The Decision to Use Benefit Transfer or Conduct Original Valuation Research for Benefit-Cost and Policy Analysis
Allen, Bryon P., Loomis, John B., Contemporary Economic Policy
Federal agencies are often required by statute or executive order to examine the costs and benefits of their decisions (Executive Order #12866, 1993; U.S. Environmental Protection Agency, 1998; U.S. Water Resources Council, 1983). Quantifying the benefits of goods and services that are traded in markets can be relatively straightforward. Doing so where markets do not exist is more difficult. State and federal agencies use a variety of tools to estimate willingness to pay for nonmarket goods or services. Executive Order #12866 requires a thorough regulatory impact analyses for all decisions involving economic effects of more than $100 million/yr. Natural resource damage assessments often evaluate multimillion dollar cleanup efforts. Economists analyzing these and smaller projects are often asked to evaluate the benefits of proposed regulations using only existing literature and benefits estimates from previous analyses (USDA Forest Service, 1989; U.S. Environmental Protection Agency, 2000, 2002). Resources (time and dollars) are seldom allocated to conducting original, regulation-specific, benefit estimation studies. In this article, we ask if society is best served to make multimillion dollar--and often multi-hundred million dollar decisions without commensurate data or case-specific original studies.
Generally, nonmarket values are estimated through original research using methods such as the contingent valuation method (Mitchell and Carson, 1989) or the travel cost method (Ward and Beal, 2000). The increasingly rich literature of valuation studies, combined with theoretical innovations, created the opportunity for less expensive and less time-consuming approaches to value estimation--a variety of techniques that can be described as benefit transfer (Bergstrom and De Civita, 1999; Brookshire and Neill, 1992; Loomis, 1992; McConnell, 1992).
Two general approaches have been adopted for estimating benefit transfer values. The first, and simplest, involves taking a single "point" estimate of value from a similar existing study and adopting it as the transferred value for the new policy analysis. A slightly more complicated version of this approach involves averaging the estimates from several similar previous studies and using that average value in the new policy analysis (for more examples of both of these approaches and further discussion of them, see Rosenberger and Loomis, 2001). Both of these approaches that involve transferring a best point estimate or average value can be thought of as benefit value transfer techniques.
The second and more sophisticated approach to benefit transfer involves estimating a meta-analysis function from a database of original estimates of the class of goods or services under investigation. When an estimate is needed for a new situation, the variables in the meta-function are set to the appropriate values and an estimate of willingness to pay is produced. Such meta-analysis functions exist for outdoor recreation (Rosenberger and Loomis, 2001; Smith and Kaoru, 1990; Walsh, Johnson, and McKean, 1989), groundwater valuation (Boyle, Poe, and Bergstrom, 1994), endangered species valuation (Loomis and White, 1996), air pollution and visibility (Smith and Osborne, 1996), and the health effects of air pollution (Desvousges, Johnson, and Banzhaf, 1998).
A number of studies have performed statistical convergent validity tests on benefit transfer values created using both value and function transfer approaches (Downing and Ozuna, 1996; Kirchhoff, 1998; Kirchhoff, Colby, and LaFrance, 1997; Loomis, 1992; Loomis et al., 1995; Smith and Huang, 1995). The results of these studies have been generally mixed, with many suggesting that benefit transfer estimates are significantly different from the original estimates in question. While these studies provide interesting information regarding the statistical validity of benefit transfer, they do not tell us anything about the opportunity costs of poor policy decisions from conducting benefit transfer instead of original studies, that is, avoiding errors in falsely adopting an uneconomical policy versus failing to adopt an economical policy. …