Determinants of further training – evidence for Germany

Abstract

We explore the relevance and development of further training in private sector firms using the German Socio-Economic Panel, which is a representative sample of German residents. We focus on formal training and explore possible individual and job-based determinants of its incidence. We also show changes over time during a 20-year observation period from 1989 to 2008. Most hypotheses are supported by the empirical evidence. Job status and firm size are the most relevant characteristics for training participation. Furthermore, our analyses reveal a general trend of rising training rates from 1989 to 2008, indicating an increased importance in the German labor market.

Keywords:

• further training
• GSOEP
• human capital
• panel data

Notes

1. In this paper, we use the terms ‘further training’, ‘continuous training’, ‘vocational training’, ‘occupational training’ and simply ‘training’ in a widely synonymous way.

2. Besides, there is some purely descriptive evidence (see Bellmann 2003 for an overview).

3. It is not useful to estimate fixed effects binary probit models because of inconsistent results (Cameron and Trivedi 2009, p. 601). Therefore, we tested the results of a binary logit fixed effects model and a binary logit random effects model against each other. For those persons with information in multiple years and variation in the dependent variable (4002 observations of 1378 persons), the result of the Hausman test is not significant, which indicates that a random effects model is more efficient compared to a binary probit model. In this paper, we stick to binary probit estimations. The qualitative results do not differ to corresponding logit estimations at all. Results for the fixed effects estimates are provided by the authors on request.

4. However, this effect is not very big and not independent from the effect of age as increasing tenure of an individual goes automatically along with an increasing age. A joint consideration of both tenure and age reveals a constantly negative influence on training participation, which means that the age effect overcompensates the effect of tenure in higher years.