Conference Schedule

The titles and authors of all paper presentations are listed below in the program overview. The respective abstracts can be accessed by clicking on the sessions

Keynote Lecture

Thursday | November 9, 2023 | 05:00 pm - 05:00 pm (CET)

• The Benefits of Linked Data for Labor Market and Educational Research
  Bernd Fitzenberger (IAB Nuremberg / FAU Erlangen-Nürnberg)
  ≡ Abstract

  Education and work are core human activities that are essential for economic prosperity and social development. To foster empirical research and evidence-based policy advice, researchers and policymakers need high-quality up-to-date data. However, life course data linking education, training, and labor market outcomes are still scarce in Germany compared to some other European countries. The lack of comprehensive linked data renders both monitoring and governing the education and training system difficult because the impact on labor market outcomes remains hard to establish.

  As one outstanding showcase of a linked dataset, the NEPS-ADIAB dataset combines survey data from the National Education Panel (NEPS) with administrative employment biographies from the Institute for Employment Research (IAB). NEPS-ADIAB can be used in particular to address research questions that require very precise wage data over the life course. However, a comprehensive linkage with administrative education data remains a challenge.

  Another showcase for the potential of linking administrative data and survey data at IAB is the evaluation of active labor market policies where the use of linked data is legally possible. Linking survey data and administrative data often requires the consent of the survey participants. This can be a role model for other fields.

  The keynote address by Prof. Fitzenberger, Director of the IAB, will elaborate the potential of linked data for labor market and educational research.

back to top

Session 1 (Chair: Nadine Bachbauer, LIfBi/IAB)

Thursday | November 9, 2023 | 12:45 am - 02:00 pm (CET)

• Intertemporal Labour Market Participation of Married Women in Germany
  Saikat Ghosh (LIfBi Bamberg)
  Christian Aßmann (LIfBi Bamberg)
  ≡ Abstract

  Female labour supply is an important part of a country’s resources and remains the focus of policy debates. In developed countries, population ageing will lead to a significant slowdown in labour force growth, which leads to challenges to the sustainability of economic growth. One of the key countermeasures could be to increase the participation of groups that are currently underrepresented in the labour market. One such is the labour market participation of women in these countries. However, studies in developed countries including Germany are sparsely available on women's intertemporal labour market participation.
  A predominant characteristic of annual participation behaviour is the high persistence in individual participation decisions (Heckman, 1978, 1981). Persistence occurs mainly due to (i) State dependency and (ii) Individual heterogeneity. Several sources of state dependence have been considered in the literature. Such as intertemporal non-separable preference for leisure (Hotz et al. 1988), human capital accumulation (Heckman, 1981), and search costs which differ across states (Eckstein & Wolpin, 1990). Factors such as fertility decisions, non-labour incomes, and state dependencies affected women’s labour market participation (Hyslop, 1999; Heckman, 1978). Being able to distinguish among them is important because they have different implications for labour market policies (Croda et al., 2011).
  This longitudinal study uses the data from NEPS-ADIAB -SC1 for a period 2012/13 to 2019 to understand women’s labour market participation in Germany. The sample consists of about 1500 married women, born between 1962 and 1995, having at least one child born in 2012/13. Different static and dynamic econometric models are used to identifying different sources of various in women’s labour market participation. The main focus was on the relationship between women's labour market participation and their fertility decisions.
  The initial findings from both static (e.g. linear probability, probit with random effect) and dynamic (dynamic panel using Arellano-Bond technique) estimations show state dependency where women’s previous labour market participation significantly positively affects the current labour market participation. Besides, the number of children in the household negatively affects the probability of labour market participation. In addition, if a woman had no child before 2012/13 (i.e. the target child was the first child) then it significantly increases the probability of returning to the labour market. Therefore, this paper presents a framework to estimate the intertemporal labour supply behaviour of women in Germany. In all specifications, we find strong state dependency and substantial effect for fertility variables.



• The Pivotal Factors? How Mothers’ Pre-Birth Employment Characteristics Shape Child Development
  Mathias Huebener (BiB Wiesbaden, IZA Bonn, DIW Berlin)
  Malin Mahlbacher (BiB Wiesbaden)
  Susanne Schmid (University of Oldenburg / University of Bamberg)
  Gundula Zoch (University of Oldenburg / LIfBi Bamberg)
  ≡ Abstract

  Prior to becoming mothers, women now spend more time in the labor market and in professional activities. This trend continues after childbirth, resulting in a significant shift in family life and early childhood environments. While a large body of literature analyzes the consequences of maternal employment on child development, limited attention has been given to the role of pre-birth labor market experience and job characteristics. Nonetheless, both factors play a crucial role in resource accumulation, women’s re-entry into the labor market, and their capacity to reconcile work and family life after birth. Therefore, pre-birth employment is likely to affect child development through several channels. First, employment can directly affect the fetus, as higher parental resources are likely to improve prenatal care and uterine conditions. Conversely, higher work-related stress may have negative effects on fetal development. These initial effects could potentially have lasting consequences on children’s later development through dynamic complementarity. Additionally, pre-birth employment can also affect later child development by shaping re-entry and post-birth employment conditions, or by affecting labor market experiences linked to parenting styles (e.g., managerial, interactive or manual tasks). However, evidence regarding the role of pre-birth employment (characteristics) in child development remains scarce.
  This study aims to fill this research gap by utilizing rich panel data on child development from the German National Education Panel Study, linked to administrative data on mothers’ employment histories (NEPS-SC1-ADIAB7521, 2012-2021; NEPS-Netzwerk et al. 2023). We analyze whether maternal work experience, employment conditions, and workplace characteristics before giving birth are related to child development over time. We examine cognitive and non-cognitive measures of child development, measured by mathematical literacy and the SDQ, through the first ten years of children’s lives. To understand potential mechanisms, we further investigate whether pre-birth employment conditions are related to re-entry, parenting behavior, well-being, and other mediators following childbirth.
  Preliminary OLS results suggest that higher pre-birth wages are advantageous for cognitive child development up to age ten, whereas precarious aspects, such as fixed-term employment correlate with more behavioral problems. However, accumulated work experience prior to birth is not related to cognitive or non-cognitive child development. Future analyses will also include firm characteristics and investigate the potential mediators contributing to the observed patterns.



• Paying Mothers for Caring: Labor Market Outcomes and Child Development
  Arnim Seidlitz (IAB Nuremberg, HU Berlin, Berlin School of Economics)
  ≡ Abstract

  In the this article, I analyze the German system of cash-for-care transfers, which includes both federal and state programs (Bundesbetreuungsgeld and Landeserziehungsgeld). The federal government implemented a monthly payment of 150 e for families with a child aged one or two, conditional on not enrolling the child to institutional care. Concurrently, four out of the 16 federal states had their own cash-for-care programs, offering higher benefits for a shorter duration.
  In my study, I aim to build upon the existing literature in several ways. Firstly, I evaluate multiple German cash-for-care policies within a single study, which is crucial as they target the same population and employ similar restrictions. Secondly, I investigate a range of outcome variables, including fertility, working hours and most importantly, early skill development and maternal investments. Thirdly, in the discussion, I aim to shed light on the mechanisms underlying these effects, exploring potential channels through which the impacts may be explained.
  Using German Microcensus data, my analysis reveals that mothers with a migrant background experience substantial and statistically significant reductions in employment when there is an increase in the available benefit amount, whereas the average population remains unaffected. There is additionally a significant increase in further fertility.
  To provide a more comprehensive understanding, I utilize data from the National Educational Panel Study (NEPS), which offers valuable insights into child-care choices, parental time allocation with the child and skill test data in subjects such as vocabulary, math and science for children aged three and above. The survey data suggest a delay in the enrollment of children in childcare and certain adjustments in maternal behavior. Specifically, mothers increase their time invested in practicing counting with their child. However, there are no statistically significant effects observed on early skill development.
  The observed effects appear to be partly attributable to changes in childcare choices and associated behavioral adjustments. However, there are other aspects of the effects that cannot be explained by changes in care decisions. One plausible explanation for these effects is the increase in disposable income resulting from the available benefits.

back to top

Session 2 (Chair: Manfred Antoni, IAB)

Thursday | November 9, 2023 | 02:20 pm - 03:35 pm (CET)

• Differences in Public Pension Entitlements within Couples - Analyses based on SHARE-RV
  Imke Herold (MEA-SHARE Munich)
  ≡ Abstract

  Old-age provision remains an important issue for western countries due to increasing life expectancy and longer retirement durations. Analyzing old-age provisions in Germany is challenging, because the system is separated into three pillars and data that covers all of them is scarce. In general, wealth and income analyses are typically based on a household level since household members are assumed of sharing resources. However, in terms of a gender perspective the individual allocation of pension rights is just as important.
  Literature shows that women collect fewer pension entitlements in the German Pension Insurance that men. Although, women also acquire pension entitlements for child-raising, a gender gap remains. Most studies focusing on the distribution of wealth within couples examine parameters collected in surveys like income, financial assets, life insurances, or building property. That is due to the fact that most surveys do not capture public pension entitlements.
  The linked data set SHARE-RV provides valuable research data by directly linking the German sample of the Survey of Health, Ageing and Retirement in Europe (SHARE) to selected administrative data of the German Pension Insurance (RV). It offers exact data from the pension accounts of the SHARE-respondents as well as important context information, household information, and information on private pension arrangements.
  Using SHARE-RV data, I want to contribute to the question of how pension entitlements are distributed within couples aged 50 years and older and what factors influence this distribution. Furthermore, I want to examine if couples with a large entitlement gap tend to compensate this with additional private pension arrangements, especially for the partner with fewer entitlements. Linked SHARE-RV data is especially useful to answer those questions, because it provides information on couples’ living conditions from the survey data as well as detailed information on collected earning points from the administrative data.



• Levels, Distributions, and Drivers of Lifetime Earnings
  Mattis Beckmannshagen (DIW Berlin/SOEP)
  Timm Bönke (DIW Berlin)
  Martina Kaplanová (DIW Berlin/SOEP)
  Carsten Schröder (DIW Berlin/SOEP, FU Berlin)
  Yogam Tchokni (DIW Berlin/SOEP)
  ≡ Abstract

  Based on the unique structure of SOEP-RV, a novel dataset comprising linked administrative and survey data (Lüthen et al. (2021)), this paper provides estimates of the full distribution of lifetime earnings in Germany for people born between 1935 and 1964. The main contribution of the paper is that we are able to compute full lifetime earnings for a sample representative for the whole West German population. The linked data combine administrative data for earnings from dependent employment, with earnings of civil servants and self-employed based on the survey data from the Socio-Economic Panel (SOEP, see Goebel et al. (2019)) allowing us to track full earnings biographies. Existing research on lifetime earnings based on administrative data could not capture civil servants or self-employed and thus had to rely on extensive sample restrictions regarding labor market participation (see e.g. Bönke et al. (2015)).
  Similar to findings in the US (Guvenen et al. 2022) and France (Garbiniti et al. 2023) we find that real median lifetime earnings have stagnated across birth cohorts. This is a result of slightly decreasing lifetime earnings for mean born after 1940, while lifetime earnings of women have steadily increased from birth cohort 1935 to birth cohort 1960. Despite this converging trend for men’s and women’s lifetime earnings, women’s median lifetime earnings remain substantially lower than men’s. The increase in female lifetime earnings over time can likely be attributed to increasing labor market participation among women. While women born between 1935 and 1939 on average were employed for 16 years of their life, for those born between 1955 and 1959 the lifetime labor market participation was at 22 years. Inequality follows a u-shaped pattern trend along birth cohorts. Gini and Theil indices are lowest for the individuals born between 1945 and 1949 and are increasing for later birth cohorts. A decomposition of the Theil index reveals that this development occurs despite decreasing inequality between men and women. However, within-group inequality among men is increasing substantially which is the main driver for overall inequality increasing for later born cohorts.
  In the remainder of the paper, we take advantage of the detailed socio-economic characteristics included in the SOEP and discover main determinants of one’s position in the lifetime earnings distribution by applying machine learning methods. This allows us to assess to what degree variables like parental background or education determine lifetime earnings over time.



• Educational Upgrading and Social Inequality – Analyses based on NEPS-SC6-ADIAB
  Anja Grauenhorst (University of Bamberg)
  Steffen Schindler (University of Bamberg)
  ≡ Abstract

  Lifelong learning has gained importance during the last decades. It could be a solution to meet the challenges of our time such as the outage of skilled labour in some industries or a changed need for qualifications due to structural and technical change. However, it might also add a new dimension to questions of social stratification and mobility. Considering that education serves as the key mediator of the association between social origins and destinations, lifelong learning adds a dynamic element to social mobility as it alters the association between social origins and education.
  This paper focuses on a specific part of lifelong learning: the upgrading of formal educational degrees after labour market entry. We want to find out whether and in which way formal educational upgrading activities after labour market entry change the strength of the association between social origins and earnings.
  Our paper replicates the study by Virdia and Schindler (2019), who found that formal educational upgrading does not alter the strength of the association between social origins and social status (measured by the International Socio-Economic Index of Occupational Status, ISEI). Following a suggestion by the authors, we vary the labour market outcome and consider earnings instead of ISEI. While effects of educational upgrading on ISEI exclusively reflect occupational changes, earnings capture within‐job effects in addition. Hence, by considering wage, we expected to arrive at conclusions about the influence of educational upgrading that are different from those of Virdia and Schindler. 0ur first results suggest that formal educational upgrading is a weaker mediator for the relationship between social origins and wage than initial degree. If degree is included in the form of educational attainment at labour market entry, this mediates more strongly the effect of social origin on the child's later wage over the life course. Our results so far are of a descriptive nature and based on the representation of developments by growth curve models, among others.
  The basis of our analyses is the linked data product NEPS-SC6-ADIAB consisting of survey data from the German National Educational Panel Study (NEPS) and administrative data from social security. The part of NEPS containing the start cohort adults (start cohort 6), includes retrospective life course data about individuals in Germany. The part with the administrative data of NEPS-SC6-ADIAB stem from social insurance notifications and is part of the Integrated Employment Biographies (IEB) of the Institute for Employment Research (IAB) (Bachbauer/Wolf 2022).

back to top

Session 3 (Chair: Clara Wolf, LIfBi/IAB)

Thursday | November 9, 2023 | 03:55 pm - 04:45 pm (CET)

• Has my Employer Noticed my Promotion? How Employers and Individuals Report the Requirement Level in the German Classification of Occupations 2010
  Ann-Christin Bächmann (LIfBi Bamberg, IAB Nuremberg)
  Basha Vicari (IAB Nuremberg)
  ≡ Abstract

  The German Classification of Occupations 2010 (KldB 2010) does not only provide information on occupational specialization, but also differentiates occupational units according to their requirement level into four groups ranging from (1) unskilled or semi-skilled activities to (4) highly complex activities (Paulus & Matthes 2013). The requirement level (5th digit of the KldB 2010 code) thus offers a new analytical potential to observe individual career advancement and hence address crucial issues of social inequality, e.g., by analyzing possible gender differences in career trajectories. To date, however, it is unclear whether the measure of the requirement level and possible changes over time are reliably reported. Depending on how occupational information is collected, different types of measurement errors may occur. For instance, survey data can be inaccurate and prone to coding errors. Administrative data can also be subject to errors, e.g., employer notifications might under-report changes in the annual occupational coding routine, including changes in the requirement level.
  In our paper, we analyze possible differences in occupational information and, in particular, the requirement level, on the basis of respondent and employer reports. For this purpose, we use the NEPS-SC6-ADIAB data that links survey data of the Adult Cohort of the German National Educational Panels Study with administrative data from the Integrated Employment Biographies (IEB) of the Institute for Employment Research.
  First comparative analyses suggest central differences in the reported requirement levels of the two data sources, with requirement levels tending to be higher in survey data. However, these differences show up only for men. This finding raises a puzzle: It remains unclear whether men either systematically over-report their requirement levels in surveys or employers systematically underreport occupational promotions and men are more likely to be promoted than women. To resolve this puzzle, we next examine the reporting differences in more detail, including relevant control variables, such as employer change. In a final step, we demonstrate how the differences in reporting affect substantive analyses of gender differences in promotion processes.



• Consenter Selectivity and Cross-Validation in the SOEP-CMI-ADIAB Sample
  Manfred Antoni (IAB Nuremberg)
  Mattis Beckmannshagen (DIW Berlin/SOEP)
  Markus M. Grabka (DIW Berlin)
  Sekou Keita (IAB Nuremberg)
  Parvati Trübswetter (IAB Nuremberg)
  ≡ Abstract

  As part of the SOEP-CMI-ADIAB record linkage project, data from the Socio-Economic Panel collected in interviews were linked to administrative data from the IAB. This study examines the quality of the successfully linked data in terms of selectivity. In addition, it identifies steps that can be taken as best practices in handling the data to minimize potential problems resulting from measurement inaccuracies or selectivity of the sample.
  1. Selectivity analyses: Selectivity and thus lack of representativeness of the SOEP-CMI-ADIAB sample can arise at three levels.

  • At the level of the question on consent: not all respondents of the Socio-Economic Panel were asked the question on consent. Were respondents with certain characteristics not asked?
  • At the level of consent: Were respondents with certain characteristics more likely to give consent to data linkage than others?
  • At the level of data linkage: Does the probability of success of data linkage vary based on certain characteristics of the respondents?

  Logistic regressions within each relevant population are used to test whether certain characteristics had significant influence at any of the three levels. Subsequently, hints are given on how to compensate for potential selectivity by creating weighting factors that take all levels into account.
  2. Cross-validation of gross compensation and part-time/full-time variable Because of the linked data set, two sources of information exist for many respondents on the same (or at least very similar) questions. For example, the administrative data contain information on employees' gross daily pay, while the SOEP asks about gross monthly pay. At the same time, the administrative data contain information on whether employees work full or part time, while the SOEP survey data contain actual as well as contractually agreed weekly working hours.
  On this basis, cross-validations can be performed. Does the SOEP information agree (approximately) with the administrative data? Are there systematic deviations and, if so, how can these possibly be explained?

   

back to top

Session 4 (Chair: Corinna Kleinert, LIfBi)

Friday | November 10, 2023 | 09:00 am - 10:15 am (CET)

• Gender Gaps in Early Lifetime Income and Education
  Tobias Brändle (Pforzheim University, IAW Tuebingen)
  ≡ Abstract

  Research Question: Gender gaps in earnings are well documented and remain high in Germany, even though various measures have been put in place to reduce them (see, e.g. Brütt and Yuan, 2022; or Seitz and Sinha, 2022). It has also become more evident that the unconditional difference in lifetime income between men and women is even larger, at around 40%, compared to 21% in contemporary earned income (the so-called gender pay gap, see Brändle et al. 2019, 2022). At the same time, however, the gender-specific educational returns of different educational paths are similar there are no clear differences in the relative returns to education (Zühlke et al. 2020). Career breaks, such as the birth of a child, have been identified as a key factor influencing income and especially lifetime income differences between genders (Steffens et al., 2019, Anderson et al., 2002, Budig and England, 2001). However, the question is whether differences in incomes are already manifesting between genders before childbirth. For this purpose, this study analyses the gender-specific differences in early lifetime earnings and the role of different educational paths. The aim is to examine the role that career choices in early life determine later differences in lifetime income between genders. Since the studies so far have mainly examined contemporary income when analysing the gender earnings gap, periods without income from work are often ignored. Likewise, income before childbirth often cannot be observed. Therefore, new insights can be expected from the analysis of full career paths and lifetime earnings.
  Data and Method: The study uses the linking of the starting cohort 6 of the National Educational Panel Study and the administrative data of the IAB (NEPS-SC6-ADIAB; for detailed information on the data see Antoni et al., 2018). This data has the advantage that both the educational path and other important control variables (such as parental background, grades, etc.) can be mapped using the NEPS survey data, and precise income information about the entire working life can be mapped using the administrative data from the IAB. To estimate the effect of early career choices on later lifetime earnings, we compare individuals with different educational paths before childbirth. We control for observable socio-economic characteristics and educational background. We differentiate between individuals pursuing vocational training or a university degree.
  Results: It becomes clear that the lifetime earnings of men and women differ significantly and that different educational paths and career breaks play an important role. Career breaks have a greater effect on lifetime earnings when entering the workforce late. Thus, people with higher education are more affected, but at the same time they try to avoid career breaks; they occur less frequently in them, especially unemployment and further education. Women with lower education take parental leave much more often and much earlier than women with higher education. The differences can be observed in particular up to the age of 30. Men have higher early lifetime incomes than women and this difference becomes even greater when children are born into the household. Educational returns are negative for early lifetime income and they tend to be lower for women, particularly when considering the first educational decision and not the highest educational attainment. The results of the study also show that there are large returns to education in lifetime income and that the consideration of cumulative income can provide important insights for educational decisions. The analyses also show that linked income and survey data offer great advantages in terms of the variety and quality of analysis options. The observation of individual educational paths, employment status and reliable income information offer a sound basis for the analyses and further research.



• Educational Expansion and the Quality of University Graduates: Evidence from Germany
  Miriam Sturm (University of Tuebingen)
  ≡ Abstract

  In recent decades, many industrialised countries have witnessed considerable increases in the shares of individuals participating in tertiary education. This has given rise a discussion about the quality effect of selection into higher education: Assuming a time-constant distribution of ability, it seems natural to expect that educational expansion draws increasingly lower academic abilities into universities. Empirical evidence on this hypothesis is scarce and hard to establish since ability is inherently unobserved. However, a decreasing average quality of college graduates should be reflected in a depressed college premium. This rationale was used by a small number of studies on the US and UK, but the results are not unanimous.
  Germany constitutes no exception from the trend towards an increasing academisation of the workforce: College enrolment rates have steadily increased over the past decades, which is particularly striking given the popularity of the dual vocational training system. Still there is no empirical evidence systematically analysing how the observed shift in educational degrees is related to potential shifts in abilities. Inspired by Carneiro & Lee (2011), the current study suggests to not only use changes in educational shares over time, but also geographic variation: Exploiting the federal organisation of the German education system, I compare the college premia received by graduates of a similar age working on the same labour market who, however, acquired their general qualifiation for higher education (the so-called Abitur) in different federal states with different Abitur shares. Following the argument above, regional differences in Abitur shares can be thought to reflect differences in average ability, which should ultimately result in lower college premia for individuals from states with high Abitur shares. Given the lack of a standardised school leaving test such as the SATs, it is particularly appealing in the German case to identify potential quality effects based on regional variation. Moreover, shifts in the shares of highly educated individuals should reflect shifts in average ability more accurately due to the low cost of education compared to the US, which the literature has focused on so far.
  The credibility of the aforementioned approach hinges on the use of high-quality wage data combined with detailed information on when and where individuals were educated. The NEPS-SC6-ADIAB combined data base of administrative and rich survey data provides both and is therefore perfectly suited for the given setting. In line with the results by Carneiro/Lee, preliminary evidence suggests lower college premia for graduates from regions with higher Abitur shares, suggesting lower average ability levels for these individuals.



• Dual Qualifiers with a Vocational Training and a Tertiary Degree: A Cohort Perspective on Their Intragenerational Income Trajectories, 1944-1984
  Jessica Ordemann (DZHW Hannover)
  ≡ Abstract

  Starting in the 1980th, a growing number of freshmen entered higher education with a previous vocational education and training (VET) degree (Lewin et al., 1996). As dually qualified graduates their number declined but still, in 2017, 19% of all graduates left higher education with a dual qualification of a VET- and a tertiary degree (Dahm & Peter, 2023). Despite the decline, the decision of dual qualifier to upskill their education at a later time in their life gave them the potential to correct their educational decisions as a ‘second chance’ (Bernardi 2012). Mostly, the upskilling within the individuals life course is accompanied by a raise in income that remains above that of direct qualifiers who do not have a vocational qualification (Ordemann, 2023). However, little is known about how the monetary returns on education have changed over time. I therefore ask: How do the income trajectories of dual qualifiers change over time and how do they compare to the trajectories of direct qualifiers. Relative education theory suggests that early cohorts of dual qualifiers should have earned less than later cohorts. Compared to direct qualifiers human capital theory suggests that they should have an advantage due to their dual qualification. This advantage should decrease over time in the labor market and should be stronger in earlier cohorts as specific human capital gained in VET should decrease in importance.
  Using the linked NEPS-SC6-ADIAB data, I shall present fixed-effects panel estimations about the income trajectories of four birthcohort of dual qualifiers (1944-1953 , 1954-1963, 1964-1973 and 1974-1983), separate for women and men, and compare them to direct qualifiers of the same cohorts.

back to top

Session 5 (Chair: Daniel Fuß, LIfBi)

Friday | November 10, 2023 | 10:45 am - 12:25 am (CET)

• On-the-Job-Training under Imperfect Competition: The Additional Burden of Working for a Low-Pay Firm
  Theresa Markefke (University of Cologne)
  ≡ Abstract

  On-the-job-training is important for firms, workers and the economy. A skilled labor force is a crucial ingredient for economic growth. This is mirrored not only by economists’ consideration of the importance of human capital in their modeling, but also by recurring public debate about the ways the skills of workers could be improved.
  Whether or not firms are willing to provide such training critically depends on the structure of the labor market as has been demonstrated in the theoretical work of Becker (1964) and Acemoglu and Pischke (1998) and Acemoglu and Pischke (1999). By now there is robust evidence that firms pay differently high wages to equally skilled workers (Abowd et al., 1999; Card et al., 2013; Card et al., 2018) and that these differences are important for wage inequality. In this paper, I study the implications of firm pay heterogeneity for the provision of on-the-job training and provide new insights into the relative incentives of firms operating in an imperfectly competitive labor market to invest in the skills of their workforce.
  To this end, I derive empirical hypotheses based on the strategic wage posting models by Manning (2003) and Fu (2011) . I then use two high quality datasets for the German labor market to test these hypotheses: the Linked employer-employee dataset (LIAB) and the linked worker-administrative dataset (NEPS-SC6-ADIAB). Using a number of different cross-sectional models, I analyze first how firms’ training activities differ depending on firms’ position in the wage distribution, and second, how this interacts with individual workers’ training activity and wages.
  Consistent with theoretical predictions, I find that higher-paying firms provide more on-the-job-training and workers employed at those firms receive more training. Wages increase after training, but the correlation is not always statistically significant. My findings imply that workers matched with a lower-paying firm will face worse career and income opportunities than workers matched with higher-paying firms due to lower on-the-job-training provided by the firm. This has immediate relevance for policymakers designing instruments to encourage employer-provided training.



• The Impact of Cognitive and Non-cognitive Skills on Selection in Training Firms, Training Completion, and Post-training Earnings Trajectories and the Important Role of the Firm
  Mahdi Gholami (LMU Munich)
  Samuel Mühlemann (LMU Munich, IZA Bonn, ROA Maastricht)
  ≡ Abstract

  This paper investigates the impact of cognitive and non-cognitive skills of ninth graders on selection in training firms, completion of dual apprenticeship training, and post-training earnings trajectories. Previous research has shown that these skills are crucial for success in the labor market, but little attention has been given to the role of the firm in shaping earnings variance. To address this gap, we propose a comprehensive examination of the relationship between cognitive and non-cognitive skills and earnings by incorporating firm characteristics (such as firm size, median wage in the firm, and unobserved time-invariant heterogeneity). Using data from the National Educational Panel Study (NEPS) and administrative records from the Institute for Employment Research (IAB), we analyze a sample of German students who were surveyed in ninth grade in 2011, their initial selection in training firms, completion of dual apprenticeship training and subsequent post-training earnings until 2019. Our study includes various competencies, such as math, reading, and science literacy, alongside non-cognitive skills, specifically the Big Five personality traits. Our findings indicate that both cognitive and non-cognitive skills significantly influence the selection process in apprenticeship training, with better skills leading to placement in higher-paying companies. Notably, large firms seem to prioritize math grades and reading competencies when selecting apprentices. Additionally, the initial wages of apprenticeship graduates are positively associated with their math competencies, as measured in 9th grade, for apprentices who leave the training firm after completing their apprenticeship. However, we find that once we account for firm characteristics, the association between cognitive and non-cognitive skills and post-training wages is no longer statistically significant. We interpret our findings that math competencies are an important signal in the market for apprentices but also in the market for skilled labor when employees do not yet have substantial labor market experience and when firms could not yet acquire relevant information about an individual’s productivity.



• Continuing Training Aspirations of Young Workers in the Context of Companies' Training Opportunities
  Martin Ehlert (WZB Berlin, FU Berlin)
  Mortimer Schlieker (BIBB Bonn)
  Matthias Siembab (BIBB Bonn)
  ≡ Abstract

  This study investigates the influence of previous educational experiences and company training opportunities on the continuing training aspirations of young workers in Germany. Continuing training has become increasingly important for young workers, as early vocational development plays a crucial role in shaping career paths. However, participation in training is unequally distributed. To learn more about the mechanisms behind this unequal distribution, which emerge from the educational trajectories of young workers, we pose two research questions: 1. To what extent do previous educational experiences influence aspirations to participate in continuing training? 2. Do company characteristics moderate the effect of previous educational experiences on continuing training aspirations?
  Theoretically, we draw on transaction cost theory with regard to individual and company utility expectations of non-formal training. Furthermore, we incorporate the value-expectation theory (Eccles et al. 1983) into our theoretical framework to analyse the longitudinal relationship between previous educational experiences and the expectations and values assigned to education. We hypothesize that positive educational experiences increase the perceived value of education, which in turn leads to stronger aspirations for non-formal training. Moreover, we expect that promoting company structures and opportunities (e.g. the qualification structure and financial support) mitigate the effect of negative previous educational experiences.
  We use data from the Starting Cohort 4 of the National Educational Panel Study (NEPS), which focuses on young adults in their first years of employment. We add company-level data from the IAB Establishment History Panel via the NEPS-ADIAB linked data. Preliminary findings based on analyses without the data linkage suggest that positive educational experiences (operationalised by educational degree) and companies’ training opportunities are strongly associated with young workers' aspirations for non-formal training. Financial support is particularly important for those with high educational degrees, suggesting a potential Matthew effect. However, certain company opportunities also seem to be beneficial for young workers with low educational degrees.
  We plan to extend the analyses in several ways. First, we will use a more fine-grained operationalisation of previous educational experiences. Second, we plan to explore the moderating effects of company factors in more detail using linked NEPS-ADIAB data. This data will provide additional insights into the qualification structure within companies, which is expected to further clarify the interplay between individual characteristics and company-specific factors. For example, larger group sizes of low-skilled workers could reduce companies’ transaction costs and increase workers’ motivation to participate in continuing training.



• Reference-Group Effects in Learning Contexts of Vocational Education and Training (VET)
  Nele Theuer (BIBB Bonn)
  Dominik Becker (BIBB Bonn)
  Katarina Weßling (BIBB Bonn, ROA Maastricht)
  ≡ Abstract

  Reference-group effects are well-established within educational psychology. According to the Big-Fish-Little-Pond-Effect (BFPLE) hypothesis, achievement-based school/class composition influences individuals’ academic self-evaluation. When controlling for individual achievement, higher (lower) group achievement is related to lower (higher) individual academic self-concept (Fang et al., 2018; Marsh et al., 2007; Trautwein et al., 2009). The BFLPE has also been found to influence educational/occupational aspirations and attainment (Göllner et al., 2018; Nagengast & Marsh, 2012).
  However, the BFLPE remains understudied in the context of VET. Since VET students allocate their time between two learning environments – VET schools and training firms –, existing insights from primary and secondary school environments cannot be generalised to VET. More precisely, it remains unclear if and to what extent reference-group effects such as the BFLPE exist in VET schools and/or firms, and whether potential effects may even interrelate across these learning environments.
  Hence, the aim of this study is to explore whether reference groups influence occupational aspirations in VET contexts by a) testing for these effects in both VET learning environments and b) advancing knowledge on potential moderators. Concretely, we hypothesize that the number of fellow VET students in firms might influence the strength of the BFLPE in both VET schools and firms (Thijs et al., 2010; Zell & Alicke, 2009). Moreover, we expect the BFLPE to be affected by students’ relative time of exposure to each learning context (Jansen et al., 2022), as measured by VET type (dual versus school-based).
  To test our hypotheses, we use data from NEPS-SC3 (Blossfeld & Roßbach, 2019). In a first step, we measure VET students’ self-evaluated achievement relative to group achievement as our predictor variable. According to Thijs et al. (2010), these subjective evaluations of relative achievement mediate objective group achievement effects. However, to best understand possible reference-group effects on occupational aspirations, we also aim to include objective measures.
  To that end, we will explore ways to link NEPS data to administrative information on VET schools and firms. The NEPS-ADIAB dataset provides objective information on reference groups in VET firms (Bachbauer & Wolf, 2022), whereas the Federal Statistical Office offers district-level data on VET schools (Destatis, 2022). We aim to test the feasibility of different data linkages to NEPS data, and to emphasize the validity of our subjective measure of relative achievements by demonstrating its considerable correlation with objective group achievement in secondary school.

back to top