願意支付價值最佳效率指標之建構與驗證

Construction and Empirical Examination of Optimal Efficiency Index of Willingness to Pay

雖然目前條件評估法在文獻上已經發展出多種不同的誘導支付模式，但是因為這些模式都存在若干缺點，所以本文乃發展一個具有四階段決策過程的近似理想誘導支付模式。此模式是一種混合二元選擇與開放式之設計，其前三階段之決策過程猶如三界二元選擇，最後第四階段則採用開放式。同時，近似理想誘導支付模式乃是依循最適設計準則，來制訂前三階段的受訪額度。而建構該模式的目的，除了可以在理論上呈現出近似理想誘導支付模式各個階段的最佳估計效率外，更重要的是可以為後續的實證研究，在面對各種可能的實證估計結果時，提供一個可以比較的對照基準。為了瞭解理論與實證間各個階段估計效率的對照結果，乃選擇受訪者平均願意支付價格之變異數做為效率指標值。由於根據理論推導，將可以獲取近似理想誘導支付模式的最佳效率狀況，所以將此時計算得到的指標值，稱為最佳效率指標值，並可由此繪製出最佳效率邊界。而最佳效率指標值與最佳效率邊界即是實證估計時所能達到效率性的極限，二者可以做為各種實證估算結果之效率性與理論上各種理想狀態的比較。為了驗證不同實證估算結果效率性之表現，本文選擇黑面琵鷺保護區多元價值之問卷調查資料，分別檢驗僅考慮估算平均願意支付價格和涵蓋受訪者特質兩種狀況下，且在不同分配函數與估計函數形式下探討其效率性。由各種實證估計的結果顯示，實證上的效率性會與最佳效率狀況有所落差，這些差異可能源自於分配函數選擇的不同、涵蓋受訪者特質之解釋變數的多寡、甚至是估計函數形式的不同。亦即，為了實證之需要，必須選擇不同於理想狀況下之條件時，條件差異愈大則偏離最佳狀況的情形愈顯著，而本文所建構的模型，可以明確指出此時所犧牲之效率值的大小。

Although various elicitation methods have been developed in contingentvaluation method for non-market goods evaluation the shortcomings haveappeared in all the existing methods. An elicitation method developed bySu (2004), named almost ideal elicitation method (AIEM), is a triple-boundeddichotomous choice followed by a continuous open willingness to payrevelation. This elicitation framework has the advantage of offering aparadigm for empirical research in examining efficiency performance ofwelfare measure. That is, an efficiency frontier can be drawn and can alsobe used as a reference frontier for all other empirical estimation.In order to examine the relative efficiency performance betweentheoretical design from AIEM and those from various other empiricalestimation designs, an efficiency index is constructed for this purpose. Aset of data in evaluating the benefit of Black-Faced Spoonbill Protected Areain Taiwan is selected.The results show that all empirical efficiency performance of estimatedmean willingness to pay is lower than that from the theoretical derivation.And the differences arise either from more respondents characteristicsincluded, different selection of distribution, or different choice of estimationfunctional forms. The loss of estimation efficiency can clearly be indicatedfrom AIEM designed in this study.