Identification and Inference for Algorithmic Frontiers with Selective Labels
It addresses the problem of evaluating fairness-accuracy trade-offs in algorithmic decision-making when labels are missing not at random, which is common in practice.
This paper develops methods to identify and perform statistical inference on fairness-accuracy frontiers when outcomes are selectively observed, providing sharp identification regions under unrestricted selection and point identification under unconfoundedness, with debiased machine learning estimators for inference.
This paper provides identification results to characterize a fairness-accuracy (FA) frontier, and statistical inference tools to test hypotheses and build a confidence set for the FA-frontier, when outcomes are observed only for selected individuals. When the selection process is unrestricted but loss is measured in specific ways, we provide a characterization of the sharp identification region of the FA-frontier. Under an assumption of unconfoundedness conditional on observables (and unrestricted loss functions), we obtain point identification and propose a debiased machine learning estimator, derive its asymptotic distribution, and show how this can be used to carry out inference for the FA-frontier. In work in progress, we extend the partial identification results to a broader class of loss functions.