基于分位数处理效应的面板数据政策评价方法

Panel Data Policy Evaluation Method Based on Quantile Processing Effect

学者们在估计不同群体受到同一政策的影响时,通常需要使用协变量或者工具变量。为了在缺少协变量和工具变量的情况下估计一项政策对不同群体产生的影响,文章将分位数处理效应与面板数据政策评价方法相结合,得到一种新的政策影响效应估计方法——基于分位数处理效应的面板数据政策评价方法,并进一步研究了政策影响估计量的方差缩减和分解方法。新方法得到的政策影响估计量具有一致性和渐近正态性,交叉验证方法能够缩减估计量的方差,政策影响估计量可以分解为共有影响和特有影响。新的政策影响估计方法为学者们在缺少工具变量和协变量的情况下研究一项政策对不同群体产生的差异化影响提供了帮助。

When estimating the effect of the same policy on different groups,scholars usually need to use covariates or instrumental variables.In order to estimate the effect of a policy on different groups in the absence of covariates and instrumental variables,this paper combines quantile processing effect with panel data policy evaluation method to obtain a new policy effect estimation method:panel data policy evaluation method based on quantile processing effect,and further studies the variance reduction and decomposition methods of policy effect estimators.The policy effect estimators obtained by the new method are consistent and asymptotically normal.The cross-validation method can reduce the variance of the estimators,and the policy effect estimators can be decomposed into common effects and specific effects.The new policy effect estimation method helps scholars to study the differentiated impact of a policy on different groups in the absence of instrumental variables and covariates.