ZAURoot (Zivot-Andrews Unit Root test)

[trubador]

Hi

I am trying to download ZAURoot adds in to my computer.l i have Eviews 6. It wants me to write something in the Format String part and I could not skip that part.

Could you please expalin me what to do?

Unfortunately, all add-ins require EViews 7.1 or above. The following link will be more helpful: Subroutine for Zivot-Andrews Unit Root Testing

[startz]

Using add-ins requires version 7.1.

[wrsuclaedu]

After running ZAURoot with the default of 4 lags, I increase the number of lags to 6. This gives me a "Near Singular Matrix" error message. If I go back and set the number of lags back to 4, instead of getting my original result, I get the "Near Singular Matrix" error message. Thereafter, no matter what number of lags I set, I still get the error message. Is there something I can do to avoid this problem? I attach a file with the exchange rate series I am using below. Many thanks for any suggestions.

[wrsuclaedu]

If any one knows how the EViews Zivot-Andrews Add-In selects lag length (what criterion), could you please post here? I have looked at the Docs that accompany the Add-In and could not find information on lag length selection criteria, other than it selects lags constrained by the user-specified maximum lags. Many thanks for any info on this.

[trubador]

It uses Akaike Information Criterion (AIC) to determine the optimum lag length. Please note that the maximum lag length is specified by the user.

[wrsuclaedu]

Thank you trubador.

A quick follow up. When I run Zivot-Andrews in STATA, and specify AIC to select the lag length, I am running into a pretty different result than I get with EViews, even if I impose the lag length (that is, one rejects unit root, the other doesn't; and the break points are far from each other). Are there more differences between the implementations than just how lag lengths are selected?

[trubador]

I am not familiar with STATA and have no idea how it handles the Zivot-Andrews unit root test. However, the source code of EViews's ZAURoot procedure is easily accessible and its results are very much in line with those of the original paper.

[raquelopez]

Hello,
My name is Raquel. I work with Eviews 7. I have never used Eviews add-ins, however I am interested in applying the Zivot-Andrews test and I have found the add-in ZAUroot to apply this test. I have saved the ZAUroot file. I find the following files inside: zauroot install, zawrap and zivotandrews, but I don´t know how to get it installed and then how to run it. Please, could yo help me?
Thank you in advance,
Raquel

[EViews Gareth]

You need EViews 7.1

[torossahin]

Dear all,
I am facing the same problem that was faced by previous users. I am using eviews 7.1 and have installed the add-in for zivot-andrews unit root test. But when I try to run it on my series, it turns out as "insufficient number of observations". Can you please kindly post a reply on the solution? I couldn't figure it out from your previous correspondences.
Thanks in advance,
Regards,
Toros

[torossahin]

Hi again,
The same error popped up as "insufficient number of observations" when I try through the subroot coding.
Shall I change the code? Can you please kindly help me?
Best regards,
Toros

[trubador]

You should post your workfile...

[torossahin]

Dear trubador,
Attahced you will find the workfile. The 2 series that I try to apply the ZAURoot test are LNGSYIH and LNIH. (Both of them are 31 observations).
Why do you think the add-in does not work?
Thanks and best regards,
Toros

[trubador]

1) If you significantly increase the lag-length, you may quickly run out of degrees-of-freedom.
2) Make sure that you are using the most recent update of EViews.

[lingqqq]

I am using Eviews7. When I download ZAURoot and installed by Eveiws, it asked me to enter the format string which can not skip.
What should I enter?
Or am I doing the wrong way of installing?