Benefit estimates for landscape improvements: sequential bayesian design and respondents' rationality in a choice experiment.

摘要

This article examines the non-market benefits of low and high impact actions on four major landscape components addressed by the Rural Environment Protection Scheme in Ireland, using a multi-attribute, stated-preference approach. A choice experiment was administered to 402 adults in 2003-04. Several methodological issues are addressed: the use of prior beliefs on the relative magnitudes of parameters, standardized description of different levels of landscape improvements via image manipulation software, adoption of efficiency-increasing sequential experimental design, and sensitivity of benefit estimates to inclusion of responses from "irrational" respondents. Bayesian design updating delivers significant efficiency gains without loss in respondent efficiency. The willingness to pay (WTP) estimates and the model fit are sensitive to the exclusion of irrational respondents. The general public is strongly in favour of the improvements in the rural landscape typically brought about by the Scheme. In conclusion, the benefits from improving the rural landscape are of considerable magnitude. It is suggested that choice experiment studies should incorporate procedures for identifying and screening out respondents with inconsistent preferences, and that WTP estimates should be evaluated for sensitivity to the inclusion and exclusion of such respondents.