EViews.com

Sorry, I meant:

'This calculates the sum of lagged t statistics for a number of variables
freeze(tablet!X) eq!X

'Add something like, if (condition for appearance of Wald) !tsa= 1 + 19 + 2 + (!D*7) + ((3 + !OL+1 + 2)*5) + (3 + !TSAL + 1 + 2)
else, !tsa=19 + 2 + (!D*7) + ((3 + !OL+1 + 2)*5) + (3 + !TSAL + 1 + 2)
'!tfp= 19 + 2 + (!D*7) + ((3 + !OL+1 + 2)*5) + (3 + !TSAL + 1 + 2) + (3 + !TFPL + 1 + 2)
'tstattsasplps1(!X)=tablet!X(!tsa,6)
'tstattfp4ps1(!X)=tablet!X(!tfp,6)

I'm attempting to store the sum of lagged t statistics for a number of PDL regression models. To do this, I am freezing each PDL regerssion output into a table and attempting to select the sum of lagged t statistics for a specific variable. Where this sum is in the table is dependent on the # of lags and degress of the polynomials, but also depends on the rest of the output of the pdl regression. For some reason, in many of my models a Wald F-Statistic and Probability(Wald F-Statistic) appear in the area with other summary statistics, while in other models these two measure do not appear. This affects the location of the sum that I am interested in, and has made it very difficult to select the correct one.

Why do these measures appear for some models and not others? I am testing many different lag lengths, with constraint=1 and a 3rd degree polynomial.

If I knew why these measure appeared, a simple confitional like the one I commented in above would probably solve my problem.

Depends on whether your model has robust standard errors or not.

eq!X.ls(cov=hac) ps1 c time pdl(ipr,!OL,!D,!C) pdl(un,!OL,!D,!C) pdl(unionprq,!OL,!D,!C) pdl(unipr,!OL,!D,!C) pdl(unionipr,!OL,!D,!C) pdl(tsaspl,!TSAL,!D,!C) pdl(tfp4,!TFPL,!D,!C)

For each regression I am running, I'm using the Newey West option, so shouldn't every model have robust standard errors calculated, and therefore have the Wald Statistic measures printed?

Also depends on whether you have a constant or not

I also have a constant in every equation (eq!X.ls(cov=hac) ps1 c time ...), is there another criterion?

Sorry about this...

Post the workfile with examples of equations with and without.

Look at eq2 and eq3.

eq2 is
!X=2, !OL=3 !TFPL=3 !TSAL=4
eq3 is
!X=3 !OL=3 !TFPL=3 !TSAL=5
Both eqs have !D=3, !C=1.
eq!X.ls(cov=hac) ps1 c time pdl(ipr,!OL,!D,!C) pdl(un,!OL,!D,!C) pdl(unionprq,!OL,!D,!C) pdl(unipr,!OL,!D,!C) pdl(unionipr,!OL,!D,!C) pdl(tsaspl,!TSAL,!D,!C) pdl(tfp4,!TFPL,!D,!C)

Eq2 has the Wald statistic measures while Eq3 does not.
I attatched my code and a workfile with the regression results. Scroll to the right for the Eqs.
Thanks for your help

Looks like a bug, we'll fix it for the next patch.

In the mean time, I guess you could check whether the appropriate cell of the table contains the text "Wald F-statistic".

Alright, that's unfortunate. Thanks for the quick responses. I've got it all working with a conditional based on the string contents of the general statistic cells.