The Gender Wage Gap and Social Skills on the Labour Market: A Machine Learning Approach

Abstract

This thesis examines a skill-based explanation of the gender wage gap in the light of social skills’ increasing importance on the labour market, bridging the gap between two prominent topics of labour economics. In my empirical strategy, I utilise a novel dataset that matches Danish register data with job descriptions, connecting workers’ skills to their wages. I find a considerable gender wage gap of 20% and decompose it with respect to workers’ different characteristics. This shows that women and men do not earn the same wage premium for different skills. Furthermore, by separating workers who work within social skill-intensive jobs from workers who do not, I am able to quantify the wage impact of social skills and their interaction with workers’ other characteristics. Hypothesizing that this effect might be driven by workers’ self selecting themselves into specific jobs based on their characteristics, I apply a generalized random forests model. Using this model, I find a wage premium of approximately 3% on social skill-intensive jobs, and this effect is higher for women than for men. The social skill wage premium exhibits a strongly heterogeneous pattern, and I find causal evidence in favour of positive spillover effects across different skill categories.