统计研究 ›› 2020, Vol. 37 ›› Issue (1): 62-73.doi: 10.19343/j.cnki.11-1302/c.2020.01.005

• • 上一篇    下一篇

样本选择参数分位回归模型及其在工资分布分解中的应用

邰凌楠 钱曼玲 田茂再   

  1.  
  • 出版日期:2020-01-25 发布日期:2020-02-28

Sample Selection Parametric Quantile Regression and Its Application in Distribution Decomposition of Wages

Tai Lingnan  Qian Manling  Tian Maozai   

  1.  
  • Online:2020-01-25 Published:2020-02-28

摘要: 在工资差距分解问题中,研究者经常会遇到样本选择偏差问题,直接忽略会导致最终估计结果产生严重偏差,同时在众多工资差距分解方法中,相比于均值分解,分布分解方法更受研究者青睐。针对参数分位回归,本文首次提出可加形式与非可加形式的样本选择参数分位回归(SSPQR)模型,并基于这两类样本选择参数分位回归模型给出修正样本选择偏差后的参数分位回归工资差距分布分解方法。运用上述方法及已有的工资分布分解方法,借助CHNS2015年度城镇数据,本文研究了我国城镇男女工资差距及差距分解问题,得出以下结论:①男女工资差距主要来源是性别歧视问题;②经过样本选择偏差修正后,实际的工资差距更大,歧视问题更严重;③男女工资差距程度在不同分位点上结果不同,换句话说,我们不能简单地仅从平均水平来判断工资差距程度;④与其他已有方法计算结果比较发现,SSPQR计算的工资差距程度更大。

关键词: 样本选择偏差, 参数分位回归, 工资差距, 分布分解

Abstract: In the study of decomposition of the wage gap, the researchers often encounter the problem of sample selection bias and ignoring it will lead to a serious bias in the final estimate. At the same time, the distribution decomposition method is more popular among researchers than the mean decomposition method. This paper firstly proposes separable and nonseparable sample selection parametric quantile regression (SSPQR), and constructs the distribution decomposition method of wage gap based on sample selection parametric quantile regression after correcting the sample selection bias. Using the above methods for decomposing wage gap and the CHNS 2015 urban data, we study the wage gap between men and women. Main research conclusions are as follows: the main source of the wage gap between men and women is gender discrimination; after correcting sample selection bias, the actual wage gap is even greater, and discrimination is even more serious; the gender wage gap has different results at different quantiles, in other words, we cannot simply judge the level of the wage gap from the average level; the result of wage gap by SSPQR is larger than other methods.

Key words: Sample Selection Bias, Parametric Quantile Regression, Wage Gap, Distribution Decomposition