The use of contingent valuation (CV) methods to estimate benefits has become increasingly common in project analysis. Ever since the NOAA Blue Ribbon Panel Report in 1993 (NOAA, 1993) recommended the use of the referendum form of CV, it seems to have become the method of choice in practical settings.

Referendum-type questions are thought to be easier to answer than the open-ended variety. But there is a downside: econometric techniques must be applied to the referendum data in order to infer the mean or median willingness to pay (WTP) of the sample and, thus, of the population of potential beneficiaries.

This is not, however, just a technical point. Its implications are demonstrated with data obtained from a referendum CV study done for a proposed sewer and wastewater treatment project designed to improve water quality in the Tietê River flowing through the city of São Paulo, Brazil. The results show that:

A factor of 4 separates lowest from highest central tendency estimates of WTP, ignoring one implausible outlier that is 14 times larger than the largest of the other figures.

This variation is ample enough to make a difference in the cost-benefit analysis results for the project under conservative assumptions.

Analysts that use referendum CV data must be sensitive to the problems they buy into, and decide how to deal with the resulting benefits uncertainty in their project analysis. If the principal use of CV survey data is to produce a mean or median estimate of WTP for Cost-Benefit analysis rather than to test for the factors influencing referendum choice responses and, by implication, WTP, nonparametric approaches have the advantage of simplicity over parametric approaches.