ON THE POWER PROPERTIES OF INFERENCE FOR PARAMETERS WITH INTERVAL IDENTIFIED SETS

This article studies the power properties of confidence intervals (CIs) for a partially-identified parameter of interest with an interval identified set. We assume that the researcher has bounds estimators needed to construct the CIs proposed by Imbens and Manski (2004), Stoye (2009), and Stoye (2020), denoted by C I α 1 $CI_{\alpha }^{1}$ upper C upper I Subscript alpha Superscript 1 , C I α 2 $CI_{\alpha }^{2}$ upper C upper I Subscript alpha Superscript 2 , C I α 3 $CI_{\alpha }^{3}$ upper C upper I Subscript alpha Superscript 3 , and C I α 4 $CI_{\alpha }^{4}$ upper C upper I Subscript alpha Superscript 4 . We also assume that these bounds estimators are “ordered”: the lower bound estimator is less than or equal to the upper bound estimator. This setup arises in economic applications involving missing data and treatment effects. Under these conditions, we establish two results. First, we show that C I α 1 $CI_{\alpha }^{1}$ upper C upper I Subscript alpha Superscript 1 and C I α 2 $CI_{\alpha }^{2}$ upper C upper I Subscript alpha Superscript 2 are equally powerful, and both dominate C I