Phillip Heiler from Aarhus University will present “Treatment Evaluation at the Intensive and Extensive Margins” joint work with Asbjørn Kaufmann and Bezirgen Veliyev.

Abstract:

This paper provides a solution to the evaluation of treatment effects in selective

samples when neither instruments nor parametric assumptions are available. We

exploit a weak conditional monotonicity assumption that allows for the presence of

a large share of units whose selection behavior is unaffected by the treatment, i.e. an

unknown subpopulation with a sparsity constraint. We provide semiparametrically

efficient smooth bounds for all principal strata in the nonparametric sample se-

lection model, i.e. at the complete extensive and intensive margins. The bounds

circumvent the irregularity of the general identification problem. Their degree of

smoothness can be chosen adaptively to navigate a trade-off between identifica-

tion and precision in finite samples without sacrificing validity of the corresponding

confidence intervals.