Modelling and Forecasting Monthly Airline Passenger Flows among Three Major Australian Cities

Abstract

The authors examine statistical properties of time series data for the monthly numbers of airline passengers for three major Australian city-pairs (Brisbane–Melbourne, Sydney–Melbourne and Sydney–Brisbane). It is found that the three time series show a co-movement in the long run, each exhibiting a stochastic trend and deterministic seasonality. The impulse response analysis is conducted to examine how these time series are related in the short run. The bootstrap method is used to test for the statistical significance of impulse response estimates. The evidence suggests that the time series for the Sydney–Melbourne route is highly influential on the others in the short run. A number of univariate and multivariate time series models are used for forecasting. It is found that univariate methods generate more accurate forecasts than multivariate models.