New approaches for determining the degree of differencing necessary to induce stationarity in ARIMA models

Abstract

In this article we present two methods for determining the degree of differencing required to induce stationarity in the data. These procedures are iterative and consist in systematically fitting increasing order ARIMA and ARI structures to the data, and then verifying that the resulting residuals behave like white noise using an autoregressive order determination criterion. Simulation results of different model structures with varying number of observations are used to evaluate the approaches.