EViews.com

Hi everybody
I think this is a very simple thing but I need to be sure I understand correctly so I want to hear what you think.
I have estimated several competing models with maximum likelihood and each of them gives me a negative value of the loglikelihood. Assume that I want to compare all the estimated models only on the basis of the loglikelihood they achieve: am I right if I interpret higher values of the loglikelihood (closer to zero) as indicating a better fitting model? :)

Thanks

Yes, the closer the values are to zero, the higher the log-likelihood function in the case of negative values. However, if you want to compare competing models, you should use information criteria instead of log-likelihood. EViews, for example, reports several of them (e.g. Akaike, Schwarz, etc.). Moreover, most information criteria are based on log likelihood function, adjusted by a penalty function. While the higher log-likelihood "may" indicate a better fit, usually the model with the smallest information criteria is the preferable one.

Thanks! that's exactly what I am doing but I needed to be sure!