Improving the Practice of Benefits Transfer: A Preference Calibration Approach

Abstract

Most applied welfare analyses for environmental policy evaluations, whether benefit-cost or natural resource damage assessments, rely on daptations of existing benefit estimates in what is described as benefits transfer rather than new research. Almost 10 years ago, David Brookshire organized a set of papers in Water Resources Research to focus attention on the practice of benefit transfer (see Brookshire and Neill [1992]). Since then, interest in research on the potential for improvement in these techniques has exploded and this volume reports on a number of innovations relying on refinements in the statistical methods used in meta analyses that often provide empirical benefit functions for transfer. Nonetheless, where evaluations of benefits transfer have taken place, current practice is generally regarded as very unreliable. This paper considers a different perspective on the practice of benefit transfers. It is one that interprets the benefit transfer problem as an identification problem. That is, the analyst must calibrate individual preferences for the environmental resources of interest based on the available empirical benefit estimates. Our proposed methodology is general. Here we apply it to oneexample – the development of consistent measures of the benefits of water quality improvements. To develop this logic we begin with a historical perspective, interpreting Harberger’s [1971] approximation using indifference curves and then suggesting this logic seems to have been a conceptual antecedent to the logic used with unit benefit transfers. However, in this case thesame desirable properties cannot be assured. Section 2 presents a detailed description ofconventional benefit transfer practices and these antecedents. Section 3 illustrates the limitations of such practices through a simple example. In Section 4, we provide a detailed description of ourproposed methodology in six steps. Using a case study from the Chesapeake Bay our proposedapproach is illustrated in Section 5 with travel cost and contingent valuation (CV) data. We thendemonstrate how the calibrated functions can be used to construct benefit estimates for a separatesituation. We discuss how the resulting benefit estimates differ from those of a more traditionalbenefit transfer practice, hereafter labeled “simple approximation.” Finally, in Section 6 we present a few methodological conclusions.