New methods to estimate models with large sets of fixed effects with an application to matched employer-employee data from Germany

Abstract

"This paper will introduce new methods to estimate the two-way fixed effects model and the match effects model in datasets where the number of fixed effects makes standard estimation techniques infeasible. The methods work for balanced and unbalanced panels and increase the speed of estimation without imposing excessive computational demands. I will apply the methods to a new and unusually detailed matched employer-employee dataset from Germany. The analysis shows that the omission of match effects leads to biased inference particularly concerning the effects of individual characteristics and underlines the importance of accurate biographic data." (Author's abstract, IAB-Doku) ((en))

Cite article

Mittag, N. (2012): New methods to estimate models with large sets of fixed effects with an application to matched employer-employee data from Germany. (FDZ-Methodenreport 01/2012 (en)), Nürnberg, 63 p.