Page 1 of 1

Estimated Restricted Models Using the GARCH Instruction

Posted: Mon Nov 27, 2006 12:58 pm
by Gregory
Boy, that GARCH instruction is nifty. Big improvement over my pages of code from years ago.

I'm estimating an asymmetric BEKK model using the GARCH instruction, and I apologize in advance for my rustiness, but I'm not quite sure how to impose restrictions on the parameter matrices so that I can do likelihood ratio tests. The usual battery of tests would be for seeing whether the general of umbrella model has significantly more explanatory power then certain nested models. I'd like to test the general against D=[0], that is, zeroing out the D matrix, then test the general against A=B=D=[0].

Any advice would be most appreciated.

Regards,

Gregory