Page 1 of 1
Chow Test OLS - How to correct an Error?
Posted: Mon Jan 12, 2015 9:48 am
by Nuttela
I am an urgent question, when estimating the Chow Test, I keep getting an error "Specification leads o singular matrix in at least one sub-sample".
Could you please help figure out what is the problem...?
I am using an OLS model with dummy variables. I have included n-1 dummy variables so the error should no be caused by dummy variables...
For regressors, I have included intercept (c) and all the independent variables x1 x2 x3 etc...
Thank you!
Best,
Nataly
Re: Chow Test OLS - How to correct an Error?
Posted: Mon Jan 12, 2015 9:49 am
by EViews Gareth
Almost certainly one of your dummy variables does not change during the sub-samples used in the Chow test.
Re: Chow Test OLS - How to correct an Error?
Posted: Mon Jan 12, 2015 9:57 am
by Nuttela
Ok. I'd like to get the Chow test output and test whether the parameters are constant. Even if one of he dummies is constant then why wouldn't it produce the output? Can I fix this somehow?
What is worrying is that in the model without dummies also produces an error saying "Near singular matrix"... Not sure how fix this and get the test outputs..?
Re: Chow Test OLS - How to correct an Error?
Posted: Mon Jan 12, 2015 10:00 am
by Nuttela
I've attached the file with the OLS model... I am trying to perform the Chow test in Model7. Maybe this could help identify the problem...I suspect I am just missing something...
Re: Chow Test OLS - How to correct an Error?
Posted: Mon Jan 12, 2015 10:03 am
by EViews Gareth
Your year dummy is a constant.
Re: Chow Test OLS - How to correct an Error?
Posted: Mon Jan 12, 2015 1:22 pm
by Nuttela
Great, the Chow test is working now since I have removed year series completely. :D :D
However, does this mean I should remove year from all the models as it is the constant (even if it changes from 2011 and 2013)?
Or is there a way to replace c with year? Or is this not necessary?
Thank you for all your help!