| Summary: | 碩士 === 國立臺灣大學 === 農業經濟學研究所 === 102 === Conducting contingent valuation method to estimate value of nature resources or environment, there are various reasons for respondents to give protest replies or reply don’t know/uncertainty. Past studies to deal with or estimate sample with these two groups of responses are either to delete the protest and don’t know/uncertainty samples or to consider them as zero values. However, deleting protest sample, especially, might overestimate the final mean willingness to pay. On the other hand, while accounting these samples as zero responses might underestimate mean willingness to pay. There are other methods to deal with protest and/or don’t know/uncertainty responses to solve the potential bias, all these methods, however, are in certain way too complicate to get appropriate estimation results, not to mention the computation of the mean willingness to pay.
Among all the elicitation methods employed in contingent valuation method, three types of responses are classified. The first type of elicitation approach is the mode of presenting willingness-to-pay value, open-ended, bidding game, et al. are belonged to this mode. The second type is pure binary choose mode. Single-bounded discrete choice and double-bounded discrete choice are members of this mode. The third type is mode of binary choose with open question follow-up. This study then structures a general model to systematically estimate all three types of elicitation modes mentioned above. Furthermore, this general model treats protest and/or don’t know/uncertainty responses sample in an appropriate way. Most importantly, the general model constructed in this study is easy to estimates and computation of mean willingness to pay will then not confront any difficulty accordingly.
Such general model is testified for two sets of data. They are a sample of benefit evaluation of Kenting National Park and a sample of benefit evaluation of Changhua Coast Wetland. The results from the general model constructed in this study demonstrate that the standard errors are lower than those accomplished by traditional models. This indicates that our general model is more efficient. As a result, the mean willingness to pay computed from traditional models is overestimate while protest responses eliminated and underestimate while protest responses are treated as zero values. The aggregate total benefits of the related natural resources are all underestimated by the traditional estimation methods. The underestimated total values are ranged between 30.57% and 141.87% for sample of Kenting National Park and between 26.41% and 60.85% for sample of Changhua Coast Wetland. It is obvious that underestimation of the total values are expected while aggregated benefits of the natural resources are the concern.
|
|---|
