Please use this identifier to cite or link to this item: https://hdl.handle.net/10419/36762 
Year of Publication: 
2010
Series/Report no.: 
IZA Discussion Papers No. 4928
Publisher: 
Institute for the Study of Labor (IZA), Bonn
Abstract: 
In empirical studies it often happens that some variables for some units are far away from the other observations in the sample. These extreme observations, or outliers, often have a large impact on the results of statistical analyses - conclusions based on a sample with and without these units may differ drastically. While applied researchers tend to be aware of this, the detection of outliers and their appropriate treatment is often dealt with in a rather sloppy manner. One reason for this habit seems to be the lack of availability of appropriate canned programs for robust methods that can be used in the presence of outliers. Our paper intents to improve on this situation by presenting a highly robust method for estimation of the popular linear fixed effects panel data model, and to supply Stata code for it. In an application from the field of the micro-econometrics of international firm activities we demonstrate that outliers can indeed drive results.
Subjects: 
Robust estimation
panel data
outliers
Stata
exporter productivity premium
JEL: 
C23
C81
C87
F14
Document Type: 
Working Paper

Files in This Item:
File
Size
134.36 kB





Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.