我国近年来收入差距持续扩大，收入不平等的加剧也成为人们关注的焦点。基尼系数作为国内常用的一种衡量收入不平等的措施，并不能很好得反映中国由于高收入人群收入份额增长而带来的收入不平等变化。本文重点分析了传统基尼系数计算方法的缺陷，进一步利用Gastwirth提出的基于中位数的基尼系数，即用收入分布的中位数替代基尼系数公式中的均值，以弥补基尼系数对收入不平等估计不准确以及受均值影响较大的问题。本文使用了CHIPS家庭入户调查数据，并对调查数据存在的高收入群体收入偏差问题进行了修正，具体方法是通过帕累托分布对前0.1%收入群体的分布进行取代。另外，对于调查数据中城乡地区抽样不均的问题，本文采取重新构建地区和省份的权重和对农村样本等比例抽取子样本的方法进行解决。

对比本文计算得到的2002年和2013年两年的基尼系数和基于中位数的基尼系数，发现基于中位数的基尼系数在这十一年间的变化要明显大于基尼系数，基于中位数的基尼系数反映的从2002年到2013年中国收入不平等的增长速度是基尼系数的两倍。另外，2013年的 均大于2002年的 ， 大于 ，在这期间，国内居民平均收入与收入中位数的差距拉大，高收入群体的收入增速要快于中等收入群体的收入增速，基于中位数的基尼系数对基尼系数的修正使它更容易反映高收入群体收入增长带来的收入不平等变化。接下来，本文将国内分为区域和省份分别计算了基尼系数和基于中位数的基尼系数，发现直辖市的基尼系数和基于中位数的基尼系数在2002年至2013年间增长更快，东部地区和西部地区的不平等程度相比其他地区增长缓慢。各省份的基尼系数和基于中位数的基尼系数的变化率在2002年至2013年间均发生增长，除四川省外，其他省份的基于中位数的基尼系数的变化率均大于基尼系数的变化率。另外，基于中位数的基尼系数反映出各省份之间收入不平等程度的差距也进一步增大。

In recent years, China's income gap has continued to widen, and the increase in income inequality has become the focus of attention. The Gini coefficient, as a commonly used measure of income inequality in China, does not reflect China's income inequality changes due to the increase in the income share of high-income people. This article focuses on the analysis of the shortcomings of the traditional Gini coefficient calculation method, and further uses the median-based Gini coefficient proposed by Gastwirth, that is, the median of the income distribution is used to replace the average value of the Gini coefficient formula to compensate for the Gini coefficient of income inequality estimates Problems of inaccuracy and greater influence by the mean. This paper uses the CHIPS household survey data, and corrects the income bias of high-income groups in the survey data. The specific method is to replace the distribution of the top 0.1% income groups through the Pareto distribution. In addition, for the problem of uneven sampling in urban and rural areas in the survey data, this paper takes the method of reconstructing the weights of regions and provinces and extracting equal proportions of rural samples to solve the problem.

Comparing the calculated Gini coefficients for 2002 and 2013 and the median-based Gini coefficients calculated in this paper, it is found that the change of the median-based Gini coefficients during these eleven years is significantly greater than that of the Gini coefficients. The Gini coefficient reflects the growth rate of China's income inequality from 2002 to 2013 is twice the Gini coefficient. In addition, AMR in 2013 were greater than those in 2002， were greater than . During this period, the gap between the average income and the median income of domestic residents widened, and the income growth rate of high-income groups was faster than that of middle-income groups The income growth rate of Gini based on the median Gini coefficient makes it easier to reflect the income inequality changes caused by the income growth of high-income groups. Next, this paper divides the country into regions and provinces, and calculates the Gini coefficient and the median-based Gini coefficient, respectively, and finds that the Gini coefficient and the Gini coefficient of the municipalities grow faster between 2002 and 2013. Compared with the inequality in the western region, the growth in other regions is slow. The Gini coefficient of each province and the rate of change of the median-based Gini coefficient have increased between 2002 and 2013. Except for Sichuan Province, the rate of change of the median-based Gini coefficient of other provinces is greater than the change of the Gini coefficient rate. In addition, the Gini coefficient based on the median reflects that the gap in income inequality between provinces has further increased.

By comparing the gap between the income share of low-income groups, middle-income groups and high-income groups, it is found that for the p-quantile population with the highest income, the smaller the p-value is selected, the more the income share of this part of the population increases from 2002 to 2013 Quickly, the change in the Gini coefficient based on the median is very close to the change in the income share of the top 5% income group. Bikini coefficient has the problem of being more sensitive to income changes of middle-income groups. The Gini coefficient based on the median can better reflect the impact of changes in income share of high-income groups on income distribution.