growth by type of job change and rank change for workers who do not change firms. The
first two columns report average wage growth for those who changed jobs within the firm and
those who did not. The last three present average wage growth of job changers by type of job
changes.
Not surprisingly, the main difference associated with a change in rank among job changers
is the average wage growth which is 16.25% for workers who receive promotions. 19 Like the
findings for the U.S, this suggests that hierarchical rank effects play an important role in the
wage determination process in German firms. There is also evidence that previous wage growth
predict promotions. The average wage growth the period before a reported change in rank or
job (columns 3 or 2) is higher than it is when there is no change in rank or job (column 5 or
column 1).
Note that the percentage of changes involving a change to a lower rank is high relative
to previous findings on demotions. However, these changes are associated with positive wage
growth suggesting that they may not in fact be demotions and may instead result from misclas-
sification in job ranks. This would not be surprising given the known sensitivity of survey data
to this type of problem. Given that rank changes are central to the estimation of the Gibbons
and Waldman model, I corrected for possible classification errors using the information on job
changes and wage growth. 20 The resulting data, which will be used for the remaining of the
analysis, present similar average characteristics for firms and individuals as the one without
corrections. Average wage growth associated with demotion is now lower (-1.52%) and average
wage growth with no change in rank (but a change in job) is now higher (4.67%).
To see whether there is evidence of individual variation in wages within a rank, I compare
average wage growth at promotion with the difference in average wages for workers in two
consecutive ranks. To do so, I need to compute wage growth at promotion at the different
ranks. Because rank definitions and subdivisions are similar across the three occupations
considered, they can be summarized in a single hierarchical job ladder using the following 4
generic rank definitions: 21
either experiencing no change in job or a change within.
19 Note also that job changers that do not experience a change in rank receive on average a wage growth of
2.94% which is higher than the average wage growth associated with no change in job suggesting that part of
the change in job would be pay related.
20Details about the correction method and resulting changes in the data are presented in Appendix B3.
21 The blue-collar occupation is originally divided into 5 ranks, distinguishing unskilled from semi-skilled work.
I grouped the two categories into one corresponding to the lower occupational rank. See appendix B for details
15