Determinants of the Degree of Loss Aversion

Determinants of the Degree of Loss Aversion


K Hjorth, Technical University of Denmark and University of Copenhagen, DK; M Fosgerau, Technical University of Denmark and Centre for Transport Studies, SE


What determines how loss averse respondents in SP exercises are? Socioeconomics, trip characteristics or strength of the reference?


Stated preference data are used extensively in transportation to measure relationships that are hard to measure based on actual choices. Data are generally analysed assuming some specification of a utility function that is taken to represent the preferences of the respondents also in real life. These preferences are assumed to be rather stable over time, such that they can be used as a basis for forecasting and policy evaluation.

It is becoming increasingly clear that this view is too simple. Specifically, evidence is being collected that preferences, at least in SP situations, are reference-dependent. Recent examples are De Borger & Fosgerau (2007), Hess, Rose & Hensher (2007), Ramjerdi & Dillén (2007).

Loss aversion has been advanced as an explanation. A clear reference situation is defined in many SP exercises. Under loss aversion, characteristics of the choice alternatives are compared to the reference one by one, the disutility of a loss is larger than the utility of a same sized gain.

This model has been shown to describe SP data extremely well. It explains the large gap between WTP and WTA that has been found repeatedly. E.g., Horowitz & McConnell (2003) review more than 200 studies and find an average gap of 7!

The issue of the current paper is to investigate which factors influence the degree of loss aversion. Using data from the Danish value of time study, we are able to investigate the influence on loss aversion of

1. socioeconomics such as sex, age, income.
2. trip characteristics such as mode, length and duration.
3. strength of the reference, comparing experiments involving chosen and alternative transport mode.

The results will have implications both for the understanding of the phenomenon of loss aversion as well as for the design of SP experiments. A general question that this research may help answering is whether it is pertinent to seek to strengthen or weaken the reference when designing experiments.


Association for European Transport