Travel demand modelling and Prospect Theory

According to Prospect Theory, the value people attach to goods depends on their relevant reference point. People tend to value goods more if they already possess them (‘loss aversion’). Prospect theory raises fundamental questions on how to undertake cost-benefit analysis. For instance, for the valuation of travel time savings due to an infrastructure project, we would need to know the reference points for all the people affected by the project. However, there are few ‘natural’ reference points in mobility behaviour. On the other hand, loss aversion tends to disappear in routine behaviour such as daily commutes or even weekly shopping.

The central value judgement in neoclassical cost-benefit analysis is that the project should be executed if total WTP exceeds total WTA.

Economic theory postulates that the WTP and WTA should be approximately equal. However, this prediction is at odds with one of the seminal contributions of behavioural economics: Prospect Theory. This theory postulates that, at an individual’s reference point, there is a large difference between WTP and WTA (Kahneman and Tversky, 1979). For instance, in a famous behavioural experiment (Kahneman et al., 1990), randomly selected participants were given a mug, and then they were then given the possibility to sell it or trade it. Kahneman et al. found that the sum participants required as compensation for a mug they owned (the WTA) was approximately twice as high as the amount they were willing to pay to obtain the mug (the WTP). Attributing a higher value to goods that one already possesses is called ‘loss aversion’.

Thus, according to Prospect Theory, the value people attach to goods depends on the relevant reference point. This reference point is generally (but not always) the status quo: people value goods more if they already possess them. This implies that choices are context dependent.

available transportation options, and the identification of areas that are currently poorly served by transport services in general (Rainwater et al., 2015).

Requesting the platforms to make the data publicly available could be seen as a compensation for their free use of the road infrastructure constructed and maintained with public money. On the other hand, in exchange for anonymised data, the transport authority could include the shared modes in its official route planning apps .

It is still possible that shared mobility will lead to a decrease in transit use in some areas without negative side-effects. For instance, there are definitely some niches where micro-transit is likely to outperform traditional transit services. This is for instance the case with commutes to employment centres in areas with low density.

Moreover, the rise of AVs will reduce the opportunity cost of travel time and the time spent in congestion, and this will further undermine the competitive position of some transit services. The correct pricing of all transport modes according to their social costs will ensure that society will be able to capture the benefits of these innovations, while avoiding some of the possible disadvantages.

There is no obvious reason why one of these possible reference points should be more relevant than the others.

Another complication is that the relevant reference point may depend on the characteristics of the individual. For instance, the preferred departure time is affected by the reference time of arrival. This in turn depends on the flexibility one has in the starting time of the work day: if one is working on an assembly line, arriving exactly on time is crucial. However, for people in managerial positions, the exact arrival time matters much less (Van de Kaa, 2010a).

In models of vehicle choice, it has been proposed to use the current car owned by a person as a reference point. However, cars are so-called multi-attribute goods: people value them for several reasons, including the provided comfort, acceleration and power, fuel efficiency, trunk volume etc. It is an open question whether the individual will consider reference levels for each individual attribute, or rather for the general impression of the good (Van Wee, 2010).

Thus, if Prospect Theory is a good description of travel choices under uncertainty, then it undermines the use of cost-benefit analysis for transport policy appraisal.

However, the supporting empirical evidence for Prospect Theory is largely based on gambling experiments in which subjects are requested to choose one of two prospects. In real world transport settings, travellers have many more options open to them. In the context of departure time, for instance, the consequences of arriving late can be remedied by calling ahead, working more efficiently, etc. (Timmermans, 2010).

Moreover, transport choices are seldom one-off. As choices are repeated over time, people gain experience and learn about the consequences of their choices. In experimental settings, loss aversion disappears after extensive instruction or when it becomes a routine (Avineri, 2012; Van de Kaa, 2007).

Thus, for regular travel behaviour such as the daily commutes or even weekly shopping, loss aversion may be a minor issue after all. For non-routine travel, such as leisure trips, there is less scope for learning, but the consequences of choosing the wrong route or leaving too late are also less dramatic.

In summary, the implications of behavioural economics for policy instrument design appear to be much more far-reaching than the implications of Prospect Theory for cost-benefit analysis.