estimating double-bounded dichotomous choice contingent valuation models using seemingly unrelated bivariate probit regressions

despite controversy, the contingent valuation method (cvm) is widely used for non-market valuation. among elicitation methods of cvm, the dichotomous choice (dc) has been paid attention. there are two types of dc methods: single-bounded dc (sbdc) and double-bounded dc (dbdc). dbdc is more efficient than sbdc. many of cvm studies in iran used dbdc but data analyses were done by logit model; that do not raise the efficiency of dbdc. the purpose of this study is to analyze dbdc data using seemingly unrelated bivariate probit regression. to do this, the dbdc cvm questionnaire designed to estimate willingness to pay (wtp) for preserving lilium ledebourii used; and filled in the center of guilan province by 177 respondents in 1390. data analysis was done in two ways: restricted and unrestricted in which in restricted models the equality of parameters in two models were imposed. results showed that number of significant variables is much more using restricted model than unrestricted model. also, wtp using first (model in which its dependent variable is the response to the first offered bid) and second (model in which its dependent variable is the response to the second offered bid) unrestricted models are 6650 and 6963 rls, respectively and is 7225 rls using restricted model. it’s proposed that the researchers of cvm use seemingly unrelated bivariate probit regression to analyze dbdc data for acquiring efficient estimates.