请问大家, 在EVIEWS里进行因果关系(GRANGER CAUSALITY)TEST的时候, 我只能RUN出时间序列数据的, 但如果是 面板数据 POOL DATA/ PANEL DATA如何用EVIEWS得到F值和概率?请看下面的简单举例.

我导师这个月出去开会了, 我发EMAIL也是自动回复, 所以直接转发下EMAIL, 大家帮忙看看, (数据和变量是为了说明问题临时乱写的) 主要就是如果是POOL DATA/ PANEL DATA如何用EVIEWS直接得出一个SINGLE表格来看F值和概率, 从而判断因果关系? 还是必须每个CROSS-SECTION DATA都RUN一次GRANGER CAUSALITY, 然后把每个股票对应的F值和P值取平均, 我有258只股票, 那不是太麻烦了. 我只会用EVIEWS RUN出单只股票的GRANGER CAUSALITY, 但我以前好象看过别人文章里把N只一起RUN出, 就用一个表格来显示F和P. 我尝试把所有数据用PANEL DATA/ POOL DATA的形式输入EVIEWS, 可以RUN出POOLED ESTIMATION,就是像模型里面的系数和常见指标都可以, 但RUN不出一个表格的GRANGER CAUSALITY.

For example, there is a stock; R denotes Return of the Stock, V denotes Trading Volume of the Stock..

Year	R	V
1995	103	2333883
1996	208	9992223
1997	306	3866688
1998	300	3000000
1999	200	34888888
2000	100	2987377
2001	988	998888348
2002	298	98888866
2003	302	2209993
2004	208	33388868

Rt=c1+ a2*Rt-1 +a3*Rt-2 +b1*Vt-1 + b2*Vt-2 + et

Vt=c2+a3* Rt-1 +a4* Rt-2 +b3*Vt-1 + b4**Vt-2 + et

We can work with a VAR model to get estimation and run the Granger Causality with Eviews to get F-Statistic and Probability (if 2 Lags).

Pairwise Granger Causality Tests
Date: 08/03/05 Time: 14:48
Sample: 1995 2004
Lags: 2
Null Hypothesis:	Obs	F-Statistic	Probability
V does not Granger Cause R	8	1.42507	0.36723
R does not Granger Cause V		1.75732	0.31250

BUT, if there are three stocks, S1, S2, S3; Ri,t denotes Return of Stock i in year t, Vi,t denotes Trading Volume of Stock i in year t.

Stock	Year	R	V
S1	1995	103	2333883
S1	1996	208	9992223
S1	1997	306	3866688
S1	1998	300	3000000
S1	1999	200	34888888
S1	2000	100	2987377
S1	2001	988	998888348
S1	2002	298	98888866
S1	2003	302	2209993
S1	2004	208	33388868
S2	1995	466	3888889
S2	1996	888	6888888
S2	1997	899	77888356
S2	1998	999	66666488
S2	1999	333	78888888
S2	2000	386	99988898
S2	2001	668	39999998
S2	2002	389	3000088
S2	2003	788	800997
S2	2004	200	6999998
S3	1995	100	3999877
S3	1996	300	30009988
S3	1997	500	5888876
S3	1998	400	4555858
S3	1999	800	33333888
S3	2000	300	28888888
S3	2001	200	18899999
S3	2002	383	2633388
S3	2003	236	2986878
S3	2004	639	38869866

Ri,t=c1+ a2*Ri,t-1 +a3*Ri,t-2 +b1*Vi,t-1 + b2*Vi,t-2 + et

Vi,t=c2+a3* Ri,t-1 +a4* Ri,t-2 +b3*Vi,t-1 + b4**Vi,t-2 + et

After importing all the above panel data, I select 'Granger Causality test' as usual and enter ‘R? T? ‘in the series list but it then shows ‘R? is not defined’.

·How can we run the Granger Causality with Eviews to get F-statistic and Prob. when dealing with panel data/ pool data??

·Can we get just a single table of F-stat. and Prob. rather than three tables for each stock ? Or we cannot get a single table but need to calculate the average of three different F-stat and Prob for each stock ? Because I hope to get a table as follows to see the Granger Causality btw R and V

Null Hypothesis:	Obs	F-Statistic	Probability
V does not Granger Cause R	..	..	..
R does not Granger Cause V		..	..

Looking forward to your answer. Many thanks!

Best Regards

有熟悉EVIEWS的同学会吗? 知道怎么用一次RUN出来吗?

谢谢大家的帮助!