tStd.Errort-StatisticProb.C-4554.8052253.323-2.0213730.0615X1-0.0012270.012050-0.1018200.9202X23.5106071.3910092.5237850.0234X31.3559166.7476780.2009460.8434X4636.6679355.14651.7926910.0932X5598.4430200.69662.9818290.0093R-squared0.99137Mean1787.4...7dependentvar10AdjustedR-squared0.988503S.D.dependentvar1691.457S.E.ofregression181.3664Akaikeinfocriterion13.47387从回归结果看,可决系数很高,通过F检验,但在显著性水平05.0??下,1X、3X和4X的回归系数并不显著,而且1X的系数符号与预期的相反,这表明该模型很可能存在多重共线性。(三)多重共线性的检验用EViews计算各解释变量的相关系数,得到相关系数矩阵:变量X1X2X3X4X5X110.99180.96840.98220.8406X20.991810.94280.96440.8058X30.96840.942810.96200.8931X40.98220.96440.962010.8640X50.840600.80580.89310.86381由相关系数矩阵可以看出,各个解释变量相互之间的相关系数较高,证实确实存在严重多重共线性。采用逐步回归的办法,去检验和解决多重共线性问题。运用OLS方法分别做Y对X1、X2、X3、X4、X5的一元回归,结果如下:变量X1X2X3X4X5参数估计0.0317225.69308249.278872390.4193071.836
全文预览