Suggested Bibliographic Reference: NAAFE Forum 2017 Proceedings, March 22-24, 2017. Compiled by Ann L. Shriver with assistance from Stefani Evers. North American Association of Fisheries Economists (NAAFE), Corvallis, 2017.
Proceedings of the 2017 Forum of the North American Association of Fisheries Economists, held March 22-24, 2017 in La Paz, Baja California Sur, Mexico.
Meta-analysis is one of the most common tools in benefit transfer. The typical way to gauge the reliability of a meta-analytic benefit transfer is to use convergent validity, which consists of cross-validation exercises. The analyst estimates a meta regression model while omitting one or more observations from a meta dataset, then uses the estimated transfer model to predict the value of the omitted observation(s). These two objects (the true value and the predicted value) are then compared, producing a prediction error. The mean of the distribution of absolute percentage prediction errors is the standard measure of reliability of the benefit transfer, but discards useful information generated by the cross-validation exercise. In particular, using point estimates to conduct benefit transfer ignores that the estimated value carries substantial uncertainty. Further, only using the mean of the transfer error distribution when assessing reliability ignores information that could potentially be useful to decision makers. The present paper proposes interval transfers as opposed to point estimate transfer for conducting benefit transfers. Interval-based estimates contain more information regarding the true value of an environmental amenity. Further, a new approach to policy making under uncertainty is proposed that can be used within the framework of meta regression models. We showcase these methods in the context of benefit transfer of the recreational value of coral reef ecosystems, an area of growing international interest that has thus far received little attention.