Welfare Evaluations in Contingent Valuation Experiments with Discrete Response Data: Comment

Abstract

In the case of private goods, individuals reveal their preferences in the market place. No similar market mechanism exists in the case of public goods. However, one obvious possibility is to ask people about their maximal willingness to pay for public goods or government expenditure (Johansson 1987, 1990, Miller). In some recent contingent market valuation studies people instead have been asked to accept or reject a specified amount of money for a change in the provision of a public good; different subsamples are confronted with different The resulting and responses usually are analyzed by logit or probit techniques (e.g., Johansson and Kristrom). This raises the question of how to compute the average willingness to pay when only yes and no responses are available. This point was advanced by Hanemann in his seminal 1984 paper on welfare evaluations with discrete responses; see also Hanemann (1987). However, we believe a minor imperfection in Hanemann's presentation may confuse and mislead practitioners. Hanemann's approach assumes that negative bids cannot occur but at the same time gives the impression that it also applies to models that allow for negative bids. We also extend Hanemann's discussion of the choice of money measure in cost-benefit analysis.