EViews.com

Hi all,

Anybody know how to "Wald" test restrictions on long run parameters within the ARDL framework? Easy and straight forward in MFIT5 but I can't seem to figure that out in EV9.5. Thanks.

There's nothing built in to do it, you'll have to estimate the long run equation manually.

Can you briefly elaborate on what you mean by manually (is that as in re-estimating the Unrestricted ECM "Test Equation" and then test the restrictions?
EV does provide the LR ARDL estimates, I am trying to "efficiently" test few restrictions on these Parameters within the ARDL framework. Your help is appreciated. Thanks

Yes, re-estimate the ECM test equation.

Thanks. I was hoping for a more fast/efficient way of doing this as I at least have 53 ARDL models to estimate and test LR restrictions.
I hope someone will chime in with an idea/ad in to do this> Thanks.

Unsure of how to go about this...in re-estimating the ECM, are we to create new variables series to do that?
I have a similar issue and I am trying to conduct Long Run Wald test to test for asymmetry on some partial sum first differenced variables similar to the Shin et al. (2014) technique. I have already estimated the equation and determined LR and SR coefficients. The resulting equation with SR and LR coefficients yields multiple lags of the dependent variable...and the LR Wald test is meant to test (positive partial sum/dependent variable) = (negative partial sum/dependent variable) ..requiring the coefficient of just one lagged term of the dependent variable. My ECM equation has multiple lagged dependent variables and I am not sure if that's because I am missing something. Has anyone run into similar issues and any thoughts on how to address it? My equations are few and I don't mind a tortuous route to solving it in Eviews.

wpascals wrote:Unsure of how to go about this...in re-estimating the ECM, are we to create new variables series to do that?
I have a similar issue and I am trying to conduct Long Run Wald test to test for asymmetry on some partial sum first differenced variables similar to the Shin et al. (2014) technique. I have already estimated the equation and determined LR and SR coefficients. The resulting equation with SR and LR coefficients yields multiple lags of the dependent variable...and the LR Wald test is meant to test (positive partial sum/dependent variable) = (negative partial sum/dependent variable) ..requiring the coefficient of just one lagged term of the dependent variable. My ECM equation has multiple lagged dependent variables and I am not sure if that's because I am missing something. Has anyone run into similar issues and any thoughts on how to address it? My equations are few and I don't mind a tortuous route to solving it in Eviews.

What you're looking for with partial positive and negative sums is the asymmetric ARDL test. This isn't yet implemented but will be in the future.

You mean Eviews does not have the capability to conduct a long run Wald test using these variables or any other variables at this time?

wpascals wrote:You mean Eviews does not have the capability to conduct a long run Wald test using these variables or any other variables at this time?

No! EViews has the capacity to do almost anything you want, however, it may not be automated at the moment and you may have to write the program yourself. What I meant to say is that your partial positive sums and negative partial sums that you are trying to compute is not part of the classical Pesaran, Shin, and Smith ARDL framework. It is part of the asymmetric ARDL framework which is the work of Shin, Yu, and Greenwood-Nimmo. These are two different theories! In this regard, the asymmetric framework is not currently automated in EViews and if you would like to do computations based on it, you will have to write your own program.