Do stated preference methods stand the test of time? A test of the stability of contingent values and models for health risks when facing an extreme event

This paper focuses on the temporal stability of dichotomous choice (DC) contingent values and models and the implication of time related aspects for benefits transfer as a viable valuation method. From a practical point of view, time plays an important role in benefits transfer. Results are used from studies often carried out years ago and the obvious question is to what extent the estimated values and models stand the test of time. In the case of value function transfer, this implies testing the robustness of the estimated coefficient estimates over time. The novelty of the temporal stability work presented here is that two identical DC contingent valuation (CV) studies of the health risks associated with bathing water quality were carried out before and during extreme weather conditions causing the closure of bathing locations for public health reasons. Differences in contingent values and models before and during the extreme event are tested. The extreme event causes the seasonal good in question to become temporarily scarcer and marginal WTP is therefore expected to be higher during the extreme event than before, but not that underlying preferences have changed. Hence, basically two things are tested: the effect of time on stated preferences for a seasonal good and the effect of the extreme event on these preferences. WTP values before and during the event appear to be robust, also when accounting for substitution effects. The results before and during the extreme event remain transferable when accounting for theoretically expected factors in a simple multivariate transfer model. However, when introducing additional and contextual 'ad hoc' factors in the model, the estimated models and their random components become significantly different and hence not transferable, suggesting that unobserved preferences may not have been randomly distributed across the two surveys after all.