Comparing the Incomparable? Issues of Lacking Common Support, Functional-Form Misspecification, and Insufficient Sample Size in Decompositions

Decompositions make it possible to investigate whether gaps between groups in certain outcomes would remain if had comparable characteristics. In practice, however, such a counterfactual comparability is difficult establish the presence of lacking common support, functional-form misspecification, and insufficient sample size. this article, authors show how decompositions can be undermined by these three interrelated issues comparing results regression-based Kitagawa-Blinder-Oaxaca decomposition matching applied simulated real-world data. The that are robust support misspecification but demand large number observations. provide consistent estimates also for smaller samples require assumptions model specification and, when lacking, model-based extrapolation. recommend any benefits from using approach first assess potential problems misspecification.