Incorporating habits into an agent based model of commuting mode choice

Abstract

Following a behavioural economics approach, which brings insights from psychology into standard economic models, the utility maximising model for the choice of journey mode is extended to include habitual behaviour. The model applies Triandis’ Theory of Interpersonal Behaviour (1977) to commuters’ mode choice decision with five choices: car, bus, rail, and cycle and the option not to travel. Following Triandis, the mode choice decision is determined by either the commuters intended mode, based on utility maximisation, or the mode they habitually used. Both are affected by facilitating conditions, such as the availability of a public transport service and the possession of a driving licence.

This paper focusses on an application of the model to demonstrate the evolution of bus patronage with and without the inclusion of habit as a behavioural influence in the context of a car-restraint policy based on periodic increases in parking charges and a policy of providing free bus travel for the elderly. The model is implemented using an agent based modelling approach; the advantages of which are further demonstrated through a refinement deploying latent class analysis to enable multiple-segmentation of the agent population. It is concluded that ABM techniques for mode-choice decisions are more resource extensive to establish, but these costs can be reduced by integration with existing analysis tools and are strongly justified by the additional functionality provided. This makes the ABM approach particularly relevant for assessing sustainable mobility policies, as the changes often relate to small groups of individuals, subtle personal characteristics and interpersonal interactions are often significant, and behavioural responses often show important time-lags following intervention.