EViews.com

I am struggling with the BaiPerron breakpoint test because any equation I estimate using LS, even the simplest one regressing e.g. output on prices from 1984Q1 to 2010Q4 gives as an error message: "R(for 'library(strucchange)') returned an error: evaluation stopped because of an error." (min subsample size 0.15, I tried however also others).

I have installed R 2.15, the strucchange package, statconn and then Eview 7.2 and the BaiPerron addin. I have even reinstalled and installed the package to see whether it is correctly installed.

I will appreciate a response! Thank you very much!

You also need StatconnDCOM

Download and install the current R&Friends on this site:

http://rcom.univie.ac.at/download.html

@JimForest: Great! Thank you so much!!! It works now!

This forum is really helpful and I want to thank the admins for their support and the idea with the add ins! I was wondering, why not create an addin conducting a Chow break point test of a VAR system of equations as .e.g in JMulti. Thank you!

Should be relatively straight forward to do the LR form of the test, just estimate over the full sample, then over the sub-samples and figure out what the LR statistic is.

@JimForest: Great! Thank you so much!!! It works now!

This forum is really helpful and I want to thank the admins for their support and the idea with the add ins! I was wondering, why not create an addin conducting a Chow break point test of a VAR system of equations as .e.g in JMulti. Thank you!

No problem @stanM. I find this test to be very useful. In fact, I didn't care too much for R. But using the Statcom to interact and bring data back into EViews is a great feature. I've used the com features with Matlab as well and it interacts faster between those two environments (EViews and Matlab).

Hey guys,

" Loading StatConnector Server... Done
Initializing R...Function call failed
Code: -2147221485
Text: installation problem: unable to load connector
Releasing StatConnector Server...Done

Test1
Method '~' of object '~' failed "

How I slove this problem? and I have reinstalled R.

Thanks!

Looks like it is an issue with statconnector. You'll have to contact them.

I had that problem the first time I installed stat comm.

I think the best way to address the problem is to install "R and Friends" and the installer should follow the install sequence that enables the process to be activated properly.

Give this a try: http://rcom.univie.ac.at/download.html#ROOo

After using this connector between EViews and R, you should be able to call R directly from EViews programs. This is well documented in the EViews documentation.

I find EViews to be my preferred means of accessing R (and Matlab).

With those three programs, there is little you are unable to do.

Hi Gareth and Jimforest,
Thanks for your help! It works well now.

I have another simple question. If I want to test structure break for a univariate(Y) of time series, what equation should I setting?
Is that "ls Y C "OK? or anything else?

Many thanks!

If you want to test for a structural break in the mean, I would suggest you just regress the dependent variable on a constant.

It is exactly what I am thinking, I will try, thanks!

Hi all,

Is the BaiPerron only detect the breakpoints? I mean it does not provide unit root test right?

If I need to do unit root test on time series that have one or more than one breaks, what should i do?

Thx!

If you want to test for a structural break in the mean, I would suggest you just regress the dependent variable on a constant.

hi If i want to test for strucutual breaks in trend and constant, can i regress both?

y c @trend ?

Yeah, well you are testing for a structural break in the residuals of a regression of that form.

Yeah, well you are testing for a structural break in the residuals of a regression of that form.

Thank you !
if price series is stock price, do I need to log the series before using this addin?