Mixed logit estimation of willingness to pay distributions: a comparison of models in preference and WTP space using data from a health-related choice experiment

Different approaches to modelling the distribution of WTP are compared using stated preference data on Tanzanian Clinical Officers’ job choices and mixed logit models. The standard approach of specifying the distributions of the coefficients and deriving WTP as the ratio of two coefficients estimation in preference space) is compared to specifying the distributions for WTP directly at the estimation stage (estimation in WTP space). The models in preference space fit the data better than the corresponding models in WTP space although the difference between the best fitting models in the two estimation regimes is minimal. Moreover, the willingness to pay estimates derived from the preference space models turn out to be unrealistically high for many of the job attributes. The results suggest that sensitivity testing using a variety of model specifications, including estimation in WTP space, is recommended when using mixed logit models to estimate willingness to pay distributions.