实验三多元时间序列分析方法
1.实验目的
了解协整理论及协整检验方法;掌握协整的两种检验方法:E-G两步法与Johansen方法;熟悉向量自回归模型VAR的应用;掌握误差修正模型ECM的含义及检验方法;掌握Granger因果关系检验方法。
2.实验仪器
装有EViews7.0软件的微机一台。
3.实验内容
【例6-2】
时间与M2之间的关系首先用单位根检验是否为平稳序列。原假设为H0:非平稳序列H1:平稳序列。用Eviews软件解决该问题,得到如下结果:Null Hypothesis: M2 has a unit root
Exogenous: None
Lag Length: 3 (Automatic - based on SIC, maxlag=13)
t-Statistic Prob.* Augmented Dickey-Fuller test statistic 5.681169 1.0000
Test critical
values: 1% level -2.579052
5% level -1.942768
10% level -1.615423
*MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(M2)
Method: Least Squares
Date: 04/16/13 Time: 10:36
Sample (adjusted): 1991M05 2005M01 Included observations: 165 after adjustments
Variable Coefficien
t Std. Error t-Statistic Prob.
M2(-1) 0.013514 0.002379 5.681169 0.0000 D(M2(-1)) -0.490280 0.074458 -6.584611 0.0000 D(M2(-2)) 0.070618 0.083790 0.842797 0.4006 D(M2(-3)) 0.387086 0.073788 5.245935 0.0000
R-squared 0.480147 Mean dependent
var 1440.03
7
Adjusted
R-squared 0.470461 S.D. dependent var 1509.48
9
S.E. of regression 1098.447 Akaike info criterion 16.8651
3
Sum squared resid 1.94E+08 Schwarz criterion 16.9404
2
Log likelihood -1387.373 Hannan-Quinn
criter. 16.8956
9
Durbin-Watson
stat 1.965242
从上图我们可以看出t-statistic的值是5.681169,大于临界值,p>a,故不能拒绝被检验的指数序列是非平稳的原假设。因此一阶差分序列进行ADF检验,结果如下图显示。
Null Hypothesis: D(M2) has a unit root
Exogenous: None
Lag Length: 8 (Automatic - based on SIC, maxlag=13)
t-Statistic Prob.* Augmented Dickey-Fuller test statistic 0.988183 0.9143
Test critical
values: 1% level -2.579587
5% level -1.942843
10% level -1.615376
*MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(M2,2)
Method: Least Squares
Date: 04/16/13 Time: 10:37
Sample (adjusted): 1991M11 2005M01 Included observations: 159 after adjustments
Variable Coefficien
t Std. Error t-Statistic Prob.
D(M2(-1)) 0.053616 0.054257 0.988183 0.3247 D(M2(-1),2) -1.526069 0.096352 -15.83852 0.0000 D(M2(-2),2) -1.519649 0.149134 -10.18981 0.0000 D(M2(-3),2) -1.225623 0.184003 -6.660869 0.0000 D(M2(-4),2) -1.237445 0.196285 -6.304319 0.0000 D(M2(-5),2) -0.972024 0.197161 -4.930093 0.0000 D(M2(-6),2) -0.810098 0.185290 -4.372060 0.0000 D(M2(-7),2) -0.605069 0.144997 -4.172983 0.0001 D(M2(-8),2) -0.333781 0.080550 -4.143781 0.0001
R-squared 0.801713 Mean dependent
var 16.0700
1
Adjusted 0.791137 S.D. dependent var 2352.91
R-squared 9
S.E. of regression 1075.320 Akaike info criterion 16.8535
6
Sum squared resid 1.73E+08 Schwarz criterion 17.0272
7
Log likelihood -1330.858 Hannan-Quinn
criter. 16.9241
Durbin-Watson
stat 1.970407
从上图我们可以看出t-statistic的值是0.988183,大于临界值,p>a,故不能拒绝被检验的指数序列是非平稳的原假设。因此二阶差分序列进行ADF检验,结果如下图显示
Null Hypothesis: D(M2,2) has a unit root
Exogenous: None
Lag Length: 7 (Automatic - based on SIC, maxlag=13)
t-Statistic Prob.* Augmented Dickey-Fuller test statistic -9.223132 0.0000
Test critical
values: 1% level -2.579587
5% level -1.942843
10% level -1.615376
*MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(M2,3)
Method: Least Squares
Date: 04/16/13 Time: 10:38
Sample (adjusted): 1991M11 2005M01 Included observations: 159 after adjustments
Variable Coefficien
t Std. Error t-Statistic Prob.
D(M2(-1),2) -8.900755 0.965047 -9.223132 0.0000 D(M2(-1),3) 6.431129 0.924672 6.955038 0.0000 D(M2(-2),3) 4.970286 0.833541 5.962861 0.0000 D(M2(-3),3) 3.802432 0.700773 5.426055 0.0000 D(M2(-4),3) 2.617058 0.544596 4.805501 0.0000 D(M2(-5),3) 1.688201 0.380559 4.436109 0.0000 D(M2(-6),3) 0.910968 0.214990 4.237257 0.0000 D(M2(-7),3) 0.325934 0.080151 4.066487 0.0001 R-squared 0.941321 Mean dependent 0.11205
var 7
Adjusted
R-squared 0.938601 S.D. dependent var 4339.32
4
S.E. of regression 1075.236 Akaike info criterion 16.8474
7
Sum squared resid 1.75E+08 Schwarz criterion 17.0018
8
Log likelihood -1331.374 Hannan-Quinn
criter. 16.9101
8
Durbin-Watson
stat 1.963915