We propose a doubly robust inference method for causal effects of continuous treatment variables, under unconfoundedness and with nonparametric or high-dimensional nuisance functions. Our double debiased machine learning (DML) estimators for the average dose-response function (or the average structural function) and the partial effects are asymptotically normal with non-parametric convergence rates. The first-step estimators for the nuisance conditional expectation function and the conditional density can be nonparametric or ML methods. Utilizing a kernel-based doubly robust moment function and cross-fitting, we give high-level conditions under which the nuisance function estimators do not affect the first-order large sample distribution of the DML estimators. We provide sufficient low-level conditions for kernel, series, and deep neural networks. We justify the use of kernel to localize the continuous treatment at a given value by the Gateaux derivative. We implement various ML methods in Monte Carlo simulations and an empirical application on a job training program evaluation
In this study, we investigate estimation and inference on a low-dimensional causal parameter in the presence of high-dimensional controls in an instrumental variable quantile regression. Our proposed econometric procedure builds on the Neyman-type orthogonal moment conditions of a previous study Chernozhukov, Hansen and Wuthrich (2018) and is thus relatively insensitive to the estimation of the nuisance parameters. The Monte Carlo experiments show that the estimator copes well with high-dimensional controls. We also apply the procedure to empirically reinvestigate the quantile treatment effect…
Higher-Order Debiased Estimators for General Treatment Models
Institute of Statistics & Big Data, Renmin University of China（中国人民大学统计与大数据研究院）；Institute of Natural Sciences, MOE–LSC, School of Mathematical Sciences, CMA–Shanghai, SJTU–Yale Joint Center for Biostatistics and Data Science, Shanghai Jiao Tong University（上海交通大学数学科学学院自然科学研究院、MOE-LSCEA-Mathematics上海中心、上交-耶鲁生物统计与数据科学联合中心）
This paper extends the identification results in Nevo and Rosen (2012) to nonparametric models. We derive nonparametric bounds on the average treatment effect when an imperfect instrument is available. As in Nevo and Rosen (2012), we assume that the correlation between the imperfect instrument and the unobserved latent variables has the same sign as the correlation between the endogenous variable and the latent variables. We show that the monotone treatment selection and monotone instrumental variable restrictions, introduced by Manski and Pepper (2000, 2009), jointly imply this assumption. Mo…
We consider the estimation of treatment effects in settings when multiple treatments are assigned over time and treatments can have a causal effect on future outcomes or the state of the treated unit. We propose an extension of the double/debiased machine learning framework to estimate the dynamic effects of treatments, which can be viewed as a Neyman orthogonal (locally robust) cross-fitted version of $g$-estimation in the dynamic treatment regime. Our method applies to a general class of non-linear dynamic treatment models known as Structural Nested Mean Models and allows the use of machine …
In their IZA Discussion Paper 10247, Johansson and Lee claim that the main result (Proposition 3) in Abbring and Van den Berg (2003b) does not hold. We show that their claim is incorrect. At a certain point within their line of reasoning, they make a rather basic error while transforming one random variable into another random variable, and this leads them to draw incorrect conclusions. As a result, their paper can be discarded.
Many causal and structural effects depend on regressions. Examples include policy effects, average derivatives, regression decompositions, average treatment effects, causal mediation, and parameters of economic structural models. The regressions may be high dimensional, making machine learning useful. Plugging machine learners into identifying equations can lead to poor inference due to bias from regularization and/or model selection. This paper gives automatic debiasing for linear and nonlinear functions of regressions. The debiasing is automatic in using Lasso and the function of interest wi…
This paper considers the practically important case of nonparametrically estimating heterogeneous average treatment effects that vary with a limited number of discrete and continuous covariates in a selection-on-observables framework where the number of possible confounders is very large. We propose a two-step estimator for which the first step is estimated by machine learning. We show that this estimator has desirable statistical properties like consistency, asymptotic normality and rate double robustness. In particular, we derive the coupled convergence conditions between the nonparametric a…
The policy relevant treatment effect (PRTE) measures the average effect of switching from a status-quo policy to a counterfactual policy. Estimation of the PRTE involves estimation of multiple preliminary parameters, including propensity scores, conditional expectation functions of the outcome and covariates given the propensity score, and marginal treatment effects. These preliminary estimators can affect the asymptotic distribution of the PRTE estimator in complicated and intractable manners. In this light, we propose an orthogonal score for double debiased estimation of the PRTE, whereby th…
This paper proposes a new class of heterogeneous causal quantities, named \textit{outcome conditioned} average structural derivatives (OASD) in a general nonseparable model. OASD is the average partial effect of a marginal change in a continuous treatment on the individuals located at different parts of the outcome distribution, irrespective of individuals' characteristics. OASD combines both features of ATE and QTE: it is interpreted as straightforwardly as ATE while at the same time more granular than ATE by breaking the entire population up according to the rank of the outcome distribution.…
Debiased machine learning is a meta algorithm based on bias correction and sample splitting to calculate confidence intervals for functionals, i.e. scalar summaries, of machine learning algorithms. For example, an analyst may desire the confidence interval for a treatment effect estimated with a neural network. We provide a nonasymptotic debiased machine learning theorem that encompasses any global or local functional of any machine learning algorithm that satisfies a few simple, interpretable conditions. Formally, we prove consistency, Gaussian approximation, and semiparametric efficiency by …
This article proposes different tests for treatment effect heterogeneity when the outcome of interest, typically a duration variable, may be right-censored. The proposed tests study whether a policy 1) has zero distributional (average) effect for all subpopulations defined by covariate values, and 2) has homogeneous average effect across different subpopulations. The proposed tests are based on two-step Kaplan-Meier integrals and do not rely on parametric distributional assumptions, shape restrictions, or on restricting the potential treatment effect heterogeneity across different subpopulatio…
This paper develops a nonparametric model that represents how sequences of outcomes and treatment choices influence one another in a dynamic manner. In this setting, we are interested in identifying the average outcome for individuals in each period, had a particular treatment sequence been assigned. The identification of this quantity allows us to identify the average treatment effects (ATE's) and the ATE's on transitions, as well as the optimal treatment regimes, namely, the regimes that maximize the (weighted) sum of the average potential outcomes, possibly less the cost of the treatments. …
In this paper, we introduce and prove asymptotic normality for a new nonparametric estimator of continuous treatment effects. Specifically, we estimate the average dose-response function - the expected value of an outcome of interest at a particular level of the treatment level. We utilize tools from both the double debiased machine learning (DML) and the automatic double machine learning (ADML) literatures to construct our estimator. Our estimator utilizes a novel debiasing method that leads to nice theoretical stability and balancing properties. In simulations our estimator performs well com…
This paper analyzes the effect of a discrete treatment Z on a duration T. The treatment is not randomly assigned. The confounding issue is treated using a discrete instrumental variable explaining the treatment and independent of the error term of the model. Our framework is nonparametric and allows for random right censoring. This specification generates a nonlinear inverse problem and the average treatment effect is derived from its solution. We provide local and global identification properties that rely on a nonlinear system of equations. We propose an estimation procedure to solve this sy…
Massachusetts Institute of Technology（麻省理工学院）；Microsoft Research New England（微软新英格兰研究院）
Many causal and policy effects of interest are defined by linear functionals of high-dimensional or non-parametric regression functions. $\sqrt{n}$-consistent and asymptotically normal estimation of the object of interest requires debiasing to reduce the effects of regularization and/or model selection on the object of interest. Debiasing is typically achieved by adding a correction term to the plug-in estimator of the functional, which leads to properties such as semi-parametric efficiency, double robustness, and Neyman orthogonality. We implement an automatic debiasing procedure based on aut…
Many economic and causal parameters depend on nonparametric or high dimensional first steps. We give a general construction of locally robust/orthogonal moment functions for GMM, where moment conditions have zero derivative with respect to first steps. We show that orthogonal moment functions can be constructed by adding to identifying moments the nonparametric influence function for the effect of the first step on identifying moments. Orthogonal moments reduce model selection and regularization bias, as is very important in many applications, especially for machine learning first steps. We gi…
