4. Benefits transfer
Estimating the so-called "non-market costs and benefits" described in equation (1) using primary research methods such as contingent valuation, the travel cost method or choice experiments is time-consuming and expensive, and cannot be justified for a scoping study of this type. In such instances, existing knowledge from research at other sites can be used in a process known as "benefit transfer". Benefit transfer entails collection of value estimates from other locations (study sites), being careful to match environments, users and proposed changes as closely as possible. The best matching results are then used as estimates of value at the site under investigation (the policy site). Alternatively, results at study sites can be averaged, or they can be adjusted to reflect better the situation at the policy site.
The three principal methods of transferring benefits from a study site or sites to a policy site (in this case, the Waitaki River) are:
- Direct transfer. In direct transfer, mean values estimated at the study site, or several study sites, are used directly at the policy site, without adjustment to reflect policy site characteristics. For example, it is conceivable that a point estimate of "existence value" benefits associated with habitat preservation in the Waitaki Catchment could be obtained by using the mean, median or range) per capita dollar value from a study undertaken elsewhere in New Zealand and multiplying this value by the number of households affected by Waitaki water allocation. As expected, extremely strong assumptions need to be invoked viz the change in environmental conditions are identical as are population characteristics (preference structure, ethnicity, age, household size, and so on).
- Benefit function transfer. This approach uses the valuation function estimated at the study site as a basis for estimating values at the policy site using policy site parameters (ie, those pertinent to the Waitaki). For example, a study may have estimated a value function for household i, as WTPi = a + bX + cY + e; where X = site/good characteristics, Y = respondent characteristics (age, income, ethnicity, etc) and e = statistical error. The estimate of value is obtained by plugging site specific values into the function.
- Meta-analysis. Meta-analysis is another form of valuation function benefit transfer. Regression analysis is applied to the results of other valuation studies completed at many sites to identify statistically the influences of site attributes on value. Meta-analysis has an advantage in that it uses information from a number of studies.
The extensive literature on benefit transfer is in general agreement that the approach can produce large errors - transferred benefits are frequently in disagreement with primary valuation studies carried out at the policy site (Brouwer, 2000; Brouwer & Spaninks, 1999; Vandenberg et al, 2001). Meta-analysis, which uses results from a large number of studies to identify the role of site attributes, is acknowledged as the most accurate benefit transfer approach. Benefit function transfer and point estimate transfer are less reliable, with benefit function transfer typically preferred to point transfer (Brouwer & Spaninks, 1999; Rosenberger & Loomis, 2003; Vandenberg et al, 2001). Typically large errors associated with benefit estimates at study sites warn of the need for caution when transferring simple point estimates. Despite lack of precision, benefit transfer is the only available indicator of non-market values in the absence of a site-specific study. It is an approach that is generally accepted as providing order of magnitude estimates of values that indicate whether further, site-specific valuation work is warranted.
