Keywords

1 Introduction

The knowledge and skills that individuals acquire are a major source of economic growth (Hanushek & Kimko, 2000). Faced with the need to adapt to new technological or organizational change, further training has gained in importance in the last decades and will likely continue to do so. Indeed, in terms of the number of participants, the training sector is already the largest education sector in Germany. Whereas there is a large literature analysing the characteristics of training participants (e.g. Fitzenberger & Muehler, 2015; Kaufmann & Widany, 2013; Kruppe & Baumann, 2019; Pischke, 2001) and the determinants of firms that provide training to their workforce (e.g. Gerlach & Jirjahn, 2001; Görlitz, 2010; Käpplinger, 2007; Zwick, 2004), there are few studies investigating regional determinants of training. Bellmann et al. (2011) and Martin et al. (2015), for example, have performed comprehensive analyses showing how a variety of regional economic factors predict regional training activities, and there are two studies investigating whether a firm’s provision of training depends on the regional population density (Brunello & de Paola, 2008; Brunello & Gambarotto, 2007). In addition, Martin et al. (2016) analyse the effect on continuous training participation of variations in community administrative practices on the local level.

This lack of evidence is surprising given that both Rzepka and Tamm (2016) and Görlitz and Rzepka (2017) show that there is a large extent of variation in the share of individual training participants on the regional level in Germany. While there are some regions with a share of training participants fewer than 25 percent, others have almost twice as many participants. The probable reason why regional factors have been largely neglected so far is that many data sets that include training information do not additionally include detailed geographical identifiers that would allow a merging of regional factors. One exception is the data from the National Educational Panel Study (NEPS): Starting Cohort 6—Adults (Adult Education and Lifelong Learning) that contain detailed training information on the individual level and offers the opportunity to assess geographical identifiers of these individuals on the regional level. As part of SPP 1646 funded by the DFG, two studies were published by Rzepka and Tamm (2016) and Görlitz and Rzepka (2017) that exploit this unique opportunity to shed light on regional determinants of training participation. This article summarizes the results of these studies. Rzepka and Tamm (2016) analyse the industry-specific local employer density as a predictor of individuals’ training incidence and training intensity. Local employer competition might affect training, because firms often finance the training of their employees. However, if they have to fear that the employee trained by them will be hired by a competing firm just after completing training, their willingness to finance it will be reduced. Thus, for firms, the higher the level of competition, the lower the incentives to finance training. Görlitz and Rzepka (2017) investigate whether training supply—that is, the number of firms providing courses and seminars to the public—affects individuals’ training participation. Both studies focus on training of employed individuals, which comprises any formal and non-formal training with a professional interest.

The remainder of this article is set out as follows. The next section describes the data and shows how regional factors can be merged to the NEPS data. Sections 15.3 and 15.4 summarize the analyses investigating either local employer competition (Rzepka & Tamm, 2016) or training supply (Görlitz & Rzepka, 2017) as determinants of training. The last section concludes the article and emphasizes the research potential of the NEPS data.

2 Data and Opportunities to Merge Regional Characteristics

Rzepka and Tamm (2016) and Görlitz and Rzepka (2017) use data from NEPS: Starting Cohort 6—Adults (Adult Education and Lifelong Learning), doi:10.5157/NEPS:SC 6:1.0.0.Footnote 1 NEPS data collection is part of the Framework Programme for the Promotion of Empirical Educational Research funded by the German Federal Ministry of Education and Research and supported by the federal states.

In the first wave in 2009/10, the data contain survey information on, for example, educational and training biographies, social demographics, and labour market outcomes from more than 11,000 individuals who were born between 1944 and 1986 (Blossfeld et al., 2011). Data include the county in which each survey participant works. This information can be used to merge regional information to the data. When analysing labour market processes, one can merge data on the county level or any higher level. Both Rzepka and Tamm (2016) and Görlitz and Rzepka (2017) aggregate the more than 400 counties to 140 local labour markets following Kosfeld and Werner (2012). The local labour markets aggregate counties that are characterized by commuter links. Counties with strong commuter flows are combined to belong to one local labour market. This regional unit represents the relevant labour market for an individual, because the decision which employer to choose and when to switch employers is often determined by the travel time to work. Of course, travel time is influenced by the distance of counties, but also by infrastructure and public transport. The definition of local labour markets based on commuter links captures both aspects.

NEPS data are well suited to analyse training processes because they cover a variety of survey questions addressing an individual’s training participation. Rzepka and Tamm (2016) and Görlitz and Rzepka (2017) define training incidence as employees’ participation in any formal and non-formal training with a professional interest (e.g. seminars and courses) within 12 months prior to the survey interview. Additionally, Rzepka and Tamm (2016) also analyse total training in hours. This aggregates the number of hours spent in training seminars or courses in the last 12 months. NEPS data also allow controlling for the ‘usual’ determinants of training on the individual level: gender, migration background, age, education, and occupation. Job-related characteristics such as tenure, temporary contract, or part-time contract and factors of the firm such as firm size and industry are available as well.

3 The Impact of Local Employer Competition on Training Incidence

According to the human capital theory (Becker, 1964), training is an investment that is made when training returns exceed the corresponding costs. Becker distinguishes between two types of training that determine whether firms or workers bear the training costs: firms will bear the costs of training that is specific to the firm, whereas employees have incentives to invest in general training that is transferable when changing jobs. The empirical literature has provided evidence conflicting with these predictions, because it shows that employers contribute substantially to general training (Loewenstein & Spletzer, 1999). Based on these findings, a new strand of training literature emerged in economics—the ‘new training literature’—in which it is assumed that labour markets are imperfect (Acemoglu & Pischke, 1998). These theories predict that employers will invest in general training if they have monopsony power to compress wages so that they can recoup their training investments by paying wages that are lower than workers’ labour productivity.

One source of monopsony power that induces firms to invest in training could be mobility constraints. If mobility costs for individuals are low—for instance, in regions with high industry-specific firm density—the risk that employees will quit after receiving employer-sponsored training and be re-employed by another firm (‘poaching risk’) will be higher. The poaching risk lowers firms’ incentives to invest in training (Gersbach & Schmutzler, 2012). Mühlemann and Wolter (2007) test this hypothesis for Switzerland and provide evidence that having a greater number of firms in a region that are in the same industry reduces the likelihood of a firm financing apprenticeship training. Similarly, Brunello and de Paola (2008) and Brunello and Gambarotto (2007) report that the population density (measured as the number of employees per square kilometre) is an important determinant of firms’ training decisions in Italy and the UK, respectively. Also, Picchio and van Ours (2011) find that a decrease in labour market frictions significantly reduces firms’ training expenditures.

Rzepka and Tamm (2016) investigate the effect of local employer competition on individual training participation in Germany. Their analysis uses data from Wave 2009/2010 of NEPS Starting Cohort 6 and defines individual training incidence and training duration as their dependent variables. Data on labour market competition come from the Establishment-History-Panel (BHP) provided by the Institute of Employment Research (IAB) and these are merged to the NEPS survey on the regional level of local labour markets. The authors used two measures of local labour market competition: (a) the density and (b) the concentration of firms in a particular NACE industryFootnote 2 per region. Density was measured by the number of firms per industry and region divided by the size of the region in square kilometres; and the concentration was calculated using the Herfindahl index. For both outcomes, they estimate separate regressions using each of the two measures of local labour market competition. Their model controls for sector- and region-specific fixed effects that absorb many sources of endogeneity and selection into local labour markets.

The analysis shows that employees are significantly less likely to participate in training, and their training duration decreases in response to an increase in local labour market competition. This supports the new training literature. In sum, it means that there may be negative externalities on training participation due to sectoral agglomeration. The authors also discuss potential remedies. For instance, firms could increasingly use payback clauses when financing further training. In practice, this means that employees would need to reimburse training costs if they move to another company before the investment is paid off.

4 Regional Training Supply as a Predictor of Individuals’ Training Participation

Görlitz and Rzepka (2017) explore the correlation between regional training supply and employees’ training participation. They measure training supply as the number of firms that offer training in the local labour market. The reason why training supply could affect participation is that training demand can be met only by training supply. If there is no training supply, individuals are constrained in their training decisions. Another reason is that training often takes place in classroom-type settings at locations distant from the workplace of employees, so that workers have to travel to the training supplier in order to participate in training. Even though this seems to be more important for self-financed training, it also matters for training financed by employers. Whereas large firms also provide training inhouse, many smaller firms co-finance training for courses that are given by an external provider. In general, travel costs have been shown to influence educational decisions. For example, the study by Card (1995) exploits differences in the proximity to college as instruments for educational attainment. With regard to training, Tuor and Backes-Gellner (2009) have shown that giving up free time for training plays a much more important role for non-participation in training than having to spend money for training.

The authors merge the survey data of Wave 2009/2010 of NEPS Starting Cohort 6 with information on training supply that comes from two distinct sources. First, they calculate the number of training suppliers based on their industry affiliation ‘Adult and other education n.e.c.’ (NACE Rev. 2, code 85.59) from the universe of all German establishments provided by the German federal employment agency (Statistik der Bundesagentur für Arbeit, 2013). This measure is referred to as ‘BA data’ in the following. Second, they use information on the number of firms supplying training in 2007 that is available in the wbmonitor data (Bundesinstitut für Berufsbildung und Deutsches Institut für Erwachsenenbildung, 2007). This information is called henceforth the ‘wbmonitor data’. Because the establishments are defined by having at least one employee who is covered by the social security system, and because many employees of training suppliers are not covered by social security, the number of training suppliers is smaller in the BA data compared to the wbmonitor data.

Görlitz and Rzepka (2017) regress training incidence on each of the two measures of training supply in separate regressions. To account for differences in the size of the local labour market, they do not analyse the number of training suppliers as an independent variable, but rather define the density of training suppliers by dividing the absolute number of training suppliers by the size of the respective local labour market in square kilometres. Results indicate that training participation is significantly higher in regions with many suppliers in the training market. Furthermore, they show that the density of training suppliers is not related linearly to training incidence. Instead, there is a statistically significant, concave relationship between the density and the incidence of training. Figure 15.1 illustrates this relationship.Footnote 3 An increasing number of training suppliers per square kilometre is initially associated with higher training levels. However, if the density of training suppliers is above average compared to other regions, a further increase in density is no longer associated with higher levels of training participation. The authors conclude that policymakers who follow the political aim of increasing training participation through subsidizing training supply should focus on regions with below-average training supply, because this is more likely to be effective than providing subsidies to all regions equally. However, they also note that a final policy conclusion requires a causal analysis of the relationship, and this remains a topic for future research.

Fig. 15.1
A line graph of the density of training suppliers versus average training participation. The dotted line represents the B A data. The dashed line represents the w b monitor data. The highest points are 11,0.41 and 15,0.39.

OLS prediction of the relationship between training participation and the density of training suppliers. Notes: The graph illustrates the predictions of the OLS regression results of Table 2 from Görlitz and Rzepka (2017). The constant of the regression is 0.29 in the BA data and 0.30 in the wbmonitor data

Full size image

5 Conclusion

This article presents the research potential of NEPS Starting Cohort 6 given by merging aggregate regional information on the county level. It also emphasizes the research potential of NEPS data by summarizing the analyses and the findings of two research studies that exploit it: Rzepka and Tamm (2016) and Görlitz and Rzepka (2017).