This preliminary look at the data suggests that German data seem to share some of the
same stylized facts as the U.S. data which makes it worth pursuing the analysis by estimating
the Gibbons and Waldman model using these data. According to the model, the joint effects
of job and individual characteristics in the wage determination process can be explained by
the assumption of the workers’ comparative advantage in a given rank and the fact that
individual skills are differently rewarded in each rank. Going further in assessing the role
of comparative advantage, the next subsection provides preliminary evidence on the joint
impact of the rank variables and individual skills on wage outcomes. Also given the primary
role of unmeasured (by the econometrician) ability (unobserved in the case of learning) in
the Gibbons and Waldman model, the next subsection provides preliminary evidence on the
presence of unmeasured ability in the wage determination process.
3.3 Preliminary Evidence on the Role of Comparative Advantage and Un-
measured Ability
In this section, I analyze whether comparative advantage based on measured ability is im-
portant. To do so, I estimate the joint effect of rank and individual characteristics in an
estimation of inter-rank wage differentials. I also consider the importance of unmeasured abil-
ity by comparing the results of an OLS estimation of the rank wage premia with the results
of a fixed-effect estimation. The idea is that if comparative advantage based on measured
individual characteristics matters and if there is evidence of unmeasured ability in the wage
determination process, one can expect to find some evidence of comparative advantage based
on unmeasured ability and therefore proceed to the estimation of the Gibbons and Waldman
model.
Differences in average individual characteristics across ranks are presented in the Appendix
C Table which reports average education, potential experience, marital status, woman and
German percentages together with the raw wage differentials (relative to the lower rank) by
rank. As observed in Table 2, the wage differentials increase within job rank in different
proportions that depend on the type of occupation, with white-collared workers showing the
highest differentials in each rank. While these rank wage premia might reflect the increasing
responsibilities and task complexity of higher rank jobs, there is a positive correlation between
rank premia and measures of individual ability such as education. From the results of the
appendix C Table, the link with other characteristics is however less clear. A global measure
of the workers’s individual characteristics would be more convenient for analyzing interaction
effects of the worker’s ability and his job rank in the wage determination process.
17