IV估计的最优工具变量选取方法

The Optimal Choice of Instruments in IV Estimation

IV估计的有限样本性质对工具变量的选取十分敏感,尤其是存在弱工具变量的情形。本文在Donald和Newey(2001)的基础上研究了常用的IV估计———2SLS的最优工具变量选取方法。首先通过对2SLS估计量进行Nagar分解,从理论上推导出估计量的近似MSE表达式;根据这一表达式,提出IV估计的最优工具变量选取准则,并证明选取准则的渐近有效性。模拟结果表明,本文提出的工具变量选取准则能够极大地改善2SLS估计量的有限样本表现。本研究为实证中面临的工具变量选择问题提供了理论依据。

It is well known that finite sample properties of IV estimators are sensitive to the choice of instruments, especially when the instruments are weak. In this paper, we propose a novel procedure to select the optimal instruments for two-stage least squares （2SLS） based on Donald and Newey （2001） . First, we derive an approximation of the mean square error （MSE） of the 2SLS estimators through Nagar decomposition. According to this approximation, we propose a criterion of choosing instruments by minimizing the Nagar approximation of the MSE and prove the asymptotic efficiency. Finally, Monte Carlo simulations show that the proposed selection criterion an improve the finite sample properties of 2SLS significantly. This research provides a theoretical basis for the selection of instruments in empirical studies.