Breaking down the Gender Pay Gap through a machine learning model

Being able to decompose the gender pay gap (GPG) and determine the contribution of each component is important to design appropriate policies to reduce it. With the aim of providing a new tool to achieve this, in this paper, we propose a decomposition approach based on a machine learning model. The tool was implemented on a population of 5,742 Argentinean IT-related workers to obtain the value of the adjusted and unadjusted GPG in a four-phase process: sample characterization, development of a wage predictor, calculation of adjusted GPG, and analysis of the explained component of GPG. According to our analysis, there is a GPG of 20%, 7,7% of which can be explained exclusively by direct discrimination while 12,3% can be ascribed to other factors, such as total years of experience, educational level, and number of people in charge.