Please use this identifier to cite or link to this item: https://hdl.handle.net/10419/222047 
Year of Publication: 
2018
Series/Report no.: 
IEHAS Discussion Papers No. MT-DP - 2018/34
Publisher: 
Hungarian Academy of Sciences, Institute of Economics, Budapest
Abstract: 
Men's labor income is on average higher than that of women practically everywhere. This gender pay gap can be decomposed into two components: on the one hand men usually work in better paid jobs (the sorting effect), and, on the other, even in the same occupation men get higher wages (the within-job pay gap). In this paper we focus on the second component, by trying to identify those jobs where the gender of the employee matters most. Using Hungarian individual data from a dataset where jobs are identified by their 3-digit employment classification code, we compute three statistical measures that turn out to entail more and more stringent criteria for variable importance. Our simplest measure is significance at the 5 percent in linear regressions. Judging by this criterion the majority of occupations have a gender pay gap. Secondly, we compute a variable importance measure defined for regression analysis, that narrows down the group of jobs where being male seems to carry definite financial advantages. Finally, we apply an alternative methodology, Random Forest regression, and calculate one of the associated variable importance measures. This new indicator reduces our looked for job categories even farther, and gives rather sharp results concerning the type of jobs where the within-job pay gap is definitely detectable. We find that gender has the most clearly distinguishable role in occupations requiring the least education. The broader categories include "Craft and Related Trades Workers", "Plant and Machine Operators and Assemblers" as well as "Elementary Occupations". Our results suggest that the vanishing of the overall pay gap in Hungary is partly due to the fact that in higher skilled jobs the occupational pay gap is not so important, whereas it obscures the fact that in lower-paid unskilled jobs it is still very much extant.
Subjects: 
gender wage gap
wages and education
Random Forest regression
JEL: 
C16
J31
J79
Document Type: 
Working Paper

Files in This Item:
File
Size





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