On Information Criteria to determine the true lag for the Autoregressive models

Abstract

In this research ,we compare between the Information Criteria (Akaike, Final Prediction Error, Schwarz, Hannan-Quinn and the Akaike's information corrected criterion which developed by Hurvish and Tsai in 1989).In order to determine which criterion can be used to determine the probability of picking up the true lag for Autoregressive model for the data generating process from several Autoregressive models, when the error term for Autoregressive model is normally distributed and when the error term has ARCH(q) model with q=1,2 and also under structural break for error term .We obtained the results for different sample sizes by using simulation.