The sample median as an estimator of population mean true willingness to pay under valuation uncertainty: a synthesis and analysis of the literature

This paper reviews the contingent valuation literature on hypothetical bias and valuation uncertainty, with special emphasis on measures of central tendency suitable for applied welfare analysis. In a theoretical scenario without valuation uncertainty, sample mean willingness to pay (WTP) is an estimator of population mean true WTP, the appropriate welfare measure. It is proposed here that this may not hold under valuation uncertainty. In this case, mean true WTP should be bounded by mean hypothetical WTP and mean actual WTP (in turn, mean actual WTP should exceed mean certain WTP). Based on a theoretical model and a review of existing empirical studies, median hypothetical WTP was found to be an estimator of mean true WTP. It is therefore proposed that contingent valuation researchers should consider using the sample hypothetical median as an estimator of population mean true WTP when supporting data on the degree of valuation uncertainty are absent. This could be a straightforward approach, e.g. in a low income country context, where the possibility of financing elaborate research designs accounting for valuation uncertainty is limited. It can also prove useful to revisit existing contingent valuation studies with these findings in mind.