Chief research scientist
The objective of this proposal is to go beyond the prior literature on career and wage dynamics by studying the interplay between within-firm careers and job mobility, which is an important unexplored area in the field of personnel economics. Our specific contributions are as follows:
First, we analyse the interplay between within-firm careers and job mobility in two complementary ways. In the first approach, we extend a work-horse career model to include endogenous across-firm mobility by adding match-specific productivity shocks and test the empirical implications of this model for worker turnover, career progress and wage dynamics. This would be the first career dynamics model in the literature to incorporate endogenous job mobility.
In the second approach, we extend models of occupational mobility to include career development and empirically test the resulting new predictions. Recent research has shown that the existing theories cannot explain all empirical regularities concerning occupational mobility. This may be because the existing frameworks do not allow for hierarchical moves. In both approaches, we use firm-level productivity shocks to generate job mobility.
Second, we develop and test new methods to distinguish empirically between the competing theoretical models of promotions. The theoretical literature has produced many different models of promotions, but distinguishing empirically among these models has proven elusive. In our earlier work, we have shown how to distinguish a particular class of promotion models from others. Now we extend that framework to cover a new and important class of models.
We use a large, linked, employer–employee data set covering white-collar workers in Finnish manufacturing. This data set is exceptionally well-suited to the research questions at hand. First, it includes thousands of firms and millions of employee observations, allowing us to study the interplay of within-firm careers with job mobility. Second, the data are rich enough to empirically distinguish among alternative models of promotions. Third, the data set provides the necessary information on productivity and other firm-level variables.
Much of the analysis will use panel data techniques for linked employer-employee data. To empirically distinguish among different models of promotion, we estimate a system of structural equations by maximum likelihood.