دانلود کتاب کلان داده در اقتصاد و مدیریت

The first part of this monograph focuses on the causal inference in observational studies. Chapter 1 starts the causal inference for a discrete treatment variable. Within the potential outcomes framework, we first introduce the average treatment effects (ATE) of a binary treatment and its identification methods under the unconfoundedness condition. To estimate the ATE, we introduce the calibration weighting method by exploiting the covariate balancing property of the propensity score and show that the calibration estimators are consistent and attain the semiparametric efficiency bound. Further, we introduce the semi-supervised machine learning method for estimating ATE. Finally, we discuss the estimation of the general causal effects of a multi-valued treatment variable based on the artificial neural networks, and show that the “curse of dimensionality” can be alleviated if nuisance functions belong to a mixed smoothness class. Chapter 2 extends the causal inference for a continuous treatment variable. We first introduce a general continuous treatment model that encompasses a wide variety of causal parameters of interest including average, quantile, and asymmetric least squares treatment effects. For this general continuous treatment model, we propose the identification method and derive the semiparametric efficiency bounds under the unconfoundedness condition. We then propose a covariate balancing estimation method that attains the general semiparametric efficiency bounds. Finally, we study various advanced causal testing problems for a continuous treatment. Chapter 3 concerns the causal inference for a continuous treatment variable in the presence of measurement errors. We first identify the average dose-response function (ADRF) for a continuously valued error contaminated treatment by a weighted conditional expectation. We then estimate the weights nonparametrically by maximizing a local generalized empirical likelihood subject to an expanding set of conditional moment equations incorporated into the deconvolution kernels. Thereafter, we construct a deconvolution kernel estimator of ADRF. We derive the asymptotic bias and variance of the proposed estimator and provide its asymptotic linear expansion, which helps conduct statistical inference. To select our smoothing parameters, we adopt the simulation-extrapolation method and propose a new extrapolation procedure to stabilize the computation.