贫困问题在中国受到很大关注,准确测度贫困非常重要。本文对大多文献采用的GQ模型拟合洛伦茨曲线来测算中国农村贫困的做法提出了质疑,结论认为当某些居民的纯收入为负数时,采用GQ模型拟合洛伦茨曲线是有偏的,以此计算的贫困指数与实际贫困存在系统性偏差。依据我国农村居民收入的分组数据,GQ模型测算的贫困发生率高估了实际数值,而贫困缺口指数和平方贫困缺口指数则被低估了。
Poverty is important in China, and how to get the real value of pov- erty is surely very vital. GQ model is usually fitted to Lorenz curve, and the figure of China rural poverty is calculated from this model by many experts. This paper disapproves of that method. When someone's income is negative, GQ model is biased for Lorenz curve, and the poverty will be inexact. According to GQ model and the group data of China rural income, the poverty head-count index is overestimated, the poverty gap index and the squared poverty gap index are underestimated.