Despite decades of recognition, gendered pay disparities have remained remarkably persistent in most organisations. In Aotearoa-New Zealand, Public Services such as Crown Research Institutes are required to report the summary statistics of gender and racial pay gaps. However, effectively addressing these pay gaps requires a more detailed understanding of the interplay between the factors underpinning them such as organisational role, tenure and gender. We propose a new methodology to disentangle the factors underlying organisational gender pay gaps using approaches from causal inference and Bayesian statistics. The method is designed to accommodate anonymised data collected at the salary band level. We demonstrate the technique using datasets simulated from a hierarchical organisation and reproduce pay gaps between gendered groups within a complex causal structure. We also model the effectiveness of specific interventions to quantify the degree of change required to reduce gender pay gaps. Although our approach cannot address the structural factors underlying pay gaps, it does provide a method to better understand pay gaps and possible solutions, as a crucial next step in achieving pay parity.
