Construction And Empirical Research Of Evaluation Model Of Dichotomous Choice CVM Based On Double-Hurdle Model

Contingent Valuation Method is one of the most widely used and the most influential non-market valuation techniques. Dichotomous choice questionnaire is considered the most advanced elicitation method of CVM among all kinds of CVM questionnaire formats. The data derived from dichotomous choice questionnaire is a kind of discrete data which contains only “Yes” or “No” and it is tough to handle and difficult to deal with. The calculation model of dichotomous choice CVM is the hotspot and the difficulty of current research.The sample data derived from dichotomous choice questionnaire usually contains zero observations. Zero observations were not treated properly by the calculation model of dichotomous choice CVM in previous studies, which lowered the accuracy of estimation result. Focused on zero observations problem, the paper analyzed the decision procedure of CVM participants and divides the decision procedure of participants into participation decision and payment decision and gave zero observations two kinds of explanations,one of which is due to the negative attitude of the respondents who fail to express a true WTP to the subject to visit, named Protest Zero, the other one of which is due to a zero WTP to the environmental good, named Genuine Zero. A double-hurdle model was constructed to involve zero observations into CVM data analysis and a parameter estimation method was proposed. In addition, the relevant double-hurdle model and the WTP double-hurdle model were built based on the original double-hurdle model, the differences of which were compared afterwards.Taking the Sanjiang Plain Wetland Ecological Protection value as an example, the mean WTP of Hei longjiang citizens calculated by original double-hurdle model is 127.93 yuan per person each year, which is lower than the result of WTP function model and higher than the results of Tobit model. The result tested and verified that Double-Hurdle model can avoid the overestimate of WTP lead by deleting zero observations and the underestimate caused by classifying all zero observations as genuine zero observations. A similar result was obtained from both the relevant double-hurdle model and the WTP double-hurdle model. Meanwhile, the results showed that the decision procedure is explained better by the double-hurdle model than by Tobit model. By comparing two hurdles, we found that factors such as income, level of acknowledgment, times of visits act differently in two hurdles. It showed that participation and payment were two kinds of mechanisms which needed to be treated accordingly. The relevant double-hurdle model, WTP double-hurdle model obtained a similar result with that of the original double-hurdle model while the significance of both models is relative lower. By comparing the log-likelihood and AIC test, the paper took WTP double-hurdle model as the final calculation model and the Sanjiang Plain Wetland Ecological Protection value is 5.116 billion yuan in 2010. The study offers new approaches for the handle of zero observations in CVM and provides reference for ecological environment management.