异方差实验报告

附件二：实验报告格式（首页）山东轻工业学院实验报告成绩课程名称计量经济学指导教师实验日期2013.5.18院（系）商学院会计系专业班级会计实验地点实验楼二机房学生姓名学号同组人无实验项目名称异方差的检验一、实验目的和要求1、理解异方差的含义后果、2、学会异方差的检验与加权最小二乘法要求熟悉基本操作步骤，读懂各项上机榆出结果的含义并进行分析3、掌握异方差性问题出现的来源、后果、检验及修正的原理，以及相关的Eviews操作方法4、练习检查和克服模型的异方差的操作方法。5、掌握异方差性的检验及处理方法6、用图示法、斯皮尔曼法、戈德菲尔德、white验证法，验证该模型是否存在异方差二、实验原理1、异方差的检验出消除方法2、运用EVIEWS软件及普通最小二乘法进行模型估计3、检验模型的异方差性并对其进行调整三、主要仪器设备、试剂或材料Eviews软件、课本教材、电脑四、实验方法与步骤一、准备工作。建立工作文件，并输入数据，用普通最小二乘法估计方程（操作步骤与方法同前），得到残差序列。1、CREATEU131回车2、DATAYX回车输入数据obsYX126487772105921039099544131105085122109796107119127406127478503134999431142691058815522118981673012950176631377918575148191963515122221163161702228801715782412718165425604191400265002018292676021220028300222017274302321052956024160028150252250321002624203250027257035250281720335002919003600030210036200312800382003、LSYCX回车用最小二乘法进行估计出现DependentVariable:YMethod:LeastSquaresDate:05/18/13Time:11:19Sample:131Includedobservations:31VariableCoefficientStd.Errort-StatisticProb.C-700.411116.6679-6.003460X0.0878310.00482718.195750R-squared0.919464Meandependentvar1266.452AdjustedR-squared0.916686S.D.dependentvar846.757S.E.ofregression244.4088Akaikeinfocriterion13.8979Sumsquaredresid1732334Schwarzcriterion13.99042Loglikelihood-213.418F-statistic331.0852Durbin-Watsonstat1.089829Prob(F-statistic)0用普通最小二乘法进行估计，估计结果如下iYˆ=﹣700.41+0.087831XiR2=0.922R=0.92F=335.82t=(-6.0)(18.2)括号内为t统计量。β1=0.087431说明在其他因素不变的情况下，可支配收入每增长1元，个人储蓄平均增长0.087431元。2R=0.92,拟合程度较好。在给定=0.05时，t=18.2)29(025.0t=2.055，拒绝原假设，说明销售收入对销售利润有显著性影响。F=335.82)9,21(F05.0=4.18,表明方程整体显著。(一).图示检验法分别绘制X、Y坐标系散点图，命令如下：ScatxyGenre2=resid^2Scatxe2出现050010001500200025003000010000200003000040000XY可以看出，随着可支配收入x的增加，储蓄y的离散程度增加，表明随机误差项ui存在异方差性。（二）斯皮尔曼等级相关系数检验命令scatxe2sortxdataxdd1（输入1-31）lsycx出现DependentVariable:YMethod:LeastSquaresDate:05/18/13Time:11:25Sample:131Includedobservations:31VariableCoefficientStd.Errort-StatisticProb.C-700.4110116.6679-6.0034580.0000X0.0878310.00482718.195750.0000R-squared0.919464Meandependentvar1266.452AdjustedR-squared0.916686S.D.dependentvar846.7570S.E.ofregression244.4088Akaikeinfocriterion13.89790Sumsquaredresid1732334.Schwarzcriterion13.99042Loglikelihood-213.4175F-statistic331.0852Durbin-Watsonstat1.943262Prob(F-statistic)0.000000sortxdataxdd1lsycxgenre1=abs(resid)sorte1datae1dd2genrr=1-6*@sum((dd2-dd1)^2)/(31^3-31))genrz=r*@sqrt(30)出现3.326即等级相关数是显著的，说明储蓄计量模型的随机误差项存在异方差性。R=0.607258z=3.326089给定显著性水平α=0.05，查正态分布表，得96.12αZ,因为Z=3.33＞1.96，所以拒绝H0,接受H1,即等级相关系数是显著的，说明储蓄计量模型的随机误差项存在异方差性。(三)、Goldfeld-Quant检验命令sortxsmpl111lsycxDependentVariable:YMethod:LeastSquaresDate:05/18/13Time:11:36Sample:111Includedobservations:11VariableCoefficientStd.Errort-StatisticProb.C-744.6351195.4108-3.8106140.0041X0.0882580.0157055.6196190.0003R-squared0.778216Meandependentvar331.3636AdjustedR-squared0.753574S.D.dependentvar260.8157S.E.ofregression129.4724Akaikeinfocriterion12.72778Sumsquaredresid150867.9Schwarzcriterion12.80012Loglikelihood-68.00278F-statistic31.58011Durbin-Watsonstat1.142088Prob(F-statistic)0.000326smpl2131lsycxDependentVariable:YMethod:LeastSquaresDate:05/18/13Time:11:39Sample:2131Includedobservations:11VariableCoefficientStd.Errort-StatisticProb.C666.3811911.25850.7312760.4832X0.0457790.0278981.6409710.1352R-squared0.230295Meandependentvar2152.909AdjustedR-squared0.144772S.D.dependentvar354.4462S.E.ofregression327.7867Akaikeinfocriterion14.58557Sumsquaredresid966997.0Schwarzcriterion14.65791Loglikelihood-78.22063F-statistic2.692786Durbin-Watsonstat2.743586Prob(F-statistic)0.135222计算F=rss2\rss1=0.2960F0.05(9,9)=3.18，说明储蓄计量模型的随机误差项存在异方差.记下第一个残差平方和：150867.9记下第二个残差平方和：966997.0。计算F=6.41，给定显著性水平α=0.05，查F分布表V1=V2=11-2=9,F0.05(9,9)=3.18,因为F=6.41＞3.18，所以接受备择假设，即储蓄计量模型的随机误差项存在异方差性。(四).White检验命令smpl2131lsycxsmpl131lsycx出现DependentVariable:YMethod:LeastSquaresDate:05/18/13Time:11:43Sample:131Includedobservations:31VariableCoefficientStd.Errort-StatisticProb.C-700.4110116.6679-6.0034580.0000X0.0878310.00482718.195750.0000R-squared0.919464Meandependentvar1266.452AdjustedR-squared0.916686S.D.dependentvar846.7570S.E.ofregression244.4088Akaikeinfocriterion13.89790Sumsquaredresid1732334.Schwarzcriterion13.99042Loglikelihood-213.4175F-statistic331.0852Durbin-Watsonstat1.943262Prob(F-statistic)0.000000WhiteHeteroskedasticityTest:F-statistic5.819690Probability0.007699Obs*R-squared9.102584Probability0.010554TestEquation:smpl131lsycxDependentVariable:RESID^2Method:LeastSquaresDate:05/18/13Time:11:46Sample:131Includedobservations:31VariableCoefficientStd.Errort-StatisticProb.C19975.9882774.930.2413290.8111X-2.1986328.094419-0.2716230.7879X^20.0001460.0001760.8300460.4135R-squared0.293632Meandependentvar55881.73AdjustedR-squared0.243177S.D.dependentvar77875.67S.E.ofregression67748.39Akaikeinfocriterion25.17675Sumsquaredresid1.29E+11Schwarzcriterion25.31553Loglikelihood-387.2397F-statistic5.819690Durbin-Watsonstat2.580140Prob(F-statistic)0.007699Obs*R-squared=9.102584X2(0.05)=6.0，所以结论是该回归模型中存在异方差.因为TR2=31×0.2936=9.1﹥0.6)2(205.0，所以结论是该回归模型中存在异方差.其中obs*R-squared等于9.102584表示的就是统计量TR2的值。(五)克服异方差命令smpl131lsycxD