2.自相关检验（数据来源于李子奈版课后习题P155.9）
运行Eviews,依次单击file→new→work file→Annual→strat1980 end2007。命令栏中输入“data y x”，打开“y x”表，接下来将数据输入其中。
开始进行LS回归，命令栏中输入“ls log(y) c log(x)”回车，即得到回归结果如下： Dependent Variable: LOG(Y) Method: Least Squares Date: 12/11/11 Time: 11:49 Sample: 1980 2007 Included observations: 28 Variable C LOG(X) R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient 1.588478 0.854415 Std. Error 0.134220 0.014219 t-Statistic 11.83492 60.09058 Prob. 0.0000 0.0000 9.552256 1.303948 -1.465625 -1.370468 3610.878 0.000000 0.992851 Mean dependent var 0.992576 S.D. dependent var 0.112351 Akaike info criterion 0.328192 Schwarz criterion 22.51875 F-statistic 0.379323 Prob(F-statistic) 杜宾瓦尔森检验法：
由结果得到，D.W值为0.379。查表得到dl=1.33,dw=1.48,D.W 图形检验法： 单击work file击窗口Genr,分别输入：e=resid e1=e(-1)；选中e、e1，右击Open as Group,在Group窗口依次单击Quick→ Graph→ Scatter，得到此图： .2.1.0E1-.1-.2-.3-.3-.2-.1E.0.1.2 由上图可知该模型存在序列相关性。 weibo.com/kouqintang 3.多重共线性检验（数据来源于李子奈版课后习题P155.10） 运行Eviews,依次单击file→new→work file→unstructed→observation 10。命令栏中输入“data y x1 x2”，打开“y x1 x2”表，接下来将数据输入其中。 逐步回归法： 以Y为被解释变量，逐个引入解释变量。首先分别让Y与x1与x2回归： 按照ls y c x1回归： Dependent Variable: Y Method: Least Squares Date: 12/11/11 Time: 12:23 Sample: 1 10 Included observations: 10 Variable C X1 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient 244.5455 0.509091 Std. Error 64.13817 0.035743 t-Statistic 3.812791 14.24317 Prob. 0.0051 0.0000 1110.000 314.2893 11.36135 11.42187 202.8679 0.000001 Coefficient 238.9949 0.049886 Std. Error 67.42146 0.003661 t-Statistic 3.544790 13.62516 Prob. 0.0076 0.0000 1110.000 314.2893 11.44656 11.50708 185.6448 0.000001 0.962062 Mean dependent var 0.957319 S.D. dependent var 64.93003 Akaike info criterion 33727.27 Schwarz criterion -54.80677 F-statistic 2.680127 Prob(F-statistic) 按照ls y c x2 回归： Dependent Variable: Y Method: Least Squares Date: 12/11/11 Time: 12:26 Sample: 1 10 Included observations: 10 Variable C X2 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat 0.958687 Mean dependent var 0.953523 S.D. dependent var 67.75602 Akaike info criterion 36727.03 Schwarz criterion -55.23280 F-statistic 2.394519 Prob(F-statistic) weibo.com/kouqintang 对比可见x1的拟合结果较好，因此选用x1作为初始回归模型加上x2再做回归， 按照ls y c x1 x2 回归： Dependent Variable: Y Method: Least Squares Date: 12/11/11 Time: 12:31 Sample: 1 10 Included observations: 10 Variable C X1 X2 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient 245.5158 0.568425 -0.005833 Std. Error 69.52348 0.716098 0.070294 t-Statistic 3.531408 0.793781 -0.082975 Prob. 0.0096 0.4534 0.9362 1110.000 314.2893 11.56037 11.65115 88.84545 0.000011 0.962099 Mean dependent var 0.951270 S.D. dependent var 69.37901 Akaike info criterion 33694.13 Schwarz criterion -54.80185 F-statistic 2.708154 Prob(F-statistic) 很明显，变量均未通过t检验，因此，x1与x2之间存在共线性 相关系数法： 进行LS回归，命令栏中输入“ls y c x1 x2”回车，即得到回归结果： Dependent Variable: Y Method: Least Squares Date: 12/11/11 Time: 12:34 Sample: 1 10 Included observations: 10 Variable C X1 X2 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient 245.5158 0.568425 -0.005833 Std. Error 69.52348 0.716098 0.070294 t-Statistic 3.531408 0.793781 -0.082975 Prob. 0.0096 0.4534 0.9362 1110.000 314.2893 11.56037 11.65115 88.84545 0.000011 0.962099 Mean dependent var 0.951270 S.D. dependent var 69.37901 Akaike info criterion 33694.13 Schwarz criterion -54.80185 F-statistic 2.708154 Prob(F-statistic) weibo.com/kouqintang 在普通最小二乘法下，该模型的可决系数与F值较大，但是两个参数估计值的t检验值较小，说明各解释变量对Y的联合线性作用显著，但各解释变量间存在共线性使得它们对Y的独立作用不能分辨，故t检验不显著。 实验四：联立方程模型的估计与检验 （数据来源于李子奈版课后习题P228.8） 运行Eviews,依次单击file→new→work file→unstructed→observation 18。命令栏中输入“data m2 gdp p cons i”，打开“m2 gdp p cons i”表，接下来将数据输入其中。 建立统计模型： 依次单击object→new object→system 装入模型： gdp=c(1)+c(2)*m2+c(3)*cons+c(4)*i m2=c(5)+c(6)*gdp+c(7)*p inst cons i p 估计模型：估计第一个方程： 依次单击quick→estimate equation→method→勾选TSLS。输入： gdp=c(1)+c(2)*m2+c(3)*cons+c(4)*i cons i p 得到下列结果： Dependent Variable: GDP Method: Two-Stage Least Squares Date: 12/11/11 Time: 12:39 Sample: 1 18 Included observations: 18 GDP=C(1)+C(2)*M2+C(3)*CONS+C(4)*I Instrument list: CONS I P C(1) C(2) C(3) C(4) R-squared Adjusted R-squared S.E. of regression Durbin-Watson stat Coefficient -173.5857 -0.049398 1.669297 0.940707 Std. Error 913.2787 0.024083 0.068490 0.043700 t-Statistic -0.190069 -2.051188 24.37286 21.52645 Prob. 0.8520 0.0594 0.0000 0.0000 102871.0 69213.19 13816310 0.999830 Mean dependent var 0.999794 S.D. dependent var 993.4180 Sum squared resid 1.554243 weibo.com/kouqintang 第一个方程的估计结果如下： GDP=-173.5856662-0.04939825085*M2+1.669297466*CONS+0.9407073699*I 继续估计第二个方程： 依次单击quick→estimate equation→method→勾选TSLS。输入： m2=c(5)+c(6)*gdp+c(7)*p cons i p 得到下列结果： Dependent Variable: M2 Method: Two-Stage Least Squares Date: 12/1111 Time: 12:45 Sample: 1 18 Included observations: 18 M2=C(5)+C(6)*GDP+C(7)*P Instrument list: CONS I P C(5) C(6) C(7) R-squared Adjusted R-squared S.E. of regression Durbin-Watson stat Coefficient 1986.222 1.809611 -146.9316 Std. Error 14474.53 0.063122 62.85453 t-Statistic 0.137222 28.66846 -2.337646 Prob. 0.8927 0.0000 0.0337 144174.3 117812.4 1.82E+09 0.992274 Mean dependent var 0.991244 S.D. dependent var 11023.89 Sum squared resid 0.389408 第二个方程的估计结果如下： M2=1986.221751+1.809610829*GDP-146.9316416*P 综上所述：该联立方程最终结果为： GDP=-173.5856662-0.04939825085*M2+1.669297466*CONS+0.9407073699*I M2=1986.221751+1.809610829*GDP-146.9316416*P weibo.com/kouqintang
相关推荐：
loading