Bellmann, L. and Bender, S. and Hornsteiner, U.
Interfirm Job Mobility of Two Cohorts of Young German Men 1979 - 1990: An analysis of the (West-)German Employment Statistic Register Sample concerning multivariate failure times and unobserved heterogeneity.
Collaborative Research Center 386, Discussion Paper 94
The OECD (1993) has documented that the majority of workers in industrialised countries can look forward to finding a stable employment relationship. However new entrants into the labor force experience high turnover. Promoting institutions which support longer tenures and worker participation (or ''voice`` in the firm) utilize strategies to encourage enterprise and employee efforts in skill formation and training. The results of the OECD (1993) study show that attachments between employee and employer are more likely to endure for Japanese, French and German workers. Furthermore Germany has the highest share of young new recruits who received any formal training from their employer. In Germany, 71.5 % of young new recruits were trained at any job within 7 years after leaving school, whereas in the U.S. only 10.2 % of young new recruits were similarly trained (cf. OECD 1993, 137). It is sometimes assumed that employment protection policies have been exogenously imposed and thus probably impair efficiency. However, research on the micro-economics of labor markets has shown that employers may be interested in long-term employment relationships (cf. Levine 1991). Here, the job training model focusing on the importance of human capital investment, specifically the job shopping and matching model stressing the process of information gathering through employment experience should be mentioned. In such models employment protection legislation has not only desirable distributional effects but also help to ensure efficient outcomes. Therefore, it is important to assess the relevance of micro- economic theories empirically. This paper provides an empirical analysis of job durations in Western Germany using information from two cohorts of new entrants to the labor force documented in the (West-)German employment statistic register sample (cf. Bender and Hilzendegen 1996). The appropriate empirical technique to study job length is event history or survival analysis. In labor market research, survival analysis has primarily focused on explaining the length of unemployment spells. Application of this technique to employment is less common 1 , because huge longitudinal data sets are needed. Apart from testing hypotheses about the effect of personal characteristics and labor demand variables (e.g. firm size and industry affiliation), we will assess the influence of heterogeneity of the members of the two cohorts on their duration profile. The applied model and estimation method allow for unobserved heterogeneity and correlation between the clustered failure times of one employee as well as for right-censored spells. Our analysis is not restricted to the beginning of the working life of the employees. The individual retirement decision is affected by employment protection and early retirement regulations which differ widely between the firms. The respective data are missing in the employment statistic register, so that the retirement decision cannot be modelled explicitly.