CMStatistics 2022: Start Registration
View Submission - CFE
A1731
Title: Decomposing earnings uncertainty using German SOEP data Authors:  Friederike Schmal - University of Muenster (Germany) [presenting]
Abstract: A new approach is suggested to decompose earnings development in a predictable and unpredictable component, adjusted for the German labor market. We examine which part of income development was already predictable for the individual when he or she just left school. It is at this point that the decision about future education and thus its influence on future income is made. For this purpose, we use the link between expectations of future income variability, later realized income, and the educational decision regarding university attendance. The difficulty at this point is that only one income trajectory per individual can be observed at a time (income with university degree or without); however, we need both trajectories for our analysis. Therefore, we develop a new approach so that we can estimate the unobserved trajectory with a likelihood function for each individual. To avoid too strong restrictions on the covariance structures and to deal with the high number of parameters to be estimated and the counterfactual income trajectory, we will apply an MCMC algorithm and Gibbs sampling. A factor model will also be included, which will allow us to make statements about the predictable components. We will apply our approach to the German SOEP data that contains not only information on income history but also educational background and numerous demographic variables.