Predicting Resource Policy Outcomes via Meta-Regression: Data Space, Model Space, and the Quest for 'Optimal Scope'

Resource-managing agencies are increasingly relying on secondary data to predict economic benefits for planned policy interventions. This `transfer of benefits' is often based on a quantitative synthesis of aggregate results for similar past interventions via Meta-Regression Models. However, this approach is generally plagued by the paucity of available studies and related small sample problems. A broadening of scope of the Meta-Regression Model by adding data from ``related, yet different" contexts or activities may circumvent these issues, but may not necessarily enhance the efficiency of transfer functions if the different contexts do not share policy-relevant parameters. We illustrate how different combinations of contexts can be interpreted as `data spaces' which can then be explored for the most promising transfer function using Bayesian Model Search techniques. Our results indicate that model-averaged benefit predictions for scope-augmented data spaces can be more robust and efficient than those flowing from the baseline context and data.