The identification region of the potential outcome distributions under instrument independence
推导了在工具变量不同独立性假设下潜在结果分布的识别区域，比较了边际独立、联合独立和单调性假设的识别力，并给出了平均处理效应的紧界和可检验的约束条件。
This paper examines the identifying power of instrument exogeneity in the treatment effect model. We derive the identification region of the potential outcome distributions, which are the collection of distributions that are compatible with data and with the restrictions of the model. We consider identification when (i) the instrument is independent of each of the potential outcomes (marginal independence), (ii) the instrument is independent of the potential outcomes and selection heterogeneity jointly (joint independence), and (iii.) the instrument satisfies joint independence and monotonicity (the LATE restriction). By comparing the size of the identification region under each restriction, we show that joint independence provides more identifying information for the potential outcome distributions than marginal independence, but that the LATE restriction provides no identification gain beyond joint independence. We also, under each restriction, derive sharp bounds for the Average Treatment Effect and sharp testable implication to falsify the restriction. Our analysis covers discrete or continuous outcomes, and extends the Average Treatment Effect bounds of Balke and Pearl (1997) developed for the dichotomous outcome case to a more general setting.