12 区域发展差距：成因与走向(国家统计局青海调查总队 陈健、祁瑞芳、梅科)

青海调查总队参赛队员：陈健、祁瑞芳、梅科目录摘要························································································································1一、研究背景······································································································2二、划分地域单元·····························································································2㈠资料来源·································································································2㈡划分依据·································································································2㈢方法和结果····························································································3三、我国区域发展差距现状与趋势····························································4㈠分析方法·································································································4㈡泰尔指数计算························································································4㈢泰尔指数所反映的区域发展差距变化情况·································5㈣预测区域发展差距走向······································································6四、我国区域发展差距成因分析·································································7㈠软因素分析····························································································71.模型选择························································································72.因素选择························································································73.数据整理························································································84.模型计算························································································85.结论·································································································12㈡硬因素分析····························································································121.地理位置因素···············································································122.历史性因素···················································································13五、政策建议······································································································14参考文献···············································································································16附录········································································································171区域发展差距：成因与走向摘要改革开放以来，中国经济持续快速增长，综合国力显著提高，人民生活水平极大改善，但同时，地区间差异也在逐渐拉大，区域发展不平衡现象日益突出。这不仅为政策制定者所注意，也引起经济界学者的广泛关注。本文通过分析近二十年来我国区域经济发展的差异，剖析差距形成原因，就如何促进区域经济协调发展问题，作进一步研究。文章共分五部分：第一部分研究背景；第二部分应用聚类分析法划分地域单元；第三部分应用泰尔指数分析我国区域经济发展差异变化的时间和空间分布特征，预测其发展趋势；第四部分从“软、硬”两方面，分析影响及形成我国区域经济发展差异的因素。这一部分首先根据区域分层建立变系数的面板数据回归模型，从“软”因素方面进行分析，其次从地理及历史性因素等“硬”因素方面进行分析；第五部分政策建议。本文创新之处：一是应用聚类分析法按经济发展水平对我国经济区域进行了新的分层；二是将变系数的面板数据回归模型应用于对区域经济发展差距的研究。本文不足之处：所选模型在趋势预测方面有所欠缺。关键词：区域差异泰尔指数面板数据主成份分析2一、研究背景改革开放以来，中国经济持续快速增长，国内生产总值从1978年的3624.1亿元增加到2007年的246619亿元，占世界经济的份额，也从1978年的1.8％提高到2007年的5.5％，仅次于美国、日本和德国，位居世界第四位。但地区之间发展速度却具有不同步性，在沿海和内陆之间、东部和中西部之间、城乡之间差距明显。尽管经济发展差距是客观存在和不可避免的，但是差距过大势必会影响到国民经济的健康发展。区域发展差距已成为中国经济社会发展的重大挑战，也引起学者们的广泛关注。近年来，我国制定了规模庞大的西部大开发战略、中部崛起计划和振兴东北工业区方针，在具体的问题上，专家学者们不断提出方法建议来促进和确保政策的实施，使国民经济在各地区能够协调发展。因此，进行我国经济区域发展差距及成因研究具有较强现实意义。二、划分地域单元㈠资料来源本文数据除每万人专利授权数是从国家数据库提取加工而成，其余均是从大赛组委会提供的数据加工而成。由于重庆直辖市1997年才正式成立，西藏自治区部分数据缺失，因此本文在分析中剔除了这两个地区的数据，仅从29个省、市(区)的数据来进行分析。文章在个别地方会用到专家学者的研究成果作为本文分析的论据。㈡划分依据3我们通常在实际研究中，习惯把全国各省区分为“东中西”三大地带来分析，这种划分方法往往会将地域内的差异“平均化”,但是由于中国幅员辽阔,各省区之间在地理条件、资源禀赋、经济和社会发展上都存在巨大差距,以此作为分析基础所得出的结论就显得过于笼统。为克服以上缺陷,一些学者开始以省区作为分析的地域单元，这虽然克服了前者缺陷,但是由于分析的地域单元与省区重合,中央政府的区域经济政策就会演变为“一省一策”，将无法对国家宏观区域经济政策有所贡献。因此,上述两种划分地域单元的分析方法都存在较为明显的缺陷。本文尝试采用聚类分析法，按各省区GDP发展水平将我国经济区域划分为三类：发达地区、中等发达地区和欠发达地区，作为地域差异研究的地域单元。如此分类，既能克服以“三大地带”作为分析地域单元所产生的区域内差异被“平均化”的缺陷，又能避免以省区作为分析地域单元导致的“一省一策”的不足。㈢方法和结果通过对1988-2007年29个省、市（区）人均GDP做快速聚类分析，初始聚类中心人工定为4个，结果见表1。表1:各聚类样本数聚类样本数聚类17聚类24聚类32聚类416有效样本29缺失样本04观察最终聚类中心(附录1)，由于聚类２、３样本数较少，可人工归划为一类，即最终聚类分三层（见表2）：表2：分层结果表第一层次发达地区（６个）北京、天津、上海、江苏、浙江、广东第二层次中等发达地区（７个）河北、内蒙、辽宁、吉林、黑龙江、福建、山东第三层次欠发达地区（１６个）山西、安徽、江西、河南、湖北、湖南、广西、海南、四川、贵州、云南、陕西、甘肃、青海、宁夏、新疆注：人均GDP分层加权计算结果见附录2三、我国区域发展差距现状与趋势㈠分析方法度量区域发展差距的常用指标有全距、库兹奈司比率、基尼系数、泰尔指数等。经比较，本文选用泰尔指数分析法，因为泰尔指数即能反映总体差异又能解释组间差异和组内差异，更适合本文的研究需要。㈡泰尔指数计算假设D为全国人均GDP的泰尔指数:logniiiiYDyP=∑logijijipiijjiiYYYTNYN⎛⎞⎜⎟⎛⎞=⎜⎟⎜⎟⎝⎠⎜⎟⎝⎠∑i=1,2,3其中，n为地区数，ijY为第i部分第j地区的GDP产量，iY为第i部分的GDP总量，Y为整个目标区的GDP总量。同样，ijN为第i部分第j地区人口数，iN为第i部分的人口总量，N为第i部分的人口总数。iP为i部分人口数占整个目标区的比例。5假设wrT和BRT分别表示组内差异和组间差异，那么：312123iwrpipppiYYYYTTTTTYYYY⎛⎞==×+×+×⎜⎟⎝⎠∑312312123loglogloglogiiBRiiYYYYYYYYYYYYTNNNNYYYYNNNN⎛⎞⎛⎞⎛⎞⎛⎞⎛⎞⎜⎟⎜⎟⎜⎟⎜⎟==++⎜⎟⎜⎟⎜⎟⎜⎟⎜⎟⎝⎠⎜⎟⎜⎟⎜⎟⎜⎟⎝⎠⎝⎠⎝⎠⎝⎠∑注：D系数及相关指数计算结果见附录3㈢泰尔指数所反映的区域发展差距变化情况图1：历年D系数0.0000.0100.0200.0300.0400.0500.0600.0701988199019921994199619982000200220042006TwrTbrD图2：组内和组间D系数百分比01020304050607080901988199019921994199619982000200220042006Twr百分比Tbr百分比6从图1、图2可以看出,我国人均GDP泰尔指数从整体上看呈现上升趋势，组间泰尔指数对总体D系数的贡献率从1994年后一直保持高位运行，并且有进一步增大的可能，即表示分层地区之间从1994年后发展差距一直很大，并且有进一步扩大可能，而同一层次的省份之间的差距略有缩小。㈣预测区域发展差距走向利用已知20年的D系列数据建立指数平滑模型，分析线性趋势，用马克威软件对Twr、Tbr进行未来４年预测（无季节变化），预测值见表3。表3：2008-2011年Twr、Tbr值预测结果年份TwrTwr百分比TbrTbr百分比D2008年0.012823.0