is based on a study of only one firm). 3 Second, nominal wage cuts are rare but real wage
cuts are much more common. Partly this is because nominal wage increases are insensitive to
inflation and zero nominal increases are not rare. 4 Third, the dynamics of wages within the
firm exhibit serial correlation in the sense that a real wage increase (decrease) today is serially
correlated with a real wage increase (decrease) tomorrow. 5 Fourth, studies that analyze the
relationship between wages and intra-firm mobility find that wage increases received by work-
ers who are promoted exceed increases reported by workers who do not receive promotions. 6
However, wage increases upon promotion are small compared to the difference in average wages
between two job levels. In other words, significant variations in wages remain within each level
so that wages are not tied to levels. Finally, wage increases forecast promotions in the sense
that those who receive larger wage increases get promoted more rapidly. 7
Collectively, these observations posed a challenge to the existing theoretical literature, as
no pre-existing theory could explain all of these stylized facts. In response to this challenge,
Gibbons and Waldman (1999) build a synthesized model which combines on the job human
capital accumulation, job assignment based on comparative advantage and learning dynamics.
The predictions of their model are consistent with most of the stylized facts found in the
empirical literature. The main contribution of this paper is to examine the explanatory power
of the Gibbons and Waldman theory over a large sample of firms.
The paper is organized as follows. Section 2 sketches the theoretical model of Gibbons
and Waldman and establishes the framework of the econometric analysis and how this relates
to the theory. Section 3 presents the data and provides a descriptive analysis of intra-firm
mobility and wage outcomes in German firms. Section 4 presents the results of the estimation,
and Section 5 concludes the paper.
3Baker, Gibbs and Holmstrom (1994 a,b).
4Baker, Gibbs and Holmstrom(1994 a,b) report this in the case of one firm, and Card and Hyslop (1997)
arrive at the same conclusions using the CPS and PSID. Peltzman (2000) reports a similar finding using BLS
data.
5Hause (1980), Lillard and Weiss (1979) and Baker, Gibbs and Holmstrom (1994 a,b).
6Murphy (1985) Baker, Gibbs and Holmstrom (1994 a,b) and McCue (1996).
7Baker, Gibbs and Holmstrom (1994 a,b). McCue (1996) finds that a high wage today is positively correlated
with promotion tomorrow, and Topel and Ward (1992) find that prior wage growth affects mobility even after
controlling for current wage.