I have a time series that I am trying to fit to a GARCH model. When I use the Student T distribution, I receive correct results, but using the default Gaussian distribution I get incorrect values for GARCH(1), ARCH(1) and Sigma. In both cases, GARCHFIT converges with no warnings. For example,
% Download the attached MAT file
load series.mat% **** CASE 1 - T Distribution
spec = garchset('C',0,'K',0.0001,'GARCH',0.9,'ARCH',0.02,'Distribution','T','DoF',3);[Coeff,Errors,LLF,Innovations,Sigmas,Summary] = garchfit(spec,series);garchdisp(Coeff,Errors);plot(Sigmas);% **** CASE 2 - Default Gaussian
spec = garchset('C',0,'K',0.0001,'GARCH',0.9,'ARCH',0.02);[Coeff,Errors,LLF,Innovations,Sigmas,Summary] = garchfit(spec,series);garchdisp(Coeff,Errors);plot(Sigmas);
returns an accurate model and volatility plot for the first case but an incorrect model and a random volatility plot in the second case.
Best Answer