2018/01/11 by Edward H. Kennedy, Sivaraman Balakrishnan, Kennedy, Edward H. +3 · 7 citations
Mathematics · Social Sciences · #Advanced Causal Inference Techniques #Electoral Systems and Political Participation #FOS: Computer and information sciences #Methodology (stat.ME) #Statistical Methods and Inference
paper · pdf · doi:10.48550/arxiv.1801.03635
openalex publication_date 2018/01/11 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
It is well-known that, without restricting treatment effect heterogeneity,\ninstrumental variable (IV) methods only identify "local" effects among\ncompliers, i.e., those subjects who take treatment only when encouraged by the\nIV. Local effects are controversial since they seem to only apply to an\nunidentified subgroup; this has led many to denounce these effects as having\nlittle policy relevance. However, we show that such pessimism is not always\nwarranted: it is possible in some cases to accurately predict who compliers\nare, and obtain tight bounds on more generalizable effects in identifiable\nsubgroups. We propose methods for doing so and study their estimation error and\nasymptotic properties, showing that these tasks can in theory be accomplished\neven with very weak IVs. We go on to introduce a new measure of IV quality\ncalled "sharpness", which reflects the variation in compliance explained by\ncovariates, and captures how well one can identify compliers and obtain tight\nbounds on identifiable subgroup effects. We develop an estimator of sharpness,\nand show that it is asymptotically efficient under weak conditions. Finally we\nexplore finite-sample properties via simulation, and apply the methods to study\ncanvassing effects on voter turnout. We propose that sharpness should be\npresented alongside strength to assess IV quality.\n