Multiple bounded discrete choice contingent valuation: parametric and nonparametric welfare estimation and a comparison to the payment card

In multiple bounded discrete choice (MBDC) surveys, respondents indicate how certain they would be to vote in favor of a policy at different prices by choosing, for example, among “definitely yes”, “probably yes”, “not sure”, “probably no”, and “definitely no” response options for each price. In estimating non-market values from MBDC data, past researchers have made markedly different assumptions with respect to the assumed correlation of within-respondent decisions (one for each price) and the correspondence of stated payment certainty to actual behavioral intentions. The first objective of this paper is to provide guidance for future research efforts by discriminating between existing models and proposing new estimators that relax some important statistical assumptions of existing models. Contrary to a previous study, results in this paper suggest that within-respondent decisions should be treated as being perfectly correlated. The second objective is to examine whether it is worthwhile to collect the additional information on payment certainty, as it may place additional cognitive burden on respondents as well as data analysts. Using data from previous studies, MBDC is compared with the payment card, a related elicitation approach that does not gauge payment certainty. This comparison provides strong and systematic evidence that “definitely yes” and “probably yes” MBDC respondents would vote “yes” while other respondents would vote “no” in the absence of the certainty categories.