The RATS Software Forum

TomDoan wrote: %MODELLARGESTROOT(mdl) gives the largest root, which is generally all you need. If you really want all the eigenvalues (there are a lot of them!!) you have to get the companion matrix (%MODELCOMPANION function) and use the EIGEN instruction. Note that there will typically be a fair number of complex eigenvalues, so you need to get the complex versions of the eigen values (CVALUES option on EIGEN). See https://estima.com/forum/viewtopic.php?p=726.

Dear Tom,

Thanks for your careful explanation. The value of the largest root would be somehow enough in my case.

According to the outcome of VAR test, I still have a few little puzzles
1、Is it that the F-test as below shows the causality relationship between two variables; for example, RDJI granger cause RFTSE meanwhile RSH failed to granger cause RFTSE ?
2、Does the causality relationship between two variables here take the effect of other variables of this VAR into consideration, which becomes the main difference with Granger–Sims causality ? Since they are all bilateral relationships.
3、Is it possible to obtain too the F-test result of VAR during the VAR-GARCH model since it looks like not showing.

Dependent Variable RFTSE
Mean of Dependent Variable 0.0173550020
Std Error of Dependent Variable 1.1930008799
Standard Error of Estimate 1.1408177200
Sum of Squared Residuals 7725.4966575
Durbin-Watson Statistic 2.0012

Variable Coeff Std Error T-Stat Signif
************************************************************************************
1. RDJI{1} 0.363121540 0.015955964 22.75773 0.00000000
2. RDJI{2} 0.029389282 0.017387165 1.69029 0.09102575
3. RDJI{3} -0.012993935 0.016960487 -0.76613 0.44362951
4. RSH{1} 0.001405407 0.006071609 0.23147 0.81695618
5. RSH{2} 0.000040871 0.006073049 0.00673 0.99463054
6. RSH{3} -0.006817941 0.006065957 -1.12397 0.26107213
7. RN225{1} -0.042158014 0.010655595 -3.95642 0.00007697
8. RN225{2} -0.023375236 0.010741682 -2.17612 0.02958527
9. RN225{3} -0.008070324 0.010137543 -0.79608 0.42601571
10. RFTSE{1} -0.192190637 0.016439594 -11.69072 0.00000000
11. RFTSE{2} -0.038710721 0.016856339 -2.29651 0.02168150
12. RFTSE{3} -0.007968455 0.016096631 -0.49504 0.62059112

F-Tests, Dependent Variable RFTSE
Variable F-Statistic Signif
*******************************************************
RDJI 177.4119 0.0000000
RSH 0.4317 0.7302714
RN225 6.3087 0.0002871
RFTSE 45.5783 0.0000000

Thank you very much!
Bill

yangdashan wrote: Dear Tom,

Thanks for your careful explanation. The value of the largest root would be somehow enough in my case.

According to the outcome of VAR test, I still have a few little puzzles
1、Is it that the F-test as below shows the causality relationship between two variables; for example, RDJI granger cause RFTSE meanwhile RSH failed to granger cause RFTSE ?
2、Does the causality relationship between two variables here take the effect of other variables of this VAR into consideration, which becomes the main difference with Granger–Sims causality ? Since they are all bilateral relationships.

First off, is there a reason you don't have a CONSTANT in the model? That would only be proper if all the series are thought to be mean zero and most of those returns series would be expected to have positive means.

No. If a four variable system, those are only helpful in one direction---if it's significant you have causality. A insignificant result really tells you nothing particularly useful. See Causality with Three or More Variables.

yangdashan wrote: 3、Is it possible to obtain too the F-test result of VAR during the VAR-GARCH model since it looks like not showing.

See the section on Causality in VAR-GARCH models on the same page.