Validation of Empirical Measures of Welfare Change: Comment

In a recent article, Sellar, Stoll, and Chavas (1985) make a technical error that causes them to misstate their closed-end estimates of willingness to pay. The correct computation is necessary to make comparisons of values across estimation methods. Also, the procedure used to estimate expected values has implications with respect to the advantages and disadvantages of applying closed-end questions in contingent-valuation studies. Continuous distributions on the interval (zero, infinity) are appropriate for analyzing responses to closed-end valuation questions, and expected values should be derived using the formula presented. If possible, the range of integration should not be truncated, and fat tails that motivate truncating of the range arise because of the selection of a range and distribution of dollar offers. The sampling procedure known as the "method of complementary random numbers" appears to minimize the potential for the estimated cumulative distribution functions having fat tails.