Generalized Multivariate Gaps: A Proposed Gender Gap

Abstract

The “Gap” between multivariable outcome distributions requires well founded aggregation of different outcome dimensions. Kobus, Kapera and Maasoumi (2024) extend the measurement of gender gaps to multiple outcomes (wages, health, leisure). Multivariate outcome distributions should be evaluated by a class of functions with transparent properties. Our general properties uniquely identify wellbeing with a function class that includes popular cases, such as the linear and the Cobb-Douglas. Our measure allows a nuanced and deeper understanding of degrees of substitution/complementarity, and its interplay with planner’s degrees of aversion to inequality and different population groups. We show that the General Gender Gap (GGG) in the US is substantively different from the univariate (income or wealth) gaps. Substitution or complementarity degrees are key, impacting the evolution of GGG over time, especially with heterogeneity in preferences. Women are worse off than men in the US, specially with lower concordance of income and leisure – a robust finding given the generality of the new measure.

Keywords: Gaps, Inequality, Gender, leisure, wages, aggregation, axioms, Optimal transport Theory

JEL Classification: C02, D30, I31