Some Sampling Properties of a Model for Income Distribution
推导了基于Burr分布的三参数收入分配模型最大似然估计的大样本性质，并用基尼系数等不平等指标检验模型有效性及不平等趋势，分析了数据分组和删截方式对估计的影响。
A recently proposed model for describing the distribution of income over a population, based on the Burr distribution, has been shown to fit better than the commonly used lognormal or gamma distributions. The current article extends that analysis by deriving the large-sample properties of the maximum likelihood estimates for this three-parameter model. Consequently, resulting confidence intervals for some measures of income inequality (including the Gini index) are used to further test the model's validity, as well as to examine apparent trends in inequality over time. Since these properties depend on the way the income data are grouped and censored, implications for choosing data-report intervals can be analyzed. Specifically, a choice between two common methods of reporting the data is shown to have an important impact on Gini index estimates.