Common features testing

For technical questions regarding estimation of single equations, systems, VARs, Factor analysis and State Space Models in EViews. General econometric questions and advice should go in the Econometric Discussions forum.

Moderators: EViews Gareth, EViews Moderator

nightwalk
Posts: 4
Joined: Tue Nov 25, 2008 6:00 am

Common features testing

Postby nightwalk » Tue Nov 25, 2008 6:22 am

Hi,

Can anyone tell me how to estimate the Engle and Kozicki (1993) common features testing step-by-step?

Thanks

ashtongate
Posts: 21
Joined: Thu Sep 25, 2008 10:39 am
Location: Washington DC

Re: Common features testing

Postby ashtongate » Tue Nov 25, 2008 8:32 am

Best way is to follow the method in Vahid and Engle (1992 I think). This can be done quite easily in eviews. Try the following:
1. Find the optimal lag length of the system (using level VARs)
2. Construct a matrix to be used to check for common cycles. It is used to find canonical correlation coefficients (or eigenvalues) it looks like:
mat1 = @inverse(matyy)*matyx*@inverse(matxx)*matxy
where y is the vector of log changes at t and x includes the lagged changes and the cointegrating relationship.
3. Tricky part -- find the eigen values of this matrix. It is non-symm so you have to use another program (eviews can only do symm)
4. Bring the eigenvalues back into eviews and set up the test statistics that use these eigenvalues
5. Now you should know the # of common trends and cycles and you can estimate the whole system using 3sls or some other estimation method.

Good luck!

nightwalk
Posts: 4
Joined: Tue Nov 25, 2008 6:00 am

Re: Common features testing

Postby nightwalk » Wed Nov 26, 2008 4:05 am

Thank you for the tips! Is there any way i can contact you for further clarifications?

ashtongate
Posts: 21
Joined: Thu Sep 25, 2008 10:39 am
Location: Washington DC

Re: Common features testing

Postby ashtongate » Wed Nov 26, 2008 7:54 am



Return to “Estimation”

Who is online

Users browsing this forum: No registered users and 36 guests