Interval Forecasting for Airline Passenger Demand

Abstract

In the past forty years, the volume of air travel has increased tenfold. While demand growth for OECD markets has slowed down in recent years, there is still a potential for markets where air services are currently not provided. It is the goal of the airline network development process to locate these potentially profitable markets. One of the main tools to support this process is passenger demand forecasting. However, experience has taught us that forecasting is inherently uncertain. The primary goal of this thesis is to present how interval forecasts of passenger demand can be created by using Monte Carlo simulations. The results show that, by combining a causal econometric prediction model with interval forecasts of time-dependent parameters, it reliable interval forecasts for demand can be produced. The secondary goal is to demonstrate the effects of using these intervals in a simplified airline network development model. By using actual airline data, it was concluded that the robustness of this process increases. This is particularly relevant for airlines, as they require confidence when making decisions about the development of their network.