Spike models in contingent valuation

Spike models allow for a nonzero probability of zero willingness io pay in referendum-style contingent valuation experiments. They include many of the often used models as special cases and allow various analyses that are not possible in standard applications. In this paper, the economic rationale for the spike models and their estimation techniques are discussed, and empirical illustrations are provided. One study suggests that mean willingness to pay is zero, and both studies strongly suggest that median WTP is zero. These results have some bearing on the continuing debate of the incentive properties of hypothetical surveys and the construction of scenarios in contingent valuation experiments.