对自然资源产生的社会福利进行货币化计量是目前资源与环境经济学研究的焦点之一,条件估值法(CVM)在评估自然资源福利的研究方面具有独特的优势。1984年提出的基于线性效用差下福利计量的参数模型及相应的表征尺度在后来的CVM研究中得到了较多的应用,但已有研究多侧重于方法的应用,对模型和尺度的理论有效性缺乏实证性的探讨。基于随机效用最大化原理对消费者偏好函数模型及相应的福利计量尺度进行了较为详细的数学推导(以Logit模型为例)。以上海某城市河流资源CVM福利计量为例,在实证分析过程中结合国外已有研究,将线性模型与基于非线性效用差下的对数模型进行了比较,给出了在CVM研究中对自然资源进行福利计量建议采用的参数模型与表征尺度。

Estimating the social welfare produced by natural resources is one of the focuses in the field of resources and environmental economics.Contingent valuation method (CVM) is one of the most dominating techniques for the estimation.Although the parametric model based upon linear utility difference was preferred in the subsequent studies since 1984 when Hanemnn incorporated logit model into CVM,yet with the fast development of computer software,the problem is advanced that whether the logit model can well fit the natural resources demand curve and whether it is appropriate for the welfare measures used in the previous studies need to be clarified.We firstly deduced in details the parametric models and the corresponding welfare measures supported by the theory of random utility maximization,then in a case study the fitnessf good comparison between linear models and loglinear models based upon ad hoc utility difference was made.It is showed in this paper that:(1)Linear models such as logit advanced by Hanemann can well fit the Willingness to pay (WTP) curve and measure the welfare change of nalural resources,and the zero WTP can be adjusted by Spike model in linear models.Although there is no significant difference in the results such as mean WTP between linear and log linear models,yet this can not cover the possible truth that the significance level of linear models may be obviously lower than the log linear ones.(2)Log models based upon non linear utility difference can especially in the tails fit the WTP curve very well,which may be more appropriate for the description of the natural resources demand curve but not private goods.There is no significant difference between log logit and log probit models.(3)Median WTP is sensitive to the distribution assumption of consumer preference,and there exists difficulty in measuring the total social welfare because of the definition of the median.We suggest that the mean WTP calculated under log logit model should be used to measure the welfare change of natural resources in contingent valuation method,which can be got by numerical integration in the interval of from zero to the maximum bid amount.(4)Log linear models effectively excluded the negative WTP and make the models more powerful for the valuation program for resources protection or improvement project,but log models can not consider the zero WTP while the linear models can,and log models will bring obvious fat tail effect in the calculation of mean WTP,so more discussion and improvement are needed for the above flaws of log models.