Asking for Individual or Household Willingness to Pay for Environmental Goods?

Abstract

The aggregate welfare measure for a change in the provision of a public good derived from a contingent valuation (CV) survey will be much higher if the same elicited mean willingness to pay (WTP) is added up over individuals rather than households. A trivial fact, however, once respondents are part of multi-person households it becomes almost impossible to elicit an uncontaminated WTP measure that with some degree of confidence can be aggregated over one or the other response unit. The literature is mostly silent about which response unit to use in WTP questions and in some CV studies it is even unclear which type has actually been applied. We test for differences between individual and household WTP in a novel, web-administered, split-sample CV survey asking WTP for preserving biodiversity in old-growth coniferous forests in Norway. Two samples are asked both types of questions, but in reverse order, followed by a question with an item battery trying to reveal why WTP may differ. We find in a between-sample test that the WTP respondents state on behalf of their households is not significantly different from their individual WTP. However, within the same sample, household WTP is significantly higher than individual WTP; in particular if respondents are asked to state individual before household WTP. Our results suggest that using individual WTP as the response unit would overestimate aggregate WTP, and thus bias welfare estimates in benefit-cost analyses. Thus, the choice of response format needs to be explicitly and carefully addressed in CV questionnaire design in order to avoid the risk of unprofitable projects passing the benefit-cost test.