1中国各地区农村居民家庭人均现金支出指标的聚类分析摘要本文引入系统聚类分析,根据2009年上半年全国各地农村居民家庭人均现金支出指标,运用中间距离法对全国31个地区进行分类。在此基础上,对我国农村居民消费结构的区域性变化进行分析,说明未来我国农村居民消费的区域差距将进一步减小,同时提出缩小这一差距和提高农村居民收入的合理性建议。关键词农村居民消费水平人均现金支出指标系统聚类分析引言消费理论和需求理论是微观经济学理论体系中的重要组成部分。在市场经济条件下,消费活动是经济活动的重点,一切经济活动的目的就是为了满足人们日益增长的消费需求;但是另一方面,消费活动又是经济活动的起点,是拉动经济增长的动力。需求对生产起着导向作用,在当前各地农村居民需求不足的情况下,关于农村居民的消费需求结构状况及其变动的研究具有总要的现实意义。一、样本数据说明根据2009年上半年国家统计局抽样调查的31个地区农村居民家庭人均现金支出,将各地区家庭用于消费的人均现金支出的所有费用又分为10大类指标:期内现金支出指标、生产费用支出指标、家庭经营费用支出指标、农业生产支出指标、牧业生产支出指标、购买生产性固定资产支出指标、税费支出指标、生活消费现金支出指标、财产性支出指标、转移性支出指标(见表1)。单位:元地区X1X2X3X4X5X6X7X8X9X10北京5318.9730.2606.4124.2232.5123.71.94093.223.5470.2天津3267.81242.81172.9203.3699.569.91.21896.16.0121.6河北2289.6779.6725.0362.7226.254.63.11361.714.9130.3山西2080.9500.0441.3254.4126.358.70.71431.24.0145.1内蒙古3338.71426.91223.5865.7316.6203.31.71617.352.2240.6辽宁3926.21671.21546.2719.9754.5125.03.81803.738.7408.9吉林4517.72129.91836.91404.1393.3293.15.11741.0202.9438.7黑龙江4424.62047.01812.21405.0368.8234.82.91699.1301.8373.9上海5487.1241.3229.937.332.611.40.14616.71.4627.6江苏3412.9632.1561.8236.1137.570.312.82438.95.5323.7浙江5437.21320.21192.4159.4724.4127.85.53587.723.7500.1安徽2390.0554.9490.7297.2116.764.26.01666.23.0159.9福建3026.0626.4566.8251.0208.159.60.72168.210.8219.8江西2199.8680.6591.7339.7160.489.04.51328.023.5163.2山东3060.9969.5842.4423.3314.6127.15.71885.716.7183.3河南2163.8583.9523.9237.1228.860.00.61448.61.4129.3湖北2222.2677.5601.9266.8195.975.63.91471.34.664.8湖南2418.9555.8497.8196.7175.258.03.81587.92.8268.5广东2895.2576.3547.4171.3216.329.01.72155.69.5152.0广西1900.9697.4589.5348.5187.6107.92.91125.61.673.6海南1902.8656.1639.2324.3165.416.90.31187.22.157.1重庆1773.4449.1406.3188.7172.842.82.51107.30.9213.5四川2514.5612.8533.5178.0270.779.36.11669.57.0219.12贵州1432.8412.5324.7167.7122.287.71.4846.31.0171.6云南1837.1623.0532.9277.2185.690.11.51089.96.8115.9西藏1022.3203.992.353.28.8111.50.2800.70.117.5陕西2261.3580.1442.6240.6128.4137.53.91496.84.6175.9甘肃1622.3473.2391.2298.374.482.00.91068.24.375.7青海1753.8447.9348.0190.192.599.81.71196.57.4100.3宁夏2592.81029.8870.5406.2359.3159.30.21325.631.3205.8新疆2705.01444.41109.7813.2236.6334.71.01126.044.489.2表1中国2009年上半年各地区农村居民家庭人均现金支出指标值注:X1是期内现金支出指标,X2是生产费用支出指标,X3是家庭经营费用支出指标,X4是农业生产支出指标,X5是牧业生产支出指标,X6是购买生产性固定资产支出指标,X7是税费支出指标,X8是生活消费现金支出指标,X9是财产性支出指标,X10是转移性支出指标。二、各地农村居民消费水平的聚类分析(一)聚类分析的基本思想聚类分析的基本思想是首先将每个样本当作一类,然后根据样本之间的相似程度并类,当计算新类于其他类之间的距离,在选择相近的并类,每合并一次减少一类,继续这一过程,直到所有样本被合并成一类为止。所以,聚类分析依赖于对观测值的接近程度(距离)或相似程度,为此须定义距离。定义不同的距离和相似性量度会产生不同的聚类结果。本文所定义的距离是系统聚类分析当中的常用方法——中间距离。聚类分析分R型聚类分析和Q型聚类分析,前者用于指标分类,后者用于样本即个体聚类。(二)聚类分析的相关理论选取n个样品,这n个样品具有m个不同数值的相同指标,计算出各指标的均值、标准差和极差,列表2如下:指标样品1X……jX……mX)1(X11x……1jx……mx1)(iX1ix……ijx……imx)(nX1nx……njx……nmx均值1x……jx……mx标准差1s……js……ms极差1R……jR……mR表2样品指标的均值、标准差和极差3其中,均值.,,2,1,11tjmjxnxntj标准差2j1ijj)(11xxnsni。对样品指标作标准化变换:jjijijsxxx*,.,2,1;,,2,1mjni,定义n个样品间的距离欧氏距离2121][)2(mtjtitijxxd,),,2,1,(nji。计算n个样品间的距离,得样品间的距离矩阵(0)D。开始n个样品各自构成一类,这n个类为)i(XGi.,,2,1ni此时类间距离就是样品间的距离。合并类间距离最小的两类为一新类,此时类的总个数n减少1类,计算新类与其他类间的距离,得到新的距离矩阵。一直合并到类的总个数为1。(三)聚类分析过程本文的目的是把31个地区进行分类,反映不同地区消费水平的差异,把距离相近的归为一类。本文利用SAS软件,选用中间距离来度量类与类之间的相似程度,聚类方法使用R型聚类分析。其聚类程序如下:dataZ;inputgroup$X1-X10;cards;北京5318.9730.2606.4124.2232.5123.71.94093.223.5470.2天津3267.81242.81172.9203.3699.569.91.21896.16.0121.6河北2289.6779.6725.0362.7226.254.63.11361.714.9130.3山西2080.9500.0441.3254.4126.358.70.71431.24.0145.1内蒙古3338.71426.91223.5865.7316.6203.31.71617.352.2240.6辽宁3926.21671.21546.2719.9754.5125.03.81803.738.7408.9吉林4517.72129.91836.91404.1393.3293.15.11741.0202.9438.7黑龙江4424.62047.01812.21405.0368.8234.82.91699.1301.8373.94上海5487.1241.3229.937.332.611.40.14616.71.4627.6江苏3412.9632.1561.8236.1137.570.312.82438.95.5323.7浙江5437.21320.21192.4159.4724.4127.85.53587.723.7500.1安徽2390.0554.9490.7297.2116.764.26.01666.23.0159.9福建3026.0626.4566.8251.0208.159.60.72168.210.8219.8江西2199.8680.6591.7339.7160.489.04.51328.023.5163.2山东3060.9969.5842.4423.3314.6127.15.71885.716.7183.3河南2163.8583.9523.9237.1228.860.00.61448.61.4129.3湖北2222.2677.5601.9266.8195.975.63.91471.34.664.8湖南2418.9555.8497.8196.7175.258.03.81587.92.8268.5广东2895.2576.3547.4171.3216.329.01.72155.69.5152.0广西1900.9697.4589.5348.5187.6107.92.91125.61.673.6海南1902.8656.1639.2324.3165.416.90.31187.22.157.1重庆1773.4449.1406.3188.7172.842.82.51107.30.9213.5四川2514.5612.8533.5178.0270.779.36.151669.57.0219.1贵州1432.8412.5324.7167.7122.287.71.4846.31.0171.6云南1837.1623.0532.9277.2185.690.11.51089.96.8115.9西藏1022.3203.992.353.28.8111.50.2800.70.117.5陕西2261.3580.1442.6240.6128.4137.53.91496.84.6175.9甘肃1622.3473.2391.2298.374.482.00.91068.24.375.7青海1753.8447.9348.0190.192.599.81.71196.57.4100.3宁夏2592.81029.8870.5406.2359.3159.30.21325.631.3205.8新疆2705.01444.41109.7813.2236.6334.71.01126.044.489.2;procprintdata=Z;run;procclusterdata=Zmethod=medstdpseudocccouttree=N;varX1-X10;idgroup;proctreedata=Nhorizontalgraphics;run;聚类结果见表3和图1。ClusterHistoryNormTMedianiNCL--ClustersJoined--FREQSPRSQRSQERSQCCCPSFPST2Diste30甘肃青海20.0006.999..60.8.0.130429山西河南20.0007.999..57.9.0.141228广西云南20.0009.998..52.8.0.161327