Gender Wage Gap Decomposition Methods: Comparative Analysis

This thesis introduces a comparative analysis of the gender wage gap decomposition methods with the Russian Longitudinal Monitoring Survey (RLMS) data for 2018. To decompose the differences in average wages, approaches based on the Oaxaca-Blinder decomposition are used. In addition to estimating an overall effect of the wages’ composition and structure, these methods allow estimating the contribution of individual variables into these effects. Apart from the mean wages, the study focuses on other distribution statistics. Using the quantile regressions, the wage gap between men and women is decomposed for the distribution parameters such as median, lower and upper deciles. The decomposition estimates of conditional and unconditional recentered influence function quantile regressions are compared. The results of the study showed that the gender gap in wages is mainly due to the returns to characteristics. Thus, the Russian labour market differently values the same characteristics in men and women. Furthermore, the largest share of the returns’ effect is observed in the median of the distribution. Our data do not confirm the hypothesis about the industry gender segregation as a dominant contributor to the gender gap in wages. The wage gap is associated primarily with the professional status of the employee, which contributes to narrowing the gap across the whole distribution. The same effect is caused by an employee's education and job tenure, but only in the lower part of the distribution. Higher wages in the private sector and the large enterprises dominated by men employees lead to an increase in the gender wage gap. A method used to correct for self-selection into employment increased the gender wage gap from 30% to 48% as compared to the model without correction.