A Multivariate Arima Model to Forecast Air Transport Demand

Forecast of air transport demand has a great influence on the development of airport master plans with respect both to airside (runways, taxiways, aprons, technological devices) and landside (boarding/landing area, waiting rooms, etc.), given that it depends on the amount of passengers during the reference time period, usually the year or more years for such aim. As a result of deregulation and increases in travel opportunity, the air demand is continuously increasing, despite some negative peaks due to political and/or market driven events that reduce the user willingness to travel. Furthermore, the offered services have quickly changed in the last years both in terms of trip organization and monetary costs, also because various alliances and mergers have occurred, together with the emergence of new air carriers on the market (Janic, 2000). As airport managers, carriers have also a great interest in the demand modelling and simulation, particularly when there is a competitive market and users can choose among different services. The task is not easy to accomplish, given the complexity of the current situation where more air carriers can compete by offering different fares, different origin/destination airports serving the same areas, different on board services and so on. Usually, the estimation of the air transport demand can be obtained by different models/methods, among which national(multi-mode) models, time series models and market surveys (RP – SP methods) are the most used. Multi-mode models proposed in the literature to forecast travel demand at a national level can be used to obtain an estimate of the air travel demand, by using random utility behavioural models to simulate mode choices (see for example Cascetta et al., 1995). However, they do not analyse the temporal evolution of demand nor the specificity of regional airports subject to increases in specific demand segments. Time series models have been widely used, among the others Melville (1998), Karlaftis and Papastavrou (1998), Abed et al. (2001), Postorino and Russo (2001), Hensher (2002), Postorino (2003), Inglada and Rey (2004), Lim and McAleer, 2002; Ling Lai and Li Lu (2005). However, mainly simple autoregressive time series models have been used, even if with explanatory variables, while there are very few examples of ARIMA models and no one on the calibration of univariate and multivariate ARIMA models in the specific topic of the air transport demand simulation for a regional airport. Finally, market surveys are a good sources of information, and airports as well as air carriers can collect data about passengers when they are waiting for travelling or on-board. However, only few airports regularly collect data about passengers, while data of air carriers are mainly based on customer satisfaction surveys or ticket sales information for financial purposes rather than demand