Restrictions on Beta on only one vector

[spollard777]

I have a question about setting up restrictions on beta to do hypothesis testing. If I try to impose restrictions on one vector and none on the other vector ( I have 2 cointegrating relationships) I get the results of zero degrees of freedom. Not sure why. Any assistance would be appreciated. I can impose restrictions on both vectors and test with no problem.

Stephen Pollard

[TomDoan]

I have a question about setting up restrictions on beta to do hypothesis testing. If I try to impose restrictions on one vector and none on the other vector ( I have 2 cointegrating relationships) I get the results of zero degrees of freedom. Not sure why. Any assistance would be appreciated. I can impose restrictions on both vectors and test with no problem.

Stephen Pollard

How many restrictions are you imposing on the one vector?

[moderator]

I believe you'd have a non-zero degrees of freedom only for a non-identifying test (see note on page 58 of the CATS manual).

If you have the Juselius book (The Cointegrated VAR Model), you can find a fairly detailed discussion of the degrees of freedom issue in Section 12.3, starting on page 212. Also Section 12.4 (page 217) for cases where no test is done because the restrictions are achieved through linear combinations of unrestricted relations.

We're working on finalizing the second half of our Handbook for the Juselius text, which walks you through the examples including those from Chapter 12. I hope to have it ready soon.

Regards,
Tom

[spollard777]

Thanks Tom,

Yes, I ordered the book about a week from the Estima website. Just waiting for it to be processed and sent. I also downloaded the first part of the handbook as well. I think you are right. I am having a difficult time understanding, but hope the book helps, formulating the hypothesis I am testing to make sure the number of restrictions and the H or R matrix is correctly entered.

Regards,

Stephen Pollard