第六章练习题及参考解答表是中国1985-2016年货物进出口贸易总额()与国内生产总值()的数据。
表中国进出口贸易总额和国内生产总值单位:亿元年份货物进出口贸易总额(Y)国内生产总值(X)年份货物进出口贸易总额(Y)国内生产总值(X)19852001 19862002 19872003 19882004 19892005 19902006 19912007 19922008 19932009 19942010 19952011 199620121997 1998 1999 20002013 2014 2015 2016资料来源:《中国统计年鉴2017》(1)建立货物进出口贸易总额的对数对国内生产总值的对数的回归方程;(2)检测模型的自相关性;(3)采用广义差分法处理模型中的自相关问题。
【练习题参考解答】回归结果自相关检验①图示法图1、2 与的散点图以及模型残差图由上面两个图可以发现模型残差存在惯性表现,很可能存在正自相关。
②DW检验由回归结果可知DW统计量为,同时,在的显着性水平下,,因而模型中存在正相关。
③BG检验阶数5432AICSIC滞后阶数从5阶减小到2阶,AIC及SIC达到最小时,滞后阶数为2阶,此时,已知,,同时P值为,在的显着性水平下拒绝原假设,即存在自相关。
表2 BG检验2阶回归结果自相关补救①DW反算法求由,可知,可得广义差分方程:表3 广义差分结果-DW反算法DW检验:由回归结果可知DW统计量为,同时,在的显着性水平下,,即已消除自相关。
BG检验:阶数5432AICSIC滞后阶数从5阶减小到2阶,AIC及SIC达到最小时,滞后阶数为2阶,此时,已知,,同时P值为,在的显着性水平下不拒绝原假设,即已消除自相关。
表4 广义差分BG检验2阶回归结果则可知,最终模型为:②残差过原点回归求Dependent Variable: EMethod: Least SquaresDate: 02/07/18 Time: 20:48Sample (adjusted): 1986 2016Included observations: 31 after adjustmentsVariable Coefficient Std. Error t-Statistic Prob.E(-1)R-squared Mean dependent varAdjusted R-squared. dependent var. of regression Akaike info criterionSum squared resid Schwarz criterionLog likelihood Hannan-Quinn criter.Durbin-Watson stat表5 残差序列过原点回归结果回归结果为:,可知。
进而得广义差分方程:lnDependent Variable: *LNY(-1)Method: Least SquaresDate: 02/07/18 Time: 20:51Sample (adjusted): 1986 2016Included observations: 31 after adjustmentsVariable Coefficient Std. Error t-Statistic Prob.C*LNX(-1)R-squared Mean dependent varAdjusted R-squared. dependent var. of regression Akaike info criterionSum squared resid Schwarz criterionLog likelihood Hannan-Quinn criter.F-statistic Durbin-Watson statProb(F-statistic)表6 广义差分-残差序列过原点回归结果DW检验:由回归结果可知DW统计量为,同时,在的显着性水平下,,因而模型已不存在自相关。
BG检验:阶数5432AICSIC滞后阶数从5阶减小到2阶,AIC及SIC达到最小时,滞后阶数为2阶,此时,已知,,同时P值为,在的显着性水平下不拒绝原假设,即已消除自相关。
Breusch-Godfrey Serial Correlation LM Test:F-statistic Prob. F(2,27)Obs*R-squared Prob. Chi-Square(2)Test Equation:Dependent Variable: RESIDMethod: Least SquaresDate: 02/07/18 Time: 21:30Sample: 1986 2016Included observations: 31Presample missing value lagged residuals set to zero.Variable Coefficient Std. Error t-Statistic Prob.C*LNX(-1)RESID(-1)RESID(-2)R-squared Mean dependent varAdjusted R-squared. dependent var. of regression Akaike info criterionSum squared resid Schwarz criterionLog likelihood Hannan-Quinn criter.F-statistic Durbin-Watson statProb(F-statistic)广义差分BG检验2阶回归结果则可知,最终模型为:③德宾两步法求构建模型Dependent Variable: LNYMethod: Least SquaresDate: 02/07/18 Time: 21:43Sample (adjusted): 1986 2016Included observations: 31 after adjustmentsVariable Coefficient Std. Error t-Statistic Prob.CLNXLNX(-1)LNY(-1)R-squared Mean dependent varAdjusted R-squared. dependent var. of regression Akaike info criterionSum squared resid Schwarz criterionLog likelihood Hannan-Quinn criter.F-statistic Durbin-Watson statProb(F-statistic)德宾两步法回归结果由此可知,,进而得广义差分方程:lnDependent Variable: *LNY(-1)Method: Least SquaresDate: 02/07/18 Time: 22:03Sample (adjusted): 1986 2016Included observations: 31 after adjustmentsVariable Coefficient Std. Error t-Statistic Prob.C*LNX(-1)R-squared Mean dependent varAdjusted R-squared. dependent var. of regression Akaike info criterionSum squared resid Schwarz criterionLog likelihood Hannan-Quinn criter.F-statistic Durbin-Watson statProb(F-statistic)广义差分-德宾两步法回归结果DW检验:由回归结果可知DW统计量为,同时,在的显着性水平下,,因而模型已不存在自相关。
BG检验:阶数5432AICSIC滞后阶数从5阶减小到2阶,AIC及SIC达到最小时,滞后阶数为2阶,此时,已知,,同时P值为,在的显着性水平下不拒绝原假设,即已消除自相关。
Breusch-Godfrey Serial Correlation LM Test:F-statistic Prob. F(2,27)Obs*R-squared Prob. Chi-Square(2)Test Equation:Dependent Variable: RESIDMethod: Least SquaresDate: 02/07/18 Time: 22:16Sample: 1986 2016Included observations: 31Presample missing value lagged residuals set to zero.Variable Coefficient Std. Error t-Statistic Prob.C*LNX(-1)RESID(-1)RESID(-2)R-squared Mean dependent varAdjusted R-squared. dependent var. of regression Akaike info criterionSum squared resid Schwarz criterionLog likelihood Hannan-Quinn criter.F-statistic Durbin-Watson statProb(F-statistic)广义差分BG检验2阶回归结果则可知,最终模型为:④科克兰·奥科特迭代法Dependent Variable: LNYMethod: Least SquaresDate: 02/07/18 Time: 22:38Sample (adjusted): 1986 2016Included observations: 31 after adjustmentsConvergence achieved after 16 iterationsVariable Coefficient Std. Error t-Statistic Prob.CLNXAR(1)R-squared Mean dependent varAdjusted R-squared. dependent var. of regression Akaike info criterionSum squared resid Schwarz criterionLog likelihood Hannan-Quinn criter.F-statistic Durbin-Watson statProb(F-statistic)Inverted AR Roots.88科克兰·奥科特迭代法回归结果DW检验:由回归结果可知DW统计量为,同时,在的显着性水平下,,因而模型已不存在自相关。