一类与洛伦兹曲线相关的不平等指数测度对比研究

发布时间：2018-01-09 03:21

本文关键词：一类与洛伦兹曲线相关的不平等指数测度对比研究　出处：《山西财经大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着经济的快速发展会伴生一系列的经济问题,不平等现象的加剧一直是经济学家研究的重点,不平等问题的存在是导致各国经济、社会、甚至政治问题的根本原因,因此各国一直以来都从未间断对于不平等的研究。国外对于不平等的研究早已有成熟的方法理论,但我国目前有关不平等指数方法的文献还停留在基尼系数、泰尔指数等传统的指数上。为了突破这一障碍,本文引进一类与洛伦兹曲线相关并具有优良性质的不平等测度指数:Bonferroni指数(简称B指数)、De Vergottini指数(简称V指数)、Zenga指数(简称Z指数)和Canberra指数(简称C指数)这四种不平等指数,它们都可以基于洛伦兹曲线的调整表示,与传统的Gini系数相结合,揭示这些指数间的关联,探讨了这几种指数所具备的一些性质,并基于CFPS微观数据库中的家庭人均收入数据运用各指数进行了测度分析,以深入了解各指数的测度差异。本文首次对一类与洛伦兹曲线相关的不平等指数进行了系统的介绍,通过文献研究法整理了各指数的构建理论、不平等曲线的定义及其不平等指数所应具备的一些性质。在此基础上通过对公式的理论推导与实证分析相结合的方法揭示了各指数之间的关联,对各指数所具备的性质作了对比分析,并阐释了这几种不平等指数的测度差异。本文的创新点主要在于通过理论分析与实证分析结合的方法从不平等测度水平值的差异、不平等动态变化测度的差异和不平等分解测度的差异三方面探究了这一类与洛伦兹曲线相关的不平等指数测度差异。从不平等测度水平值的差异中得出:不同的不平等指数构造理念不同,测度结果也就会存在一定的差异,分析各指数的构造理念可知这几种不平等指数在数量上存在一种大小关系:Gini(7)C(8)?B?Z?V。从不平等动态变化测度差异中发现:各不平等指数所关注的群体不同,基尼系数对中高等收入群体更敏感,B指数和C指数对低收入群体较敏感,V指数对高收入群体较敏感,而Z指数对高收入群体与低收入群体同样关注。对于不平等分解测度的差异,主要从指数分解的彻底性和分解方法的不同两方面分析了差异。从各指数分解的彻底性上来看,只有Z指数具备加和分解性,可以分解彻底,而其他指数都分解不干净,有不可解释的余项,政策指导意义不明确。此外,只有Z指数具备子要素和子群的联合分解这一独特性。从各指数分解方法的不同可得出:仅采用各组收入份额进行加权分解会大大减弱组间不平等贡献,若采用各组人口份额进行加权分解则会提高组间不平等贡献,且采用收入份额加权与人口份额加权的结果可能会截然相反。最后据本文对各指数的差异分析得到关于不平等指数的选用及其方法上的一些启发,并分析了本文的不足之处。
[Abstract]:With the rapid development of economy will accompany a series of economic problems, the increase of inequality has been the focus of economists, the existence of inequality is leading to the national economy, society. Even the root causes of political problems, so countries have never stopped the study of inequality. Foreign research on inequality has long been a mature theory of methods. However, the literature about inequality index method in our country still stays on the traditional index such as Gini coefficient, Thiel index and so on. In order to break through this obstacle. In this paper, we introduce a class of inequality measure index: Bonferroni index (B index), which is related to Lorentz curve and has good properties. De Vergottini index (Z index) and Canberra index (C index) are four kinds of inequality index. Both of them can be combined with the traditional Gini coefficient based on the adjustment representation of Lorentz curve to reveal the correlation between these indices and to discuss some properties of these indices. And based on the data of household per capita income in CFPS microcosmic database, this paper uses each index to measure and analyze. In order to understand the measurement difference of each index, this paper firstly introduces a class of inequality index related to Lorentz curve, and collates the construction theory of each index by literature research method. Based on the definition of inequality curve and some properties of inequality index, the relationship between the indices is revealed by combining the theoretical derivation of formula with empirical analysis. The properties of each index are compared and analyzed. The innovation of this paper mainly lies in the difference of the measurement level of inequality through the combination of theoretical analysis and empirical analysis. This paper explores the difference of inequality index measure related to Lorentz curve from three aspects: the difference of the measure of dynamic variation of inequality and the difference of measure of inequality decomposition. Different inequality index constructs different ideas. There will be some differences in the measurement results. By analyzing the structural concepts of each index, we can see that these inequality indices have a size relation in quantity. B? Z? From the difference of dynamic change of inequality, it is found that the groups concerned with inequality index are different, and Gini coefficient is more sensitive to middle and higher income group than B index and C index to low income group. V index is more sensitive to high income group, while Z index pays attention to high income group as well as low income group. The difference is mainly analyzed from two aspects: the thoroughness of exponential decomposition and the difference of decomposition methods. From the perspective of the thoroughness of each exponential decomposition, only Z exponent has additive and decomposable property, which can be decomposed thoroughly. The other indicators are not clean, there is an unexplained remainder, policy guidance is not clear. In addition. Only Z index has the uniqueness of the joint decomposition of subelements and subgroups. From the different methods of exponential decomposition, it can be concluded that using only the income share of each group for weighted decomposition will greatly reduce the contribution of inequality among groups. If the population share of each group is weighted and decomposed, the contribution of inequality among groups will be increased. And the results of weighted income share and population share may be completely opposite. Finally, according to the analysis of the differences of each index, some enlightenments on the selection and method of inequality index are obtained. The deficiency of this paper is analyzed.
【学位授予单位】：山西财经大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F113.8

【参考文献】