendo et al., 2017). As we can rely on detailed labour market histories and household information, unobserved attitudes or motivation should not greatly alter our results. Although it would nevertheless be preferable to directly control for such characteristics, they are not available in the administrative data we use, and sample sizes in alternative survey data are not sufficient for our focus group. Moreover, an important advantage of the administrative data over survey data is that it does not suffer from panel attrition or recall error.

Table 1 and Tables A4–A6 show that the standardized difference in means after balancing (weighting) is far below 0.1 for each covariate in each sample, which indicates that balancing was successful (Austin, 2009).

As a robustness check, Table B1 provides results using propensity score matching methods. Findings are almost identical to those using entropy balancing (Table 2).

Table 2. Average Treatment Effects on the Treated (ATTs) in percentage points of parents’ further vocational training on children’s outcomes. Entropy balancing.

	Children’s outcomes at ages^a							N Treated	N Controls
	17	18	19	20	21	22	23
Age 14–17 at parents’ programme participation								3,514	186,659
Children’s outcomes:
Apprenticeship			1.9**	2.9***	2.5***
Regular employment			−0.8***	0.1	0.8
Pre-vocational training			1.0*	1.3**
Age 14–15 at parents’ programme participation								1,983	105,518
Children’s outcomes:
Apprenticeship	0.2	−0.1	2.1**	3.2***	2.5***
Regular employment	−0.1	−0.5**	−1.3***	−0.5	0.8
Pre-vocational training	0.3	0.6	1.8**	1.8***
Age 16–17 at parents’ programme participation								1,951	105,826
Children’s outcomes:
Apprenticeship			1.8*	2.6**	2.7***	1.7*	0.6
Regular employment			−0.3	0.8	1.0	1.0	1.8*
Pre-vocational training			0.2	1.2	0.9	0.8

*p < 0.10.

**p < 0.05.

***p < 0.01.

^aChildren’s apprenticeships and regular employment are measured at the time points of their birthdays for each of the ages given in the table. Children’s pre-vocational training participation is measured as any pre-vocational training participation within the year following each birthday.

We used the medeff program by Hicks and Tingley (2011) for the mediation analysis. In our case, the mediator is the proportion of days that parents spent in employment between 1 June 2007 (the sampling time point) and the respective time of outcome measurement (the child’s respective birthday). In the case of lone parents, the mediator variable was the lone parent’s proportion of days spent in employment. In the case of two-parent families, the mediator variable was the average of both parents’ proportion of days spent in employment. We applied mediation analysis to the sample weighted by entropy balancing weights.

Results

The results in Table 2 show that parents’ programme participation indeed affects the training and employment outcomes of their 14- to 17-year-old children. For the full group, we find significantly positive effects on children’s apprenticeship enrolment at ages 19–21 (1.9–2.9 percentage points), an increase by 8–11% of the base rate (see Table A1).

For the younger subgroup, we likewise find significantly positive effects of parents’ programme participation on children’s probability of being in an apprenticeship at ages 19–21 (2.1–3.2 percentage points). Further vocational training participation by parents with 16 and 17 year old adolescents positively influences their children’s future apprenticeship opportunities at ages 19–22, raising them by 1.7–2.7 percentage points. Thus, overall, our findings for effects on children’s apprenticeship opportunities support hypothesis 1 that parents’ further vocational training positively affects their adolescent children’s later apprenticeship opportunities as young adults. The strength of these effects does not greatly differ between the two age subgroups.

Parents’ further vocational training negatively influences children’s regular employment probability at age 19 for the full group and at ages 18–19 for the younger subgroup (Table 2). For the older subgroup, we find a marginally significant positive effect on children’s regular employment at age 23 (1.8 percentage points). One explanation for the negative employment effects at ages 18–19 might be that parents’ further vocational training extends children’s education time and prevents children’s early employment entries directly after mandatory education. In any case, our findings for ages 18–19 contradict hypothesis 4 that parents’ further vocational training would positively affect children’s regular employment opportunities as young adults, and our findings for age 23 provide limited support for hypothesis 4.

In the full sample, parents’ participation in further vocational training positively influences their children’s probability of participating in pre-vocational training at ages 19 and 20 (1–1.3 percentage points, Table 2). Our findings for the younger subgroup similarly indicate a positive effect of 1.8 percentage points for ages 19–20. For the older subgroup, we find no significant effects on children’s probability of pre-vocational training participation. Thus we find support for hypothesis 5 for the younger but not the older subgroup.

Table A2 in the supplementary appendix shows effects of parents’ further vocational training on additional outcomes. Negative effects are found on children’s unemployment as well as UB II receipt while not participating in the labour market. In contrast, positive effects are found on children not receiving UB II while not participating in the labour market, a category which includes student loan recipients. Thus, altogether, there appears to be a tendency for parents’ further vocational training to deter children’s early employment, unemployment and benefit receipt and to keep children in education longer.

Table A3 shows results from a robustness check using an alternative time point specification for measuring the outcomes. Instead of children’s exact birthdays, we measured outcomes on the first of October, assigning children to their normative school cohorts. This leads to measuring outcomes about nine months later than in our original estimates. Although effects appear shifted to the left by a year, findings for apprenticeship opportunities are quite robust to this alternative specification. However, we no longer find significantly negative employment effects at young ages.

Table 3 shows the results of the mediation analysis, indicating which part of each effect in our main analyses is a direct effect of parents’ further vocational training participation and which part is a mediated effect via parents’ employment. The sum of the average mediated effect and the average direct effect is the complete ATT. For children’s apprenticeship opportunities, we find significantly positive direct as well as mediated effects of parents’ further vocational training for ages 19–21 for the complete sample, as predicted by hypotheses 2 and 3. We find similar effects for both subsamples for ages 20–21. The direct effect is almost always larger than the mediated effect. Thus a possible interpretation is that participating in further vocational training especially raises the importance that parents attribute to education and that this is conveyed to their children. However, the mediated effect via parents’ employment is significant as well. The mediated effect might be interpreted as an effect of parents’ greater financial resources or perceptions of self-efficacy due to higher employment rates.

Table 3. Mediation analysis. Mediated and direct effects of parents’ further vocational training participation on children’s outcomes. ATTs in percentage points.

	Children’s outcomes at ages^a							N treated	N controls
	17	18	19	20	21	22	23
Age 14–17 at parents’ programme participation								3,514	186,539
Children’s outcomes:
Apprenticeship
Average mediation			0.6***	0.7***	0.4***
Average direct effect			1.3*	2.3***	2.1***
Regular employment
Average mediation			0.0	0.3***	0.5***
Average direct effect			−0.9**	−0.1	0.3
Pre-vocational training
Average mediation			−0.4**	−0.3***
Average direct effect			1.4**	1.7***
Age 14–15 at parents’ programme participation								1,983	105,461
Children’s outcomes:
Apprenticeship
Average mediation	0.3***	0.7***	0.9***	0.9***	0.5***
Average direct effect	−0.1	−0.7	1.3	2.3**	2.1**
Regular employment
Average mediation	0.0	0.1**	0.1	0.2***	0.6***
Average direct effect	−0.1	−0.5**	−1.3***	−0.7	0.2
Pre-vocational training
Average mediation	−0.1*	−0.2**	−0.3***	−0.3***
Average direct effect	0.5	0.8	2.1***	2.1***
Age 16–17 at parents’ programme participation								1,951	105,749
Children’s outcomes:
Apprenticeship
Average mediation			0.4***	0.6***	0.4***	0.0	0.1
Average direct effect			1.3	2.0*	2.4**	1.8*	0.6
Regular employment
Average mediation			0.1	0.3***	0.6***	0.9***	1.2***
Average direct effect			−0.3	0.5	0.4	0.1	0.6
Pre-vocational training
Average mediation			−0.3***	−0.3***	−0.3***	−0.2**
Average direct effect			0.5	1.5*	1.2*	1.0*

*p < 0.10.

**p < 0.05.

***p < 0.01.

^aChildren’s apprenticeships and regular employment are measured at the time points of their birthdays for each of the ages given in the table. Children’s pre-vocational training participation is measured as any pre-vocational training participation within the year following each birthday.

With respect to children’s employment, we find significantly positive mediated but not direct effects of parents’ further vocational training (Table 3). Thus it seems that parents’ further vocational training only positively affects children’s employment under the condition that it first raises their own employment opportunities. This is especially true for younger ages of the child, where direct effects of parents’ further vocational training on children’s employment are negative, perhaps because the direct effect is to keep children in education longer. A possible explanation for positive mediated effects on children’s employment might be that these effects are network effects. Employed parents may have more information on job openings and application strategies. Altogether, our findings only partly support hypothesis 4 that there would be positive direct and mediated effects on children’s employment.

Discussion

This is the first paper to analyse intergenerational effects of an ALMP. The focus is on intergenerational effects of further vocational training for means-tested unemployment benefit recipients in Germany. We study the influence of parents’ participation in further vocational training on their children’s pre-vocational training, apprenticeships, and regular employment at a later point in time. Our article aims to contribute to an understanding of whether ALMPs can reduce intergenerational transmissions of unemployment.

Characteristics of parents’ employment situation impact their children’s well-being and educational outcomes (Kaiser et al., 2017; Tulk et al., 2016). An important aspect of the work-family nexus is how children are affected by their parents’ labour market integration. For instance, parental unemployment has been found to lower children’s optimism about their educational opportunities (Lindemann & Gangl, 2018), and to lead to the perception that efforts may not pay off (Andersen, 2013). In turn, parents’ unemployment can diminish children’s educational outcomes (Kallio et al., 2016; Lindemann & Gangl, 2017).

We expected the employment opportunities of unemployed parents receiving means-tested benefits to improve as a consequence of further vocational training. Parents’ employment can enable them to invest in their children’s education, and can also raise their children’s self-esteem and ambitions. Therefore, we expected a positive effect of parents’ further vocational training on their children’s apprenticeship and employment outcomes, mediated via parents’ improved employment opportunities.

Moreover, beyond a mediated effect, we expected direct effects of parents’ further vocational training participation on their children’s apprenticeship opportunities, as parents’ training may raise the importance attributed to education in the family. We used rich administrative data and focused on adolescents who were 14–17 years old when their parents participated in further vocational training. Outcomes for the young adults are evaluated at ages 17–23, depending on their age during their parents’ programme participation. We compared families from participant and non-participant groups using entropy balancing methods.

The results show that parents’ further vocational training positively influences children’s apprenticeship opportunities. The larger part of this effect is a direct effect of parents’ further vocational training. This finding of a predominant direct effect may be interpreted as indicating that further vocational training increases the importance parents attribute to education, and that they convey this greater valuation of education to their children. We also find evidence of a smaller but likewise significantly positive mediated effect of parents’ further vocational training via parents’ increased employment levels. An interpretation is that employed parents are better able to invest in their children’s education. Parents’ employment may also raise parents’ and children’s levels of self-esteem and perceptions of self-efficacy, which can be beneficial for children’s apprenticeship opportunities. The direct effect we measure may also capture the impact of further mediators not accounted for in this analysis, such as parents’ working conditions, work-family compatibility, occupation or income level.

In contrast to the positive effects on children’s apprenticeship opportunities, depending on the specification, we find non-significant or negative effects of parents’ further vocational training on children’s employment at young ages. Possibly, children whose parents participate in further vocational training remain in school longer and enrol in an apprenticeship rather than entering employment directly at an early age. For the oldest age for which we measure children’s outcomes, age 23, we find a marginally significant positive effect of parents’ further vocational training on children’s employment rate. Data for longer observation periods is not yet available.

A topic for future research is to study effects of parents’ further vocational training on children’s longer-term employment opportunities, for instance on children’s employment opportunities in their mid- and late twenties. Moreover, future research can examine effects of parents’ participation in other types of ALMPs, such as job creation programmes or employment subsidies. Analyses of programmes with greater numbers of participants (for instance workfare programmes) could differentiate by further subgroups, for instance by parents’ and children’s gender. Future analyses could also investigate whether outcomes differ for younger children, as pressure on parents of young children receiving means-tested benefits to enter employment has risen since the Hartz reforms in Germany (Jaehrling, 2015). In addition, future qualitative research could expand upon the study by Shiffman (2011) for the United States, to gain direct insight as to whether and how parents’ education participation may lead to greater involvement in their children’s education in various country and social policy contexts.

Altogether, our findings imply that investments in further vocational training for parents receiving means-tested unemployment benefits are particularly rewarding. Such vocational training investments not only improve the employment opportunities of societal groups facing the greatest employment disadvantages, but also counteract negative impacts of parents’ unemployment on their children’s educational outcomes, resulting in improved apprenticeship opportunities for the next generation as well.

Open questions remain. Due to means-tested benefit recipients’ low formal qualifications on average, the jobs they take up may not always be good quality jobs, or jobs characterized by high work-family compatibility. Previous research has shown that characteristics of parents’ employment situation, such as non-standard schedules, impact their children’s well-being and educational outcomes (Kaiser et al., 2017; Tulk et al., 2016). However, by raising qualification levels, further vocational training not only improves job opportunities, but also opportunities for better quality jobs (Dengler, 2016). Thus a question for future research is to investigate the role of further mediators such as job quality and work-family compatibility.

Acknowledgements

We are very grateful to Silke Anger, Bob Hart, Barbara Hofmann, Malte Reichelt, Rüdiger Wapler, Jürgen Wiemers, Joachim Wolff, as well as two anonymous reviewers for very helpful comments and advice.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Cordula Zabel is a senior research scientist at the Institute for Employment Research in Nuremberg, Germany. She was a PhD student at the Max Planck Institute for Demographic Research in Rostock and received her PhD in Demography from the University of Rostock in 2006. Her research interests are in the fields of household dynamics, labour market transitions, and social policy.

Eva Kopf is a senior research scientist at the Department for Senior Citizens and Intergenerational Affairs of the City of Nuremberg, Germany. She previously worked at the Institute for Employment Research in Nuremberg. She received her PhD in Economics from the Friedrich-Alexander-University Erlangen-Nuremberg in 2011. Her research interests are in the fields of social policy, household divisions of labour, care of the elderly and intergenerational affairs.