A Comparison of Willingness to Pay Estimation Techniques From Referendum Questions

Referendum style willingness to pay questions have been used to estimatepassive use values. This referendum question format method may beproblematic for many reasons, including the statistical techniques used toestimate willingness to pay from discrete responses. This paper comparesa number of parametric, semi-nonparametric and nonparametric estimationtechniques using data collected from US households regarding Federalprotection of endangered fish species.The advantages and disadvantagesof the various statistical models used are explored. A hypothesis test forstatistical equality among estimation techniques is performed using ajackknife bootstrapping method. When the equality test is applied, themodeling techniques do show significant differences in some possiblecomparisons, but only those that are nonparamentric. This can lead toconflicting interpretations of what the data show. Resource managers andpolicy analysts need to use caution when interpreting results until anindustry standard can be developed for estimating willingness to pay fromclosed ended questions. Copyright Kluwer Academic Publishers 2001