USING INSTRUMENTAL VARIABLES FOR INFERENCE ABOUT POLICY RELEVANT TREATMENT PARAMETERS
USING INSTRUMENTAL VARIABLES FOR INFERENCE ABOUT POLICY RELEVANT TREATMENT PARAMETERS
复制标题
DOI：
--
复制
发表时间：
2016
期刊：
影响因子：
--
通讯作者：
M. Mogstad;Andrés Santos;Alexander Torgovitsky
中科院分区：
文献类型：
--
作者：
M. Mogstad;Andrés Santos;Alexander Torgovitsky
我们提出了一种使用工具变量（IV）的方法来推断除受手头工具影响的个体之外的个体的因果效应。政策相关性和外部有效性的问题取决于我们可靠地做到这一点的能力。我们的方法利用了IV估计和许多处理参数都可以表示为相同潜在边际处理效果的加权平均值的洞察力。由于权重是已知或确定的，对IV估计的了解通常会对未知的边际处理效果施加一些限制，从而对感兴趣的处理参数的逻辑允许值施加一些限制。我们展示了如何从IV估计中提取有关兴趣平均效应的信息，更一般地说，从一类类似IV的估计中提取信息，其中包括TSLS和OLS估计，以及许多其他估计。我们的方法有几个应用。首先，它可以用来构造实际或假设的政策变化的平均因果效应的非参数界限。其次，我们的方法允许研究人员灵活地结合形状限制和参数假设，从而使编译器的平均效应外推到不同或更大的群体的平均效应。第三，我们的方法提供了规范测试。除了检验正确指定模型的零值外，我们还可以使用我们的方法来检验无选择偏差、无增益选择和工具有效性的零假设。重要的是，使用我们的方法的规格测试不要求治疗效果对具有相同观察值的个体是恒定的。为了说明我们方法的适用性，我们使用挪威的行政数据来推断家庭规模对儿童结果的因果影响。我们感谢2016年考尔斯基金会夏季计量经济学会议的听众。Bradley Setzler提供了优秀的研究助理。†芝加哥大学经济系；统计挪威;国家经济研究局。加州大学圣地亚哥分校经济系该研究部分由美国国家科学基金会资助SES-1426882。§西北大学经济系研究部分由国家科学基金会资助SES-1530538。对芝加哥大学经济系的热情好客表示感谢。
We propose a method for using instrumental variables (IV) to draw inference about causal effects for individuals other than those affected by the instrument at hand. The question of policy relevance and external validity turns on our ability to do this reliably. Our method exploits the insight that both the IV estimand and many treatment parameters can be expressed as weighted averages of the same underlying marginal treatment effects. Since the weights are known or identified, knowledge of the IV estimand generally places some restrictions on the unknown marginal treatment effects, and hence on the logically permissible values of the treatment parameters of interest. We show how to extract the information about the average effect of interest from the IV estimand, and more generally, from a class of IV-like estimands which includes the TSLS and OLS estimands, among many others. Our method has several applications. First, it can be used to construct nonparametric bounds on the average causal effects of an actual or hypothetical policy change. Second, our method allows the researcher to flexibly incorporate shape restrictions and parametric assumptions, thereby enabling extrapolation of the average effects for compliers to the average effects for different or larger populations. Third, our method provides specification tests. In addition to testing the null of correctly specified model, we can use our method to test null hypotheses of no selection bias, no selection on gains and instrument validity. Importantly, specification tests using our method do not require the treatment effect to be constant over individuals with the same observables. To illustrate the applicability of our method, we use Norwegian administrative data to draw inference about the causal effects of family size on children’s outcomes. ∗We thank the audience at the 2016 Cowles Foundation Summer Econometrics Conference. Bradley Setzler provided excellent research assistant. †Department of Economics, University of Chicago; Statistics Norway; NBER. ‡Department of Economics, University of California at San Diego. Research supported in part by National Science Foundation grant SES-1426882. §Department of Economics, Northwestern University. Research supported in part by National Science Foundation grant SES-1530538. The hospitality of the Department of Economics at the University of Chicago is gratefully acknowledged.