One of the main interests in transport modelling work is to produce monetary valuations of various commodities. The prime example in this context is the valuation of travel time savings, i.e. how much is a traveller willing to pay for a reduction in travel time. However, there is also strong interest in the valuation of other components, such as frequency, delay, reliability, and safety. A monetary valuation for a component x is typically obtained on the basis of choice experiments that include this component x as well as a travel cost component, where a respondent is encouraged to trade between changes in this attribute and changes in cost. This stated preference approach is however by no means exclusive, and a large share of valuation work is still based on more direct approaches, such as transfer pricing, or revealed preference approaches.

Independently of the specific approach used, it is often the case that monetary valuations from one body of work are used to monetise non-monetary valuations in another study. As an example, a new study may present respondents with trade-offs between improvements in safety and improvements in travel time, but without including a travel cost component. After using data from this study to produce relative valuations of travel time and safety improvements, it is common practice to use an existing monetary valuation for one attribute, say time, to obtain a monetary valuation for the other attribute, in this case safety. This approach makes the significant although arguably defensible assumption that such monetary sensitivity isare transferable from one valuation to another. Given the reliance on such benefit transfer pricing procedures, it is important to test whether monetary valuations of different goods are consistent, i.e., that money is indeed fungible in the context of valuing transport benefits.

Although transferability and trade-offs of/between benefits are simply the basis of using DCM, it appears that the cost attribute presents a particularly worrying difficulty for this method.

In this paper, we present the findings from an experiment aimed at testing the validity of this assumption. In particular, we will examine trade-offs in choices made (1) between safety and money, (2) between time and money, and (3) between safety and time. For (1) and (2) subjects will be offered choices between various amounts of money and degrees of safety to determine a marginal valuation of safety, and the same will be done for time. Thus, we elicit a S/£ ratio, and a T/£ ratio. By taking the ratio of ratios, we get an elasticity of substitution between safety and time. We then elicit a direct elasticity of substitution between safety and time by eliciting choices between safety and time for particular journeys.

Standard expected utility theory suggests that the elasticity of S/£ over S/T should be the same as it is in the elasticity of substitution between S/T (where in the latter, choice tasks safety and time are traded off against each other directly without a cost attribute). If the elasticities across the valuation methods are not the same, we can reject the (null) hypothesis that money is fungible, and must entertain some alternative hypothesis explaining why different valuation tasks elicit different monetary benefits for time and safety.

An approach to developing an alternative hypothesis might draw from the study of preference reversals in behavioural economics, where choices between goods produce different valuations than direct monetary valuations of those goods (specifically, when a subject is presented with a choice between A and B, she chooses A, but when presented with a monetary valuation of A and B, she values (£)B as higher than (£)A). As applied in the transport case, we might find that safety, or time, is valued differently when traded off directly against one another, than when they are valued using money/travel cost.

If we find that the cost attribute elicits elicits different valuations of goods than when those same goods are traded off against one another without cost, it has profound implications for the valuation of all transport benefits, particularly the value of time. The DfT uses a standard value of time, and uses this to value the benefits in new projects such as reliability, comfort, frequency, or other benefits. The DfT also uses a value of preventing a statistical fatality (VPF) figure to estimate transport benefits. Are they consistent?

This paper suggests that they are not, with time saving being valued more highly when valued directly through willingness-to-pay than when inferred from the marginal rate of substitution with safety. Conversely, safety is valued less highly then valued directly through willingness-to-pay than when inferred from the marginal rate of substitution with time saving.