Deriving Willingness-to-Pay Estimates of Travel-Time Savings from Individual-Based Parameters

There is a small but growing literature that promotes the derivation of distributions of willingness-to-pay (WTP) estimates using information specific to each individual observation. These are referred to as individual conditional distributions, in contrast to approaches that rely on unconditional distributions that use random assignment in the construction of WTP distributions within a sampled population. The interest in alternative specifications is in large measure attributed to the search for empirical ways of deriving the WTP distribution that satisfies a behaviourally acceptable sign and range over the entire domain. In this paper we examine both conditional and unconditional approaches to establishing WTP distributions within the context of a mixed logit model. We find that calculating WTP measures from ratios of individual-level parameters in contrast to drawing them from unconditional population distributions empirically reduces the incidence of extreme values. Our results suggest that although problematic estimates cannot be ruled out, the use of the extra information on each individual's choices is a valuable input into the derivation of WTP distributions.