POLICY LEARNING WITH OBSERVATIONAL DATA

被引:170
作者
Athey, Susan [1 ]
Wager, Stefan [1 ]
机构
[1] Stanford Univ, Stanford Grad Sch Business, Stanford, CA 94305 USA
基金
美国国家科学基金会;
关键词
Double robustness; empirical welfare maximization; minimax regret; semiparametric efficiency; REGRET TREATMENT CHOICE; SEMIPARAMETRIC EFFICIENCY; PERFORMANCE GUARANTEES; TREATMENT REGIMES; TREATMENT RULES; MODELS; IDENTIFICATION; INFERENCE; RATES;
D O I
10.3982/ECTA15732
中图分类号
F [经济];
学科分类号
02 ;
摘要
In many areas, practitioners seek to use observational data to learn a treatment assignment policy that satisfies application-specific constraints, such as budget, fairness, simplicity, or other functional form constraints. For example, policies may be restricted to take the form of decision trees based on a limited set of easily observable individual characteristics. We propose a new approach to this problem motivated by the theory of semiparametrically efficient estimation. Our method can be used to optimize either binary treatments or infinitesimal nudges to continuous treatments, and can leverage observational data where causal effects are identified using a variety of strategies, including selection on observables and instrumental variables. Given a doubly robust estimator of the causal effect of assigning everyone to treatment, we develop an algorithm for choosing whom to treat, and establish strong guarantees for the asymptotic utilitarian regret of the resulting policy.
引用
收藏
页码:133 / 161
页数:29
相关论文
共 86 条
[1]   Semiparametric instrumental variable estimation of treatment response models [J].
Abadie, A .
JOURNAL OF ECONOMETRICS, 2003, 113 (02) :231-263
[2]   Estimation of possibly misspecified semiparametric conditional moment restriction models with different conditioning variables [J].
Ai, Chunrong ;
Chen, Xiaohong .
JOURNAL OF ECONOMETRICS, 2007, 141 (01) :5-43
[3]  
[Anonymous], 2018, ARXIV181004778
[4]  
ARMSTRONG T. B., 2013, TECHNICAL REPORT
[5]  
ARMSTRONG T. B., 2017, ARXIV171204594
[6]   Beyond LATE: Estimation of the Average Treatment Effect with an Instrumental Variable [J].
Aronow, Peter M. ;
Carnegie, Allison .
POLITICAL ANALYSIS, 2013, 21 (04) :492-506
[7]  
ATHEY S., 2021, ECONOMETRICA S, V89, DOI [10.3982/ECTA15732, DOI 10.3982/ECTA15732]
[8]   GENERALIZED RANDOM FORESTS [J].
Athey, Susan ;
Tibshirani, Julie ;
Wager, Stefan .
ANNALS OF STATISTICS, 2019, 47 (02) :1148-1178
[9]   Recursive partitioning for heterogeneous causal effects [J].
Athey, Susan ;
Imbens, Guido .
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 2016, 113 (27) :7353-7360
[10]   The Big Data Newsvendor: Practical Insights from Machine Learning [J].
Ban, Gah-Yi ;
Rudin, Cynthia .
OPERATIONS RESEARCH, 2019, 67 (01) :90-108