The gap-closing estimand: A causal approach to study interventions that close disparities across social categories
Disparities across social categories such as race, gender, and class are central in social stratification. Questions about the manipulability of these constructs, however, hinder their placement within a causal framework. This paper sidesteps the manipulability problem by advancing a different causal question: what gap across categories would persist under an intervention to equalize a treatment variable? The proposal makes three contributions. First, I distinguish between a local intervention to randomly sampled units (empirically tractable) and a global intervention to the population (the less tractable target of prior research). Second, I formalize stochastic treatment assignment rules that can improve credibility. Third, I provide a doubly-robust estimator. I illustrate by examining the pay gap by class origins under an intervention to equalize class destinations. The paper concludes with implications for practice: gap-closing estimands provide tools for the rigorous study of inequality across social categories that could inform policies to close gaps.