无条件分位数处理效应方法及其应用

Unconditional Quantile Treatment Effects and Application in Policy Evaluation

研究目标:政策评估在我国经济社会研究中备受关注,但所采用的方法以平均处理效应居多,分位数处理效应鲜见。与后者相关的理论和应用研究与国外尚存在较大差距,亟待跟进。研究方法:采用蒙特卡罗模拟方法比较了条件分位数处理效应(CQTE)与UQTE方法和其他UQTE方法之间的异同。研究发现:如果回归中只包含常数项和处理变量,CQTE与UQTE是等价的;如果个体在处理组和控制组中的相对位置保持不变,CQTE和UQTE都可以得到无偏估计。研究创新:通过最低工资标准提升政策对居民工资分布影响的实例说明UQTE方法在政策评价中的应用。研究价值:基于断点回归和双差分模型对新的UQTE拓展方法进行探索性展望。

Research Objectives： Policy evaluation is one of main concerns in economic and social re- searches of our country. However, much attention has been paid to average treatment effect, rather than quantile treatment effect （QTE）, and there is a big gap between domestic and foreign research on the estimation and application of QTE, which need to be followed. First of all, we introduce three existing literature, and use Monte Carlo simulation experiments to compare CQTE with UQTE method, and among UQTE methods. Results show that if only constant term and treatment variable are contained in regression, it is t equivalent be- tween CQTE and UQTE; If the relative place of individuals is fixed in treatment group and control group, both two methods can get unbiased estimation. Then we research the policy effect of minimum wage increasing on residents＇ wage distribution, for an example to illus- trate application of UQTE in policy evaluation. Finally, we conclude and give some possible directions like UQTE based on Regression Discontinuity, Difference in Difference and oth- ers.