本文之目的爲闡述、彙整、比較並對照目前在效益評估上,使用頻繁的假設市場價值評估法中單界及雙界二元的選擇式資料,在學理詮釋上的差異,進而檢視理論上的差異是否對不同函數形式與估計模型之實證操作有不同的啓示,並檢視雙界二元選擇所獲得之效益估計效率性比單界選擇爲佳在效用差異詮釋下的結果,是否同時呈現於支出差異的理論詮釋上。本實證所用的資料,則是來自評估墾丁國家公園資源經濟效益的調查研究,而除了將單界與雙界的二元選擇資料,以效用差異與支出差異爲詮釋外,兩種詮釋亦同時在Logit與Probit模型下設定爲線型、線型對數、半對數、雙對數等多種函數形式,完成了前述多個面向的整合與比較分析。累積過去的經驗及本研究結果對後續採用此類型資料進行效益研究的啓示是,理論上互爲對偶關係之效用差異及支出差異,在實證上反映了極大的不同,以支出差異做爲單界或是雙界二元選擇行爲的詮釋,實證上要獲取效益的願付價格或信賴區間的可及性不僅相對簡易、也相對容易成功,且雙界各模型所估計之平均值的變異數均較小,且雙界平均值信賴區間值均呈現較集中,也就是較有效率的結果。相對的,將資料詮釋爲效用差異,則可估計之願意支付價格平均值與區間推定值則相當有限。
The purpose of this study is to provide a systematic analysis for the most commonly used discrete type of contingent valuation data for its theoretical plausibility and empirical feasibility, applicability, and convenience under utility difference and expenditure difference interpretation with linear, linear-log, semi-log, double-log functional specifications and Probit and Logit estimation models. The results indicate that the duality between the utility difference and expenditure difference has different implications in the estimation, computation of mean willingness to pay, and confidence intervals. The expenditure differences are empirically all feasible and applicable for all the specifications, but there are only limited numbers of cases that are feasible and applicable under the utility difference specification. Furthermore, the efficiency of mean willingness to pay and confidence interval of mean willingness to pay for double bounded data under all the specifications for the expenditure difference is consistently higher than their counterparts of single bounded data.