32492 observations (6171 workers). Appendix A describes the data selection in more details
and provides sample means of the main variables used in the analysis hereafter.
The GSOEP contains two sources of information to describe workers’ careers within a firm.
First, the survey contains a question on job changes. Each year, individuals are asked to report
whether they have experienced a change in job situation since the previous year’s survey. This
question makes it possible to identify workers’ careers within and across firms. 15 Appendix B1
provides information on inter and intra firm mobility frequencies by experience and associated
wage growth. Overall, 11.2% of the observations report mobility, 2.9% of the observations
report intra-firm mobility. Although intra-firm mobility is low, it is comparable to the U.S. as
a proportion of the total reported moves. 16
Second, there is a question in which individuals are asked to identify their current position
with a choice among five categories: blue-collar, white-collar, civil servant, trainee and self-
employed. I considered the first three given that self-employment is not relevant for the analysis
and that the trainee category is not in itself an occupation. 17 Each position is subdivided into
a hierarchical structure of job levels or ranks according to the level of skills and responsibilities
required for the job. Appendix B2 describes in more details the occupational rank variables
and provides average characteristics by job changes and rank changes.
3.2 Summary Statistics on Intra-Firm Mobility and Wage Outcomes
A natural starting point before assessing the importance of the comparative advantage and
learning assumptions in explaining mobility and wage dynamics is to see whether the German
data exhibit the stylized facts of the U.S. data which motivated the Gibbons and Waldman
model.
Because the question on job change within the firm does not provide information on the
type of job change experienced, I use the information on job rank comparing current and
previous job rank to categorize job changes as promotions. 18 Table 1 presents average wage
order to exclude outliers for wages without losing observations on low wage workers such as trainees.
15When considering the sample of workers remaining with their firm over the period (reporting either a
job change within the firm or no change in job situation), the sample size becomes 11159 observations (3487
workers).
16McCue (1996) uses the PSID and finds that about 1/4 of the reported moves are promotions within a firm.
17Individuals identified as trainees at any point during the sample period were excluded unless they reported
the occupation for which they were training.
18 The question on job change is used to identify the sub-sample of workers who remain within their firms,
14