Real Time Business Cycle Dating with DFMS

[danon]

Hi dear Mr Tom, I am working on Markov-switching Dynamic factor model.
I would like to know if you have a detailed examples for REAL TIME marov switching like in the attached paper (Codes are available but in GAUSS).
Need to buy that one if it exists.
Best regards

Danon

[TomDoan]

Am I missing something, or is this a 2012 working paper that's never been published?

The Kim filter is covered in detail in the Structural Breaks and Switching Models e-course. Aruoba-Diebold-Scotti uses mixed frequency data. It looks like what they're doing is a combination of the two of those. However, I would be somewhat skeptical about how well this actually would work---the Kim filter has accuracy problems to start (for computing an approximate log likelihood), and using it with a state-space model as vague as a factor model seems like a bad idea.

[danon]

Hi Tom,
I do understand, i am better use the GAUSS code available!?
Best regards

Danon

[TomDoan]

If you want to pursue that, yes. I just think you'll find that it's not worth the effort.