Discrete choice experiments for valuing biodiversity conservation in Finland

This study deals with stated preference (SP) surveys for measuring the public's willingness-to-pay (WTP) for conservation of forest biodiversity in Finland. Different econometric models for double bounded contingent valuation (CV) and contingent rating and ranking (CR) data are examined, alternative ways to address parameter heterogeneity are analyzed, and pooled econometric models are used in testing for the equality of CV and CR data.; Data were collected using a mail survey that was sent in 1999 to 1740 Finns. Part of the sample received a CV questionnaire, others were sent a CR questionnaire. The CV and CR questionnaires were identical except for the valuation question. The CV respondents answered two consecutive WTP questions, whereas the CR respondents rated the status quo and three new conservation programs. Data were analyzed by using fixed and random coefficient logit (FCL and RCL, respectively) models. From statistical viewpoint, the RCL models were superior. Model specification was examined by using Likelihood Dominance Criterion by Pollak and Wales (1991) for non-nested models. The preferred models were estimated on combined CV and CR data, and used in testing for invariance between the CV and CR data. Successful pooling of the CV and CR data required estimation of data-specific scale factors. Models for combined CV data and CR data on the preferred choice accepted the pooling hypothesis, whereas the models for pooled CV data and CR data on two and three ranks rejected it.; Although models that accounted for observed heterogeneity were preferred over homogenous FCL models, the RCL formulation with unobserved heterogeneity performed statistically better than the models with observed heterogeneity. Sensitivity of the WTP to the magnitude of conservation was dependent on the estimated response model. WTP was calculated for a variety of models. The RCL model for pooled CV data and CR data on the preferred choice was chosen as the preferred model for evaluating conservation policies. The WTP estimates for the 25-, 50-, and 75-percent conservation programs were {dollar}43, {dollar}79, and {dollar}72.; Suggested future research included (i) employing the data pooling method in examining further the differences between SP methods, (ii) modeling preferences by novel distributional specifications, (iii) modeling simultaneously the unobserved and observed heterogeneity, (iv) sorting the data, perhaps along the lines of environmental and other attitudinal responses, into groups that have different parameter distributions.