Exploring Predictive Factors for Gender Pay Gaps among Korean Doctoral Holders Using Random Forest Regression Models

This is an emprical study that explores the predictors and the patterns influence of gender wage gap among domestic doctorate holders in Korea. Using the 2021 KCDH data, a random forest analysis was conducted on 3,573 employed individuals aged 21 to 65, with 151 to 153 explanatory variables predicting the wage levels of doctorate holders. The results showed that both gender variables were included in the top 20 major variables. Among these, men had a positive relationship with wages, while women had a negative relationship. A subsequent random forest analysis was performed on male and female doctorate holders seperately to idendify key predictors of wages. The findings reveaed that, compared to men, women placed greater importance on variables such as weekly working hours at their current job, employment status (e.g., non-regular employment), methods of funding doctoral studies, and mismatches between major and occupation or downward mobility. The analysis also highlighted that women were found to be in a more disadvantageous position than men when it comes to the motherhood wage penlty. The study discusses theses findings and presents policy implications aimed at promoting the career development of female doctorates.