Retrieve stats from Ramsey RESET or Chow forecast tests

[andrew.lau]

Hello there,

I'm using Eviews8 and have been trying to write a program that creates a table with the results of different diagnostic tests (Ramsey's RESET test, Chow breakpoint and forecast tests, ARCH LM test, White's heteroscedasticity test and Breusch Godfrey autocorrelation test) for several equations. Usually I do this by freezing the result of the tests and then retrieving the results from the frozen table e.g.

freeze(mode=overwrite,frozen_table) equation1.auto(2)
table1(3,2) = frozen_table(3,2)

However, I noticed that with the Ramsey's RESET test and Chow forecast test, the position of the F-statistic and likelihood ratios may change based on the length of a specification for an equation. The attached tables show that the F-statistic is located in B8 in one table and B9 in the other table.

I was wondering if there was a way to standardize the tables so that the F-statistic and likelihood would be located in the same position across all the tables for the Ramsey's RESET test and Chow forecast test? Or if there was another way to retrieve the F-statistic and likelihood ratio (value, df, probability)?

[EViews Gareth]

Unfortunately not. You'll have to calculate where the statistic will be.

[cntantamis]

Hi,
I have also encounter the same problem for the RESET (with 2 fitted terms). What I noticed is that, depending on the number of independent variables your equation has, the location of the p-value for the RESET test is in rows 7-8-9. Thus, I could use the following set of if checks

freeze(zztest2) {zzt1}.reset(2)
'Depending on the number of independent variables in each equation, the location of the RESET pvalue can be
' in rows 7, 8, or 9 of the 4th column. Use the fact that the first column on the respective row should contain "F-statistic"
if @left(zztest2(7,1),1)="F" then
zzBGRBPGAQApvalue(!i+1,3)=@val(zztest2(7,4))
endif
if @left(zztest2(8,1),1)="F" then
zzBGRBPGAQApvalue(!i+1,3)=@val(zztest2(8,4))
endif
if @left(zztest2(9,1),1)="F" then
zzBGRBPGAQApvalue(!i+1,3)=@val(zztest2(9,4))
endif

To explain: {zzt1} is the name of my equation, !i corresponds to the number of the equation, and zzBGRBPGAQApvalue is the matrix name where I stored the p-value.

I understand that it is not the best approach, but it seems to serve its purpose.

[andrew.lau]

Ahhh thanks for the responses, I will try this out.

Cheers!

[chisylvia]

Hi all, please I'm using Eviews9 how do I write a program that creates a table with the results of different diagnostic tests (Ramsey's RESET test, Chow breakpoint and forecast tests, ARCH LM test, White's heteroscedasticity test and Breusch Godfrey autocorrelation test).

[EViews Gareth]

http://forums.eviews.com/viewtopic.php?f=5&t=1638

http://forums.eviews.com/viewtopic.php?f=5&t=18