A methodology for evaluating reliability of benefit of infrastructure development projects by use of discrete choice models

近年来出版了许多实用的成本效益分析手册，以评价各类基础设施发展项目的效率，以满足日益增加的问责制需要。与此同时，公众对自然环境的关注日益增加，对基础设施建设项目的环境影响进行评估是不可避免的。成本效益分析手册还包括对项目造成的环境质量变化的效益测量的说明。虽然评估项目所衡量的效益和成本的可靠性是非常重要的，以便使项目评估更负责任，但这些手册既没有很好地建立也没有考虑到评估成本效益分析可靠性的方法。本研究的重点是对涉及环境质量变化的基础设施建设项目效益可靠性评估方法的方法论发展及其在项目评估中的应用。分别使用两种不同类型的离散选择模型，即陈述偏好模型（state preference model， SP）和揭示偏好模型（revealed preference model， RP）来描述条件评价调查中被调查者的问卷调查行为和用户前往自然环境丰富地区的行为。该研究提出了评估基于RP或基于SP的离散选择模型的效益可靠性的方法。对于SP模型，研究成功地得到了效益估计置信区间的精确估计。然而，对于RP模型，由于模型是非线性的，因此测试了Talor-Maclaurin展开、Edgeworth展开等近似方法以及Bootstrap方法、Jack-knife方法和Monte Carlo方法等基于仿真的方法。通过蒙特卡罗仿真验证了所提方法可靠性评估的准确性。结果表明，基于模拟的方法适合足够数量的试验。本文还通过琵琶湖和Kizu河的实例研究验证了方法的适用性。在利用离散选择模型评估效益可靠性的基础上，提出了含噪声的工程效率评判方法，并提出了一种图形化的评判方法。少

Many manuals of practical cost-benefit analysis to evaluate efficiencies of various types of infrastructure development projects have been published in recent years as to respond to increasing needs for accountability. Simultaneously, growing public interest in natural environment makes inevitable to assess environmental effects of an infrastructure development project. The cost-benefit analysis manuals also contain explanations about benefit measurements of environmental quality change due to the projects. Although it is very important to assess reliability of the measured benefit and cost of projects as to make the project evaluation more accountable, methodology to assess reliability of cost-benefit analysis is neither well established nor considered in these manuals. This study focuses on methodological development of a methodology to assess reliability of benefit of a infrastructure development project which involves environmental quality change and its application for project ado … More ption judgment. Two different types of discrete choice models are used for describing respondents' behavior in questionnaire surveys in contingent valuation surveys and users' behavior to make trip to the area which has rich natural environment, i.e., stated preference model (SP) and revealed preference model (RP), respectively. The study proposed methods to evaluate reliability of benefits measured either by a RP based or a SP based discrete choice model. For SP model, the study successfully could derive exact estimator of confidence intervals of estimated benefits. However, for RP model, since the model is non-linear, approximation methods such as Talor-Maclaurin expansion, Edgeworth expansion and simulation based methods such as Bootstrap method, Jack-knife Methods and Monte Carlo based method are tested. The accuracy of evaluation of reliability obtained by proposed methods are checked by Monte Carlo simulation. The results suggest that simulation based methods fit well for enough number of trials. Applicability of the methods are also examined by case studies in the Lake Biwa and Kizu river. Methodology to judge project efficiency with noise is also proposed based on the methods to assess the reliability of benefits measured by use of discreate choice model and a graphical method to help the judgment is also proposed. Less