BaiPerron (Bai-Perron breakpoint test - Requires R)

[jaykody]

Trying to run Bai-Perron on time series regression output and getting an error. I can use other plugins for R but for some reason this one is not working.

I keep getting this error:
"R (for 'time') returned an unknown error"

Sound familiar to anyone? Thoughts?

[EViews Gareth]

Can you post your workfile?

[jaykody]

After trading messages with Gareth, I had a variable named 'time' in my file. This apparently caused some conflicts, once it is changed it runs through. His reply below in the event anyone else has this problem. Thanks for the help...

R doesn't let you have a variable called TIME, which you do. Change the name of your TIME variable to something else and it will run.

[wrsuclaedu]

I am curious if anyone has encountered the following. While running B-P, on some series I get results with the default sample size (0.15). On other series I get an error message that says "Syntax Error", and no result. If I then go back and reduce the sub-sample size from the default to something shorter, say 0.10, I then usually get a result (and it is always more than one break). Does "Syntax Error" effectively mean zero breaks at the chosen minimum sub-sample length? Or is there something else going on? Many thanks in advance for any insights or suggestions.

[EViews Gareth]

Could you post your workfile, and the specification you're doing?

[wrsuclaedu]

Could you post your workfile, and the specification you're doing?

Thank you for your reply.

Please find attached the data for two variables/series.

The first case, using the series ordinaryrevreal. I OLS regress it on Year. Then I go to the addin menu and select the B-P addin. With the sub-sample length set at its default of 0.15, it returns the Syntax error. If I go back and set it to 0.10, it returns results with two breaks.

The second case, using the series primexpendreal. I OLS regress it on year, and then under the B-P addin I leave the subsample length at the default of 0.15. It returns results with no Syntax error.

These are just two series. I've tried to run the break on about 10 series thus far, and I'm batting .500 on getting results vs. the Syntax error message at the 0.15 subsample length.

[EViews Gareth]

Thanks. Small bug that would give that error whenever only one breakpoint was chosen. Re-download the add-in, and it should now work.

[wrsuclaedu]

Thanks. Small bug that would give that error whenever only one breakpoint was chosen. Re-download the add-in, and it should now work.

Works like a charm now. Thank you very much for your help.

[ramzan]

how i can install add inns with evews 7

[EViews Gareth]

You need EViews 7.1

[hmvera]

Before running bp test, you will need to install statconnDCOM from http://rcom.univie.ac.at/download.html.

Once installation is complete, run the statconn "Server 01 - Basic Test" program, usually found under Start->All Programs->statconn->DCOM->Server01 - Basic Test, to verify that everything is working. Click on the Start R button to begin the test.

Once you have run the test, you should be able to run EViews and use the xopen command to open a connection to R.
xopen(type=r)
This command will open a connection to R from within EViews and will also open an R output log window in EViews.

[inaiya]

Hi

I am new to econometrics, I need some assistance, I want use the BaiPerron test to determine structural break in a sample of 48 observations. I followed all the suggestions on how to use the add-in including installation of R. However, when I run the test I got this error message:

"R returns an error. The most likely cause of this is that your equation did not have enough observations to perform the test, or that one of the tested breakpoints caused a singularity".

What must I do next. Please assist. Thanks

Ismaeel

[EViews Gareth]

post your workfile.

[inaiya]

the file is attached.

[EViews Gareth]

You cannot use breakpoint tests when you have a dummy variable such as the one you have - it will cause singularities (as the error message indicates).