Page 1 of 1
Simple GARCH(1,1) Log Likelihood First Order Derivatives for Conditional Variance Functions Parameters
Posted: September 13th, 2011, 12:23 pm
by Dziemian
Hi,This is actually my first post here so I would like to say 'Hello!' to everyone. Now let's get to the point.I'm working on GARCH(1,1) code in excel and I'm in need of analytical first order derivatives of GARCH(1,1) Log Likelihood for conditional variance function (their final form) to compare it to my own results or is it OK to obtain them using finite difference method? I would be grateful for any help.Additionally I have few other questions:- What is the approximately correct value of first order derivatives for each observation? Should it be close to zero (for example <0.01)?- Similar question about value of 'Direction Vector' in BHHH algorithm?- What is the best way to calcualte step size in algorith mentioned above?Many thanks in advance!
Simple GARCH(1,1) Log Likelihood First Order Derivatives for Conditional Variance Functions Parameters
Posted: September 22nd, 2011, 6:57 pm
by AVt
Guess you can find it in some of the library codes, anyway here some stuff in C
Simple GARCH(1,1) Log Likelihood First Order Derivatives for Conditional Variance Functions Parameters
Posted: September 26th, 2011, 2:58 pm
by Dziemian
AVt,Many thanks for that. I'm going to try this out as soon as i get back from work .Regards,Dziemian
Simple GARCH(1,1) Log Likelihood First Order Derivatives for Conditional Variance Functions Parameters
Posted: October 1st, 2011, 3:02 pm
by Alan696
I've got RATS Algo if needed
Simple GARCH(1,1) Log Likelihood First Order Derivatives for Conditional Variance Functions Parameters
Posted: October 4th, 2011, 12:13 pm
by Dziemian
Alan969,It could be useful as well so feel free to post it here. Thanks in advance.
Simple GARCH(1,1) Log Likelihood First Order Derivatives for Conditional Variance Functions Parameters
Posted: October 25th, 2011, 10:05 pm
by Alan696
There is actually an integrated command othwerwise check RATS manual 7.0. You can also check on
www.estima.com