Are Recreation Values Systematically Underestimated? Reducing Publication Selection Bias for Benefit Transfer

This paper shows how reported recreation values can be systematically underestimated when they are derived from price coefficients. Simulations show that these publication selection biases can be very large. Because selected price coefficients are transformed into estimates of value, conventional meta-analytic methods used to deal with publication selection will often make the bias worse. Simulations also show that an alternative, meta-regression estimator, Root-n MRA, can greatly reduce this potential bias and has lower MSE than alternative meta-analytic methods. This method uses the square root of a study’s sample size as a proxy for the standard error of welfare measures, thus avoiding simultaneity bias associated with welfare measures and standard errors of price coefficients. Methods for detecting publication selection are illustrated by applying alternative estimators to the outdoor recreation valuation literature, in general, and freshwater fishing, in particular.