GMM non-nested tests.

[pete]

I'm trying to test 2 New keynesian Phillips curves. Both equations have the same dynamics but have different driving variables and both are estimated by GMM.

For the purposes of comparing and discriminating between these two specifications, I wondered therefore if anyone has any code to perform non-nested GMM tests. I was thinking of perhaps the model selection tests of Rivers-Voung 2002, Hall-Pelletier-2007 or encompassing tests appropriate in a GMM context.

Many thanks in advance.

[TomDoan]

Doesn't the Hall-Pelletier paper show that the ranking is somewhat unreliable because of the sensitivity to the choice of weighting matrices?

[pete]

Tom
Yes, indeed, most of these tests have some fragility or another. The Rivers-Voung test is just one possible candidate among the class.
If anyone has any advice or code for GMM non-nested coding, I’d still certainly be interested to hear about it. Thanks a lot.
Regards.

[Tranquyl]

Hi,

I am also interested in testing two GMM models. May I know if you managed to develop a code for this please?
Any help is appreciated.

Rgds,