1、51999M0417.310789. 041120. 930;壽28531. 252.1561999M0517. 7710559. 741279. 330. 9428467. 632. 2671999M0617.9210970.81689. 430. 9627906. 052. 3181999M0720.110655.151601.4628452. 9891999M0821.2810829. 281627.1228692. 842. 79101999M0923. 7910336. 951570.70. 9528727. 562. 54111999M1022. 6710729. 861504.
2、560. 9328792. 282. 73121999M1124. 7710877. 811434. 970飾28247. 222.36131999M1226. 0911497.121366. 580. 9927381. 772. 36142000M0127. 0110940. 53153528522. 222. 42152000M0229.310128. 311714. 581. 0229165.22. 66162000M0329. 8910921.921800. 231. 0429098. 4717?nnriMndK Rd1H733 qi1如391 HA只gRlWI 37q nd(2)将电
3、子表格数据导入eviewsFile-open-foreign data as workfile,得到数据的Eviews工作文件和数据序列表。Views EViersFile Edi t Object View Proc Qmck Options W1 ndow HelpO lorkf ile:数据口 Group: UNTITLEDZES IViewProc|Object| Print|ScView | Proc | Object Print | Name | Freeze j | Default i Sort | Transpose Edit+/-1 Smpl+/-1 TitlRange: 1
4、999M01 2008bSample:ObSCOILFUTUREDOWUEUROP12.499358.831134.67kc0 coilfuture0 dow0 nagas0 opec0 resid0 shindexlabel time0 ueurope0urmb12.029306.581090.08词14.681158.0517.3010789.041120.9317.7710559.741279.3310970.801689.4320.1010829.2823.7910336.9515707022.6710729.861504.5624.7710877.811434.9726.091366
5、.5827.0110940.531535.0029.3010128.311714.5829.891800.232000M0425.5410733.911836.322000M0528.8110522.331894.55C Untitled / NewPac2000M0631.5310447.891928.112000M07297210521.982023.54|v|2000M082.计算变量间的相关系数在 窗【I 中 4谕入命令:cor coilfuture dow shindex nagas opec ueurope urmb, 点 击回车键,得到各序列之间的相关系数。结果表明Coilfut
6、ure数列与其他数列存 在较好的相关关系。 Croup: UBTITLED forkfile: Jfc#: :Untitledyiew|Prodob血t| PrntNane|Fieeze| 刃mpte| Sheet | Sials| Spec |CorrelationoevyURMB1.0000000.6454250.5707940.7766690.905918-0774546-0.913769 aDCW0.6464261 oooooo07642540.3963450 596469-0 526129-06886246.5787840.7542541 .oooooo0.2727380.34426
7、6-0.39170607377220778117-0.62346-0.5950230.8059180.5994690.778117-0718323-0.606976 .-0774548-0.5261390.642346-0 7183230.671706-0 912763-0.688624-0727722-0 596023.0 60EQ760 6717061 oooooo vuin.3 时间序列的平稳性检验(1)观察coilfuture序列趋势图在eviews中得到时间序列趋势图,在quick菜单中单击graph,在series list对 话框中输入序列名称coilfuture,其他选择默认操
8、作。图形表明序列随时间变化 存在上升趋势。(2)对原序列进行ADF平稳性检验quick-series statistics-unit root test,在弹出的 series name 对话框中输入需要检 验的序列的名称,在test for unit root in选择框中选择level,得到原数据序列的 ADF检验结果,其他保持默认设置。O Series: COILFUTURE Workfile:数据,:Unti-tled |View IProc | Object | Properties | Print |Name | Freeze Sample| Genr | Sheet | Grap
9、h | Stats| Ident|Augmented Die key-Fuller Unit Root Test on COIL FUTURENull Hypothesis: COILFUTURE has a unit rootExogenous: ConstantLag Length: 0 (Automatic based on SIC, MAXLAG=12)t-StatisticProb*Auamented Dickey-Fuller test statistic0.9700920.9961Test critical values:1 % level5% level10% level-3.
10、488063-2.386732-2.580281MacKinnon (1996) one-sided p-values.Augmented Dickey-FullerTest Equation Dependent Variable: D(COILFUTURE) Method: Least SquaresDate: 12/26/11 Time: 21:37Sample (adjusted): 1999M02 2008M08Included observations: 115 after adjustments得到序列的ADF平稳性检验结果,检测值0.97大于所有临界值,则表明序 列不平稳。以此方
11、法,对各时间序列依次进行ADF检验,将检验值与临界值比 较,发现所有序列的检验值均大于临界值,表明各原序列都是非平稳的。(3)时间序列数据的一阶差分的ADF检验quick-series statistics-unit root test,在 series name 对话框中输入需要检验的序列的 名称,在test for unit root in选择框中选择Ind difference,对其一阶差分进行平 稳性检验,其他保持默认设置。I Tin it Root TestTest type| Augmented Dickey-FullerOKCancelAugmented Dickey-Fulle
12、r Unit Root Test on D(COILFUTURE) D(COILFUTURE) has a unit root-7.3034410.0000-3.488585-2.886959-2.580402Augmented Dickey-FullerTest EquationDependent Variable: D(COILFUTURE,2) Method:46 1999M03 2008M08 114 after adjustments得到序列的ADF平稳性检验结果,检测值7.8远小于所有临界值,则表明 序列一阶差分平稳。以此方法,对各时间序列的一阶差分依次进行ADF检验, 将检验值与
13、临界值比较,发现所有序列的检验值均小于临界值,表明各序列一阶 差分都是平稳的。由此可知,以上时间序列均为一阶单整时间序列。4.Granger因果检验(1)quick-group statistics-granger causality test,出现如下对话框,输入各序列名 称,点击oic以此得到输入序列之间的单项或双向因果关系。(2)滞后阶数釆用Eviews推荐的滞后阶数Lag Specification(3)得到与coilfuture序列相关的Granger因果检验结果。存在dow到coilfuture 的单向因果关系;存在shindex到原油期货价格的单向因果关系;存在原油期货 价格到n
14、agas的双向因果关系;存在原油期货价格到OPEC产量的单向因果关系; 存在ueurope到原油期货价格的单向因果关系;存在ermb到原油期货价格的单 向因果关系。Pairwise Granger Causality Tests Date:55 1999M01 2008M08 Lags: 2Null Hypothesis;ObsF-StatisticProb.DOW does not Granger Cause COILFUTURE COILFUTURE does not Granger Cause DOW1144.518320.605B30.01300.5476SHINDEX does no
15、t Granger Cause COILFUTURE COILFUTURE does not Granger Cause SHINDEX3.508131.102090.03340.3358NAGAS does not Granger cause COILFUTURE COILFUTURE does not Granger Cause NAGAS1.606462.451890.20530.0909OPEC does not Granger Cause COILFUTURE COILFUTURE does not Granger Cause OPEC0.109093.707610.89670.02
16、77UEUROPE does not Granger Cause COILFUTURE COILFUTURE does not Granger Cause UEUROPE2.551650.231810.08260.7935URMB does not Granger Cause COILFUTURE COILFUTURE does not Granger Cause URMB2.544540.992810.08320.37395.协整检验对上述的7个单整时间序列采用EG (Engle-Granger)两步法进行协整检验。(1) 在工作表窗口选取 coilfuture、dow、shindexnag
17、as、opec ueurope urmb,然后单击右键,选择open,点击as group,得到所有序列数据。 Group: OTTITLED lorkfilef -WifledX(2)在新窗口 中点击 proc,选择 make equation,使用 Engle-Granger (EG)两步 检验法进行回归,得到回归结果:O Equation: UHTITLED Workfiles gSWistijrtit.View!Proc|Object I PrintjName| Freese, Estimate Forecast |StatsResids| COILFUTURE Method: 22:
18、Sample; 1999M01 2008M08 11 6VariableCoefficientStd. Error-0.0008160.000786-1.0388770.3012MAGAS1.4903800.3755823.9681870.00010.0035680.0005586.391991-0.0014660.001 0221.4352840.1541-1 2.246635.870970-2.0859630.0393-49.591243.055834-16.22838C351.383530.2871711.60173R squared0.952580Mean dependenivar46
19、.0B198Adjusted R-squared0.949970S.D. dependentvar26.64445S.E. of regression5.959677Akaike info criterion6.466357Sum squared resid3871.434Schwarz criterion6.632522I nn likPlihnnrirtd!Hnnqn-Quinn rritrfi(3 )在新窗口中点击proc,选择make residual series,得到残差项时间序列RESIDOlo RESID01 Workfile:UntilledXView I Proc | Ob
20、ject |Properti |Print |Name |Freeze (Default Sort | Edit+/-1 Srnpk/-1 LatRESID01Last updated: 12/26/11 -22:40T1Modified: 1999M01 200SM08/makeresid口-16.34587-18.27189-14.07738-6.627166-5.926538-2.665586-2.823574-3.279900-0.883294-2.5386202.5660447.6239124.1783003.784423nnnhrtnd圈|(4)对该序列RESID01进行ADF检验
21、(如上所述)。若残差项平稳,则存在协整关系。否则,不存在。山结果可知,检验值-5.298明显小于所有临界值,则残差项RESID01平稳,即原油期货价格与选定的相关连续经济变量存在着长期均 衡关系。 RESID0LMe训Object Print |Name |Freeze | (Default T Sort| Edit4卜| Smpl+卜|LatAugmerrted Dickey-Fuller Unit Root Test on RESID01 RESID01 has a unit root 1 (Automatic based on SIC, MAXLAG=12)Auamented Dickey-Fullertest statistic-5.297907*Mackinnon (1996) one-sided p-values. D(RESID01)Method:436.误差修正模型(1)对所有
copyright@ 2008-2022 冰豆网网站版权所有
经营许可证编号:鄂ICP备2022015515号-1