Locally-weighted meta-regression and benefit transfer

Meta-regression models (MRMs) are commonly used within benefit transfer to estimate will-ingness to pay for environmental quality improvements. In virtually all benefit transfers of this type, a single regression model is fit to all source points in the metadata, and used to produce out-of-sample predictions for all possible policy-site applications. Despite the advantages of this approach over other types of benefit transfer, the predictive accuracy of these MRMs generally leaves room for improvement. In this paper we propose a locally-weighted regression approach to MRM estimation to enhance the accuracy of benefit transfer predictions in an environmental valuation context. We introduce the concept of locally-weighted meta-regression, provide econometric underpinnings, and discuss the construction of weight functions. We illustrate the use of cross-validation to decide between weight functions, and show how this framework can be applied in an actual benefit transfer setting. For our empirical application on willingness-to-pay for water quality improvements, we find that the proposed approach brings substantial gains in predictive accuracy in a leave-one-out setting, and measurable improvements in predictive efficiency for benefit transfer.