Comparison of New Approach Criteria for Estimating the Order of Autoregressive Process

Abstract

This paper compares the performance of the former model order determination criteria with new Approach in terms of selecting the correct order of an autoregressive time series model in small and large samples. The considered criteria are the Akaike information criterion (AIC); Scwarz information criterion (SIC), the Hannan − Quinn criterion (HQ) and the three new median criteria. Our results show that the median between AIC and Scwarz (MAS) criterion performs best in terms of selecting the correct order of an Autoregressive model for small and large samples sizes. Scwarz, Hannan Quinn, the median of Akaike and Hannan Quinn (MAH) criteria and median of Scwarz and Hannan Quinn can be said to perform best in large samples. The comparison of the six model selection criteria was in terms of their percentage of number of times that they identify the true model. The simulation results indicate that overall, the proposed Median of Akaike and Scwarz (MAS) information criterion showed very good performance in this simulation study. So, we recommend median of Akaike & Schwarz (MAS), Median of Schwarz & Hannan Quinn (MSH), and median of Akaike &Hannan Quinn information criterion as reliable criteria to identify the true model.