cointegration with 2 structural breaks

[TomDoan]

Again, your series as you describe them do not fit into the K&P theorem. You would have to check with Perron to see if somehow they've extended that to more general regressors. But as the paper is written, they have to be I(0) or I(1) without breaks---all that's allowed are error processes that allow for a functional CLT.

[elhampa]

Hello Tom,

Thanks for all your support. Would you please advise me whether the test proposed by Kejriwal and Perron {Kejriwal, Mohitosh & Perron, Pierre, 2010. "Testing for Multiple Structural Changes in Cointegrated Regression Models," Journal of Business & Economic Statistics, American Statistical Association, vol. 28(4), pages 503-522}, is robust if the dependent variable is I(0).

[TomDoan]

So far as I can tell, the paper is silent on whether the dependent variable is I(0) or I(1). However, you really need to address questions like this to Prof. Perron. This is the third time I've suggested that. Maybe he has some power points that describe this paper in a less technical fashion. Note that we have nothing at all to do with this paper. We didn't write it; we have no examples of its use. If you really need to understand the theory behind it, you need to talk to the people who wrote it.

[elhampa]

Hello Tom,

Would you please advise me whether lagged dependent and independent variables can be inserted in the model using Kejriwal, Mohitosh & Perron, Pierre, 2010. "Testing for Multiple Structural Changes in Cointegrated Regression Models," Journal of Business & Economic Statistics, American Statistical Association, vol. 28(4), pages 503-522. ?

[TomDoan]

Look carefully at equation (1). That's what's covered. It's basically a static regression (as in an Engle-Granger regression). Any other dynamics are in the error term. Again, we have no examples of this. So far as I can tell, Perron doesn't have any examples of it. PLEASE, if you really want to figure out how to use this, contact Prof. Perron.