Application of sample selection model to double-bounded dichotomous choice contingent valuation studies

Modeling households' behavior with the data from a contingent valuation (CV) survey is often complicated by sample non-response, which can cause non-response bias and sample selection bias, leading to inconsistent parameter estimates and a distorted mean willingness-to-pay estimate. This paper reports the results of empirical tests for both biases using household survey data in which the double-bounded dichotomous choice CV question involved the benefit of a tap water quality improvement policy in Korea. No non-response bias, but sample selection bias, is detected in the sample. To correct for sample selection bias, a sample selection model is employed. The authors also discuss how failure to correct for bias may distort aggregate benefit estimates.