Using Observations of Marginal Willingness to Pay in Willingness-to-Pay Meta-Regressions

Abstract

Benefits transfer using meta-regression is one of the most widely applied approaches for evaluating changes in willingness to pay and welfare under water-related policies. In practice, however, analysts face a classic trade-off between the consistency of value observations and the number of observations available for estimation. Restricting observations to be more comparable—in terms of both the type of welfare measure and the nature of the environmental change being valued—typically results in a smaller usable sample from the literature. Here, we examine the performance of a middle-ground approach in which all value measures are restricted to be Hicksian welfare measures derived from stated preference studies, while both marginal willingness to pay (MWTP) and non-marginal willingness to pay (WTP) estimates are incorporated into a single meta-regression framework. Overall, we find strong evidence supporting the joint use of WTP and MWTP observations, though results depend critically on how the two measures are combined. Ensuring that WTP and MWTP are integrated in a theoretically consistent manner, rather than either pooling them without adjustment or controlling for their difference using a simple dummy variable, yields superior performance according to both within-sample fit and out-of-sample predictive accuracy. Our findings suggest that our approach can enhance the efficiency of meta-data collection and improve the reliability of benefits transfer applications in water management and policy evaluation.