New interval forecast for stationary autoregressive models

Abstract

In this paper, we proposed a new forecasting interval for stationary Autoregressive, AR(p) models using the Akaike information criterion (AIC) function. Ordinarily, the AIC function is used to determine the order of an AR(p) process. In this study however, AIC forecast interval compared favorably with the theoretical forecast interval in an out of sample forecast performance. A simulation study was used to demonstrate the procedure.

Keywords: Autoregressive, Akaike information criterion, interval forecast, sieve bootstrap

International Journal of Natural and Applied Sciences, 6(4): 471 - 477, 2010