Estimating willingness to pay for improved water quality in the presence of item nonresponse bias

This article deals with the problem of item nonresponse in contingent valuation surveys using a payment-card method, by applying a grouped-data sample-selection estimation technique that is capable of imputing the missing values conditional upon a respondent's decision to answer a willingness-to-pay question. The advantage of the technique lies in its ability to utilize all of the information in the sample, permitting a more efficient estimation in the presence of item nonresponse bias. The major determinants of willingness to pay appear to be household income, number of children, education, perception of existing water quality, and identification with environmental issues.