在中国区域内差异对区域发展的影响

Effects of Intraregional Disparities on Regional Development in China

摘要

本文分析的是中国在1989年和2001年之间发展和影响的区域性差异性。分解分析法表明intra-provincial差异对总区域不平等作出了重大贡献。在论文的第二部分,观察intra-provincial差距的影响区域经济发展是解决。利用省级面板数据对工业增长、资本和就业不平等的影响在工业增长估计影响技术效率和技术水平结果显示intra-provincial差距的一个重要积极影响省级工业增长,与因果关系不平等的增长。此外,看来inequality-growth关系不是线性的,而是极端不平等的变化的影响更强,比适度增加不平等更重要。这些结果是强劲的替代模型规范和控制变量。

This paper analyzes the development and effects of intra-provincial regional disparities in China between 1989 and 2001. A decomposition analysis shows that intra-provincial disparities contribute significantly to total regional inequality. In the second part of the paper, the impact of the observed intra-provincial disparities on regional economic development is addressed. Using provincial panel data on industrial growth, capital and employment, the impact of inequality on industrial growth is estimated as affecting technical efficiency and level of technology. The results show a significant positive effect of intra-provincial disparities on provincial industrial growth, with causality from inequality to growth. Moreover, it appears that the inequality-growth relationship is not a linear one, but rather that the impact of extreme changes in inequality is stronger and more significant than a moderate increase in inequality. These outcomes are robust to alternative model specifications and control variables.

关键词：不平等、 分解 、增长 ,面板数据 ,中国

Keyword: Inequality, Decomposition, Growth, Panel Data, China

目录

1. 测量和数据问题 3

2． 数据源和选择 4

3． 方法 4

3.2分解分析 5

4分解结果和解释 6

结论 7

1. 测量和数据问题

高等院校学生学习成绩的评价是非常重要的。大学生学习成绩综合评价是教育评价的重要内容，通过课程成绩综合评定，不但可以反映学生各种能力和掌握知识的程度，掌握学生专业知识和技能水平的高低差异状况，而且可以反映学生成绩水平在本班级所处的层次，也是教育管理工作中评优定先、评定各种奖学金的重要指标。 因此，科学、合理、公平地对学习成绩进行综合评价，不但有利于优化教学管理，推进高校学风建设，而且可以有效地避免学生在择优选先的竞争中产生不必要的矛盾。

当前在高校教学管理实践中，对学生学习成绩综合评价的方法主要有原始分累加法、简单算术平均法、平均学分绩法、平均学分积法等，这些方法对学生成绩进行综合评价及排名都缺乏公平合理性。由于各门课程在教学体系中地位的不同、课程难易程度不同、教学质量不同、考试的难易程度不同，课程成绩间虽然分数相等，而实际则不同，不同课程成绩之间不具有可比性，也不具备简单算术运算的功能，原始分累加法和算术平均法则忽视了这一点，严重缺乏科学性和公平性；平均学分绩法和学分积法貌似公平，其实各课程学分的设定缺乏科学依据，主观性太强。

针对以上的情况，探索科学合理的学习成绩综合评价方法具有重要的现实意义。

选择一个不平等的措施

各种指标可用于分析收入差距。虽然没有共识的文学不平等的措施是最可取的,适当的指标可以选择考虑两个主要评估标准:

最常见的方法来判断的适用性不平等的措施是通过比较他们的行为和一些公理理论推导等更好的性能的措施。通过这种方法来解决的主要问题是排序的合理性标准,这对于任何有意义的不平等是必要的比较(见范字段和1978:1978)。
提出了多个系统的公理,不同的广泛的数量和强度要求的公理。通常包括:
(没有收入以外的其他个人特征确定排序原则),
——独立或收入规模同质性(增加收入相同的积极的标量不会改变不平等)
——人口独立或人口同质性(复制每个收入不可或缺的次数不会改变不平等),
——原则或Pigou-Dalton条件转移(转移从富有到贫穷的人减少测量不平等)。

令人惊讶的是,只有少数措施能满足普遍接受公理。这些包括变异系数、基尼系数、阿特金森的措施,最后广义熵家庭的措施,最突出的example.14 Theil指数

(2)给定一个不平等的措施满足上面提到的公理一般条件,最终选择合适的测量可以考虑特定上下文中的每个测量的实用性。这不仅涉及问题如数据可用性和计算的复杂性,而且一个不平等的衡量可以提供额外的属性,并主要取决于特定目的的测量计算。摘要sub-provincial区域不平等的重要性应当评估;因此一个适当的不平等的一个重要属性测量将是它的可分解性。两种类型的解决分解:分解的子组(如地区、人口的子组,等等)和收入来源(如工资收入、财产、补贴,等等)。16,分解为子群组件的目的是。为了这个目的,一个理想的测量需要两个分解性质:17子群一致性(这意味着积极响应能力整体不平等的衡量不平等水平的变化组成团体,看到森1997,p . 157)作为最低要求;和添加剂可分解性(总体不平等是所有团体之间和群体内部的不平等)的总和作为额外的限制。这两个属性在一起只是满足Theil指数和MLD指数。18因此,这些措施之一将是应用于下面的分解分析。基尼系数不是可分解的子群一致性;指数的分解会产生一个额外的交互项,只要子组重叠,与地区收入一样。然而,考虑到广受欢迎,否则有利的基尼系数的属性,它将提供作为总不平等的衡量和跨区域不平等的比较。

2． 数据源和选择

在这项研究中,国际米兰和区域内差异综述了通过分析样本215个城市超过13年的时间里从1989年到2001年。本文使用的数据主要取自“新中国五十年的城市”由中国国家统计局(National Bureau of Statistics)发布(1999 b),一组历史数据在城市层面,主要是基于数据从各个问题的城市统计年鉴。在这个报告的时间序列数据来源来自1990年到1998年。
一般来说,在中国城市数量的增加。虽然在1991年,只有479个城市在城市统计年鉴》数据,这一数字已升至662个城市在2001.20这一增长主要是由于许多县城城镇到县城城市的升级。21的广泛和不同覆盖的城市在这些出版物中,我不得不限制示例一组一致的城市以保证可比性。因此,只有城市的所有必要的信息是可用的整个时期。例外只有在这种情况下,以前独立城市合并或者只是名称发生变化,至于这些变化成为明显的数据来源。这数据选择减少了，观测的数量只有215人(662年与662年相比可用的观测)。然而,即使采取了谨慎,谨慎的数据选择,可能发生小矛盾,因为一些时间序列的相同位置显示相对较高的波动性,特别是人口数据的情况下,和可能的显著改变大小的司法管辖区或inner-provincial组织政府。这些来源的主要数据包括:

——人口数据:年终人口(nianmo宗庆后renkou);和年平均人口(年pingyue renkou,只为1989;其他值计算平均连续两年);
收入数据:GDP(guonei shengchan我们)和值/百分比为二级,三级部门;
——就业数据:年终总就业(nianmo quanbu congye renyuansu),二、三级部门和百分比。
分解的目的,省是全世界分为三个宏观地区后影响常见的分类。-

沿海:北京、天津、河北、辽宁、上海、江苏、浙江、福建、山东、广东、广西、海南;
——中央:山西、内蒙古、吉林、黑龙江、安徽、江西、河南、湖北、湖南;
西方:四川、重庆、贵州、云南、西藏、陕西、甘肃、青海、宁夏、新疆。
其中,北京、天津、上海和省青海被排除在分析sub-provincial水平由于无效的数据,和重庆和西藏由于缺失的数据。

3． 方法
我上面描述的数据集,计算基尼系数和Theil指数,将后者分解为三个区域不平等的组件:全世界之间的不平等个宏观地区,影响全世界不平等在个宏观地区省份之间的影响,在省和不平等。下一节介绍的定义应用措施。

3.1总体不平等的措施:
测量总不平等,两项措施将应用:基尼系数和广义熵衡量Theil指数的形式。
森(1997)后,基尼指数G通常定义为:

（1）

并且 = Gini coefficient, = number of individuals, = average income (mean), = income of person , = income of person

然而,这个公式是专为个人数据。因为数据在考虑这是分组,此外,分区不平等,将使用下列公式计算的基尼分组数据:

（2）

并且 = number of groups, = cumulative population, = total population, = cumulative income , and = total income.

广义熵类不平等的措施如第二个指标被定义为:

（3）

对于所有的 ; 并且，对于 Theil Index :

（4）

对于分组数据,Theil-Index是一个典型的计算方法。

（5）

并且。 = total income of the group, = absolute frequency of population in the group; and as total income over all groups, and as total population.

这Theil指数比较每组的相对收入份额全部人口的相对份额。这两个指标将值0到1之间完美的平等或浓度分别为收入。自这两个指标满足上述分配公理,结果不应该对不同的命令替代分布相同的数据集。然而,他们不同的权重附加到一个特定的收入分布,因此在他们的基本测量不平等。Theil指数,灵敏度转移在不同收入类定义的参数alpha;= 1;这意味着低收入群体的相对加大。在不平等的分析中,这种统计特性通常被认为是积极的,因为类似的收入转移被认为更重要的个人效用较低的分布。基尼系数的敏感性取决于个人的相对位置相比其他个人(所谓的无关紧要的替代品)。因此,如果有更多的人在较低的收入分配的情况通常是这样——这些低收入将更有重量。

另一个可能的指标之间的区别是他们的敏感性不同的样本大小。汗和布伦(2001,第163页)声称Theil指数样本大小非常敏感。减少产生的问题属性,观察的数量随着时间一直保持不变;然而,obvisously,组之间的直接比较复杂。

3.2分解分析
一)基本概念
上面定义的Theil指数可以很容易地分解为组组件之间的内部和不同群体的收入接收器,或者——在这种情况下,地区。这种分解是常见的公式。

（6）

剩余内容已隐藏，支付完成后下载完整资料

Effects of Intraregional Disparities on Regional Development in China:

Inequality Decomposition and Panel-Data Analysis

By Reuter, Ulrich*

ABSTRACT

This paper analyzes the development and effects of intra-provincial regional disparities in China between 1989 and 2001. A decomposition analysis shows that intra-provincial disparities contribute significantly to total regional inequality. In the second part of the paper, the impact of the observed intra-provincial disparities on regional economic development is addressed. Using provincial panel data on industrial growth, capital and employment, the impact of inequality on industrial growth is estimated as affecting technical efficiency and level of technology. The results show a significant positive effect of intra-provincial disparities on provincial industrial growth, with causality from inequality to growth. Moreover, it appears that the inequality-growth relationship is not a linear one, but rather that the impact of extreme changes in inequality is stronger and more significant than a moderate increase in inequality. These outcomes are robust to alternative model specifications and control variables.

Keywords: Inequality, Decomposition, Growth, Panel Data, China

JEL-Codes: O11, O15, O40, O53, R11

June 23, 2004

II. 2. Measurement Issues and Data

II. 2.1. The choice of an inequality measure

Various indicators can be used in the analysis of income disparities. Although there is no consensus in the literature on which inequality measure is the most preferable, appropriate indicators could be chosen considering two main evaluation criteria:

- consistency with distributional and welfare axioms;

- practicability considerations.

(i) The most common way to judge the applicability of inequality measures is by comparing their behavior with some axioms theoretically derived as preferable properties of such measures. The main issue addressed by this approach is the reasonability of the ordering criterion; which is necessary for any meaningful inequality comparison (see Fields and Fei 1978: 315).

Several systems of axioms have been proposed, differing widely in the number and strength of their required axioms. Usually they include:

- anonymity (no personal characteristics other than the income determine the ordering principle),

- scale independence or income homogeneity (multiplying all incomes with the same positive scalar does not change inequality)

- population independence or population homogeneity (replicating each income an integral number of times does not change inequality),

- the transfer principle or Pigou-Dalton condition (transfers from a richer to a poorer person do reduce the measured inequality).

Surprisingly, only a few measures can satisfy the commonly accepted axioms. These include the coefficient of variation, the Gini coefficient, the Atkinson class of measures, and finally the generalized entropy family of measures, with the Theil index as the most prominent example.14

(ii) Given an inequality measure fulfils the above stated axiomatic general conditions, a final choice of the appropriate measure can be made considering the practicability of each measure in the specific context. This relates not only to problems like data availability and complexity of calculation, but also to additional properties that an inequality measure can offer, and depends mainly on the specific purpose for which the measure is calculated. In this paper, the importance of sub-provincial regional inequality shall be assessed; therefore an important property of an appropriate inequality measure would be its decomposability. In the context of decomposability, two types of decomposition are addressed: decomposition by subgroups (e.g. regions, population subgroups, etc.) and by income source (e.g. income from wages, property, subsidies, etc.).16 In this paper, decomposition into subgroup components is intended. For this purpose, a desirable measure requires two decomposition properties:17

- subgroup consistency (which means the positive responsiveness of the overall inequality measure to changes in the inequality levels of constituent groups, see Sen 1997, p. 157) as a minimum requirement; and

- additive decomposability (overall inequality is the sum of all between-groups and within-groups inequality) as an additional restriction.

These two properties together are only satisfied by the Theil index and the MLD index.18 Therefore,

one of these measures will be applied in the following decomposition analysis.

The Gini coefficient is not decomposable in the sense of subgroup consistency; a decomposition of the index will produce an additional interaction term as long as the subgroups are overlapping, as is the case with regional incomes. However, given the wide popularity and the otherwise favorable properties of the Gini index, it will be supplied as a measure for total inequality and cross-regional inequality comparisons.

II. 2.2. Data Sources and Selection

In this study, inter- and intraregional disparities are reviewed by analyzing a sample of 215 cities over a 13 year period from 1989 to 2001. The data used in this article is mainly taken from “Fifty Years of Cities in New China” published by the National Bureau of Statistics of China (1999b), a collection of historic data on the city level, mainly based on data from the various issues of the Urban Statistical Yearbooks. The time series data reported in this source cover the period from 1990 till 1998.

I also used the Urban Statistics Yearbooks for additional data (volumes 1990, 2000, 2001 and 2002) and for the crosschecking and verification purposes (various issues).

Generally, the number of cities in China increases over time.

剩余内容已隐藏，支付完成后下载完整资料

资料编号：[286979]，资料为PDF文档或Word文档，PDF文档可免费转换为Word