The RATS Software Forum

Dear Tom,
Dear All,
I am working on the example file from Switching Model. Is it possible to get some help on the design of a test about the number of regimes/states ? I am thinking about either teh Hansen (1992) or BIC, Hannan Quinn or AIC criteria ?
Can I get some help about that ?
Thank you very much in advance,

Max Marzo

MaxMarzo2013 wrote:Dear Tom,
Dear All,
I am working on the example file from Switching Model. Is it possible to get some help on the design of a test about the number of regimes/states ? I am thinking about either teh Hansen (1992) or BIC, Hannan Quinn or AIC criteria ?
Can I get some help about that ?
Thank you very much in advance,

Max Marzo

In practice, it seems to be mostly information criteria. And what also happens in practice is that if n regimes seems adequate, trying to fit n+1 as a comparison fails miserably.