计量经济学上机实验报告(异方差性)

整理文档很辛苦,赏杯茶钱您下走!

免费阅读已结束,点击下载阅读编辑剩下 ...

阅读已结束,您可以下载文档离线阅读编辑

资源描述

《计量经济学》上机实验报告三题目:检验异方差性实验日期和时间:班级:学号:姓名实验室:实验环境:WindowsXP;EViews3.1实验目的:掌握异方差性的检验及处理方法实验内容:建立并检验四川省农村人均纯收入(X)与人均生活费支出(Y)的函数模型实验步骤:一:检验异方差性⒈图形分析检验⑴观察销售利润(Y)与销售收入(X)的相关图(图1):SCATXY01000200030004000010002000300040005000XY从图中可以看出,随着农村人均纯收入的增加,人均消费支出不断提高,但是离散程度不大,这不能说明变量之间可能存在递增的异方差性,需要进行别的检验方法。⑵残差分析首先将数据排序(命令格式为:SORT解释变量),然后建立回归方程。在方程窗口中点击Resids按钮就可以得到模型的残差分布图(或建立方程后在Eviews工作文件窗口中点击resid对象来观察)。-100010020001000200030004000808590950005ResidualActualFitted上图显示回归方程的残差分布有明显的扩大趋势,即表明存在异方差性。⒉Goldfeld-Quant检验⑴将样本安解释变量排序(SORTX)并分成两部分(分别有1978到1988共11个样本合1998到2008共11个样本)⑵利用样本1建立回归模型1(回归结果如图3),其残差平方和为1285.598。SMPL19781988LSYCXDependentVariable:YMethod:LeastSquaresDate:11/25/13Time:08:53Sample:19781988Includedobservations:11VariableCoefficientStd.Errort-StatisticProb.C-19.6653411.22945-1.7512290.1138X0.9699700.03946624.577610.0000R-squared0.985320Meandependentvar241.7309AdjustedR-squared0.983688S.D.dependentvar93.57987S.E.ofregression11.95175Akaikeinfocriterion7.962598Sumsquaredresid1285.598Schwarzcriterion8.034942Loglikelihood-41.79429F-statistic604.0588Durbin-Watsonstat0.823583Prob(F-statistic)0.000000⑶利用样本2建立回归模型2(回归结果如图4),其残差平方和为29558.82。SMPL19982008LSYCXDependentVariable:YMethod:LeastSquaresDate:11/25/13Time:08:55Sample:19982008Includedobservations:11VariableCoefficientStd.Errort-StatisticProb.C54.5240062.053210.8786650.4024X0.7590360.02353432.252280.0000R-squared0.991422Meandependentvar1976.724AdjustedR-squared0.990469S.D.dependentvar587.0207S.E.ofregression57.30893Akaikeinfocriterion11.09776Sumsquaredresid29558.82Schwarzcriterion11.17010Loglikelihood-59.03766F-statistic1040.209Durbin-Watsonstat1.124021Prob(F-statistic)0.000000⑷计算F统计量:12/RSSRSSF=29558.82/1285.598=22.992,21RSSRSS和分别是模型1和模型2的残差平方和。取05.0时,查F分布表得F0.05(11-1-1,11-1-1)=3.18,而F=22.992F0.05(11-1-1,11-1-1)=3.18,所以存在异方差性⒊White检验⑴建立回归模型:LSYCX,回归结果如图DependentVariable:YMethod:LeastSquaresDate:11/25/13Time:08:50Sample:19782008Includedobservations:31VariableCoefficientStd.Errort-StatisticProb.C71.6140718.157593.9440290.0005X0.7614450.01088169.982270.0000R-squared0.994113Meandependentvar1030.406AdjustedR-squared0.993911S.D.dependentvar850.2099S.E.ofregression66.34631Akaikeinfocriterion11.28999Sumsquaredresid127653.1Schwarzcriterion11.38251Loglikelihood-172.9949F-statistic4897.518Durbin-Watsonstat0.401829Prob(F-statistic)0.000000⑵在方程窗口上点击View\ResidualTests\HeteroskedastcityTests,在弹出的对话框中选择White,并点击OK,检验结果如图WhiteHeteroskedasticityTest:F-statistic1.631313Probability0.213725Obs*R-squared3.235218Probability0.198372TestEquation:DependentVariable:RESID^2Method:LeastSquaresDate:11/25/13Time:09:02Sample:19782008Includedobservations:31VariableCoefficientStd.Errort-StatisticProb.C730.63222245.2710.3254090.7473X6.0093293.3491881.7942640.0836X^2-0.0015010.000910-1.6483110.1105R-squared0.104362Meandependentvar4117.843AdjustedR-squared0.040388S.D.dependentvar6279.224S.E.ofregression6151.116Akaikeinfocriterion20.37842Sumsquaredresid1.06E+09Schwarzcriterion20.51719Loglikelihood-312.8655F-statistic1.631313Durbin-Watsonstat0.883445Prob(F-statistic)0.213725其中F值为辅助回归模型的F统计量值。取显著水平05.0,由于χ0.05(2)=5.993.235218,所以不存在异方差性。实际应用中可以直接观察相伴概率p值的大小,若p值较小,则认为存在异方差性。反之,则认为不存在异方差性。⒋Park检验⑴建立回归模型(结果同图5所示)。⑵生成新变量序列:GENRLNE2=log(RESID^2)GENRLNX=log(X)⑶建立新残差序列对解释变量的回归模型:LSLNE2CLNX,回归结果如图7所示。DependentVariable:LNE2Method:LeastSquaresDate:11/25/13Time:09:05Sample:19782008Includedobservations:31VariableCoefficientStd.Errort-StatisticProb.C4.9412511.8100772.7298560.0107LNX0.3688680.2677471.3776730.1788R-squared0.061427Meandependentvar7.406587AdjustedR-squared0.029063S.D.dependentvar1.538091S.E.ofregression1.515576Akaikeinfocriterion3.731808Sumsquaredresid66.61211Schwarzcriterion3.824324Loglikelihood-55.84303F-statistic1.897982Durbin-Watsonstat0.853362Prob(F-statistic)0.178844从图7所示的回归结果中可以看出,LNX的系数估计值不为0且不能通过显著性检验,即随即误差项的方差与解释变量不存在较强的相关关系,即认为不存在异方差性。⒌Gleiser检验(Gleiser检验与Park检验原理相同)提示:打包保存时自己的文件夹以“学号姓名”为文件夹名,打包时文件夹内容包括:本实验报告、EViews工作文件。

1 / 5
下载文档,编辑使用

©2015-2020 m.777doc.com 三七文档.

备案号:鲁ICP备2024069028号-1 客服联系 QQ:2149211541

×
保存成功