Technology Category
Reasoning under Uncertainty: CausalityMachine Learning: Causal LearningGame Theory and Economic Paradigms: Mechanism Design
Non-Existent Outcomes in Research on Inequality: A Causal Approach
🤖 AI Summary
In causal inference for socially stratified outcomes—such as wages, which are observable only among employed individuals—conventional censoring approaches discard non-respondents (e.g., the unemployed), leading to biased estimation when treatment simultaneously affects both outcome existence and outcome magnitude. This paper proposes a principal stratification–based separable estimation framework that decomposes the overall causal effect into two components: (i) the treatment effect on outcome existence (e.g., employment status), and (ii) the treatment effect on the outcome value within the latent stratum of those for whom the outcome is potentially observable. Under control for observed confounders, the method integrates regression modeling with simulation-based inference to achieve consistent estimation. By explicitly accounting for selection-on-outcome-existence, it mitigates bias arising from selective observability. Empirically applied to estimating the labor-market effects of parenthood, the approach substantially improves causal identification accuracy. This work establishes a novel paradigm for causal inference with existence-dependent outcomes.
Technology Category
Application Category
📝 Abstract
Scholars of social stratification often study exposures that shape life outcomes. But some outcomes (such as wage) only exist for some people (such as those who are employed). We show how a common practice -- dropping cases with non-existent outcomes -- can obscure causal effects when a treatment affects both outcome existence and outcome values. The effects of both beneficial and harmful treatments can be underestimated. Drawing on existing approaches for principal stratification, we show how to study (1) the average effect on whether an outcome exists and (2) the average effect on the outcome among the latent subgroup whose outcome would exist in either treatment condition. To extend our approach to the selection-on-observables settings common in applied research, we develop a framework involving regression and simulation to enable principal stratification estimates that adjust for measured confounders. We illustrate through an empirical example about the effects of parenthood on labor market outcomes.
Problem
Research questions and friction points this paper is trying to address.
Addressing bias from dropping non-existent outcomes in causal inference
Estimating treatment effects when outcomes exist conditionally
Developing principal stratification methods with confounder adjustment
Innovation
Methods, ideas, or system contributions that make the work stand out.
Principal stratification for non-existent outcomes
Regression and simulation framework
Adjusting for measured confounders
🔎 Similar Papers
💼 Related Jobs
