搜索
实验报告课程名称：计量经济学实验项目：实验五异方差模型的检验和处理实验类型：综合性□设计性□验证性专业班别：姓名：学号：实验课室：指导教师：石立实验日期：广东商学院华商学院教务处制一、实验项目训练方案小组合作：是□否小组成员：无实验目的：掌握异方差模型的检验和处理方法实验场地及仪器、设备和材料实验室：普通配置的计算机，Eviews软件及常用办公软件。实验训练内容（包括实验原理和操作步骤）：【实验原理】异方差的检验：图形检验法、Goldfeld-Quanadt检验法、White检验法、Glejser检验法；异方差的处理：模型变换法、加权最小二乘法(WLS)。【实验步骤】本实验考虑三个模型：【1】广东省财政支出CZ对财政收入CS的回归模型；（数据见附表1：附表1-广东省数据）【2】广东省固定资产折旧ZJ对国内生产总值GDPS和时间T的二元回归模型；（数据见附表1：附表1-广东省数据）【3】广东省各市城镇居民消费支出Y对人均收入X的回归模型。（数据见附表2：附表2-广东省2005年数据）（一）异方差的检验1.图形检验法分别用相关分析图和残差散点图检验模型【1】、模型【2】和模型【3】是否存在异方差。注：①相关分析图是作因变量对自变量的散点图（亦可作模型残差对自变量的散点图）；②残差散点图是作残差的平方对自变量的散点图。③模型【2】中作图取自变量为GDPS来作图。模型【1】相关分析图残差散点图模型【2】相关分析图残差散点图模型【3】相关分析图残差散点图【思考】①相关分析图和残差散点图的不同点是什么？残差散点图更能看得出是否存在异方差②*在模型【2】中，自变量有两个，有无其他处理方法？尝试做出来。（请对得到的图表进行处理，以上在一页内）04008001,2001,6002,00005001,0001,5002,0002,500CZCS04,0008,00012,00016,00020,00024,00028,00005001,0001,5002,000CSE205001,0001,5002,0002,5003,0003,5004,00005,00010,00015,00020,00025,000GDPSZJ04,0008,00012,00016,00020,00024,00028,00005,00010,00015,00020,00025,000GDPSE224,0008,00012,00016,00020,00024,00028,00032,0004,0008,00012,00016,00020,00024,000YX02,000,0004,000,0006,000,0008,000,00010,000,00012,000,0005,00010,00015,00020,00025,00030,000XE22.Goldfeld-Quanadt检验法用Goldfeld-Quanadt检验法检验模型【3】是否存在异方差。注：Goldfeld-Quanadt检验法的步骤为：①排序：②删除观察值中间的约1/4的，并将剩下的数据分为两个部分。③构造F统计量：分别对上述两个部分的观察值求回归模型，由此得到的两个部分的残差平方为21ie和22ie。21ie为较大的残差平方和，22ie为较小的残差平方和。④算统计量)2)(,2)((~2221*kcnkcnFeeFii。⑤判断：给定显著性水平0.05，查F分布表得临界值)()2)(,2)((kcnkcnF。如果)()2)(,2)((*kcnkcnFF，则认为模型中的随机误差存在异方差。（详见课本135页）将实验中重要的结果摘录下来，附在本页。前7个：DependentVariable:X1Method:LeastSquaresDate:05/30/14Time:15:26Sample:17Includedobservations:7CoefficientStd.Errort-StatisticProb.Y-0.1006660.093257-1.0794510.3297C9534.7081265.1477.5364430.0007R-squared0.188998Meandependentvar8243.949AdjustedR-squared0.026798S.D.dependentvar1108.275S.E.ofregression1093.324Akaikeinfocriterion17.06679Sumsquaredresid5976790.Schwarzcriterion17.05134Loglikelihood-57.73376Hannan-Quinncriter.16.87578F-statistic1.165214Durbin-Watsonstat0.498709Prob(F-statistic)0.329683后7个：DependentVariable:X1Method:LeastSquaresDate:05/30/14Time:15:27Sample:1118Includedobservations:8CoefficientStd.Errort-StatisticProb.Y-0.5311420.224692-2.3638660.0560C24718.362614.8859.4529440.0001R-squared0.482217Meandependentvar19428.64AdjustedR-squared0.395920S.D.dependentvar4923.100S.E.ofregression3826.362Akaikeinfocriterion19.54953Sumsquaredresid87846273Schwarzcriterion19.56939Loglikelihood-76.19814Hannan-Quinncriter.19.41558F-statistic5.587862Durbin-Watsonstat0.656404Prob(F-statistic)0.055988（请对得到的图表进行处理，以上在一页内）3.White检验法分别用White检验法检验模型【1】、模型【2】和模型【3】是否存在异方差。Eviews操作：先做模型，选view/ResidualTests/WhiteHeteroskedasticity(nocrossterms/crossterms)。摘录主要结果附在本页内。模型【1】HeteroskedasticityTest:WhiteF-statistic1.704061Prob.F(1,26)0.2032Obs*R-squared1.722264Prob.Chi-Square(1)0.1894ScaledexplainedSS3.165058Prob.Chi-Square(1)0.0752TestEquation:DependentVariable:RESID^2Method:LeastSquaresDate:05/30/14Time:15:34Sample:19782005Includedobservations:28CoefficientStd.Errort-StatisticProb.C1368.250878.66281.5571960.1315CS^20.0012660.0009701.3053970.2032R-squared0.061509Meandependentvar1940.891AdjustedR-squared0.025414S.D.dependentvar4080.739S.E.ofregression4028.552Akaikeinfocriterion19.50895Sumsquaredresid4.22E+08Schwarzcriterion19.60411Loglikelihood-271.1253Hannan-Quinncriter.19.53804F-statistic1.704061Durbin-Watsonstat1.686396Prob(F-statistic)0.203192模型【2】HeteroskedasticityTest:WhiteF-statistic4.703427Prob.F(2,25)0.0185Obs*R-squared7.655216Prob.Chi-Square(2)0.0218ScaledexplainedSS12.87226Prob.Chi-Square(2)0.0016TestEquation:DependentVariable:RESID^2Method:LeastSquaresDate:05/30/14Time:15:35Sample:19782005Includedobservations:28CoefficientStd.Errort-StatisticProb.C-1311.4142086.268-0.6285930.5353GDPS^2-1.26E-052.00E-05-0.6305180.5341T^220.427359.8654832.0705880.0489R-squared0.273401Meandependentvar3461.910AdjustedR-squared0.215273S.D.dependentvar7240.935S.E.ofregression6414.371Akaikeinfocriterion20.47143Sumsquaredresid1.03E+09Schwarzcriterion20.61416Loglikelihood-283.6000Hannan-Quinncriter.20.51506F-statistic4.703427Durbin-Watsonstat1.930056Prob(F-statistic)0.018458模型【3】HeteroskedasticityTest:WhiteF-statistic14.65680Prob.F(1,16)0.0015Obs*R-squared8.605673Prob.Chi-Square(1)0.0034ScaledexplainedSS13.32535Prob.Chi-Square(1)0.0003TestEquation:DependentVariable:RESID^2Method:LeastSquaresDate:05/30/14Time:15:36Sample:118Includedobservations:18CoefficientStd.Errort-StatisticProb.C-632050.0656900.8-0.9621700.3503X^20.0083160.0021723.8284200.0015R-squared0.478093Meandependentvar1232693.AdjustedR-squared0.445474S.D.dependentvar2511199.S.E.ofregression1870002.Akaikeinfocriterion31.82522Sumsquaredresid5.60E+13Schwarzcriterion31.92415Loglikelihood-284.4269Hannan-Quinncriter.31.83886F-statistic14.65680Durbin-Watsonstat1.938966Prob(F-statistic)0.001481（请对得到的图表进行处理，以上在一页内）4.Glejser检验法用Glejser检验法检验模型【1】是否存在异方差。分别用残差的绝对值对自变量的一次项iCS、二次项2iCS，开根号项iCS和倒数项iCS1作回归。检验异方差是否存在，并选定异方差的最优形式。摘录主要结果附在本页内。（请对得到的图表进行处理，以上在一页内）（二）异方差的处理1.模型【1】中CZ对CS回归异方差的处理已