Meta-analysis and benefit transfer for resource valuation-addressing classical challenges with Bayesian modeling

The use of meta-regression models based on existing studies to estimate the value of resources at a new policy site has become a popular alternative to collecting original data in recent years. There are two prevalent dilemmas associated with classical meta-regression models: The difference in the available set of regressors across source studies and the treatment of methodological explanatory variables in the construction of benefit transfer functions. In this study we illustrate how these issues can be addressed efficiently within a Bayesian meta-regression framework. We find that the Bayesian model, in contrast to its classical counterpart, can estimate a relatively large set of parameters, including indicators of unobserved study heterogeneity, with reasonable accuracy even when the underlying meta-sample is small. The incorporation of information from regressor-deficient source data in the specification of Bayesian priors leads to a better model fit and tighter welfare estimates for Benefit Transfer in our application of freshwater angling.