Key Messages 
  • Environmental goods and services are almost always difficult to quantify in economic or monetary terms. Benefit transfer is a method which can prove useful when a primary valuation study may be unfeasible.
  • There are three approaches to benefit transfer: unit benefit transfer, value function transfer and meta-analytic function transfer.
  • There can numerous limitations to using benefit transfer in adaptation, including: data and methodological heterogeneity, correlation issues from the use of one primary study and the generalisation error associated with any benefit transfer.

Context

Where no previous economic information is available on the value of an environmental good or service, there are several methods that could be used to measure it in monetary terms. These methods can be classified into three main categories. The first group includes market-based approaches that reflect the real preferences or costs (or benefits) to individuals. The second group consists of revealed preference approaches that are based on the observation of individual choices through which people show their preferences in relation to the good or service under valuation. Finally, the stated preference approach is used to simulate a market by using surveys on hypothetical changes in the provision of goods or services caused by policy changes.

However, resources or time constraints may limit or prevent new primary valuation studies from being undertaken; instead, a benefit transfer (BT) method could be used. A benefit transfer consists of taking an estimate from previous research (i.e. the value provided by salt marsh ecosystems in a certain location) and transferring it to value an analogous ecosystem. The site from which values are taken is known as “study-site” and the place where values are being transferred to is called “policy-site”.

Policy and methodological developments 

Use of benefits transfer has increased over time, especially in publicly-funded projects. The governments of Canada, the United States and the United Kingdom have developed a joint database of primary valuation studies. This database (found at www.evri.ca) compiles original studies for use in benefits transfer applications around the world. While database approaches to calculating benefits transfers increase the ease and applicability of benefits transfer in project valuation, experts caution against applying values from other studies without careful analysis of any necessary adjustments required for the policy site (OECD, 2006).

Brookshire (1992) notes that the level of accuracy required from a benefits transfer depends on the intended use of the data, with compensatory damages requiring the highest confidence in values, followed closely by policy decisions. Less binding uses, such as early screening or scoping of policy alternatives may remain useful even with high transfer errors.

There are three ways in which a benefit transfer can be developed: unit value transfer, value function transfert and meta-analysis function transfer.

Unit value transfer

The more basic approach, called unit BT, consists of the assumption that the single value of an non-market good in the study-site is approximately equal to that in the policy-site. This value is thus directly transferred, making adjustments when necessary, usually through currency or income (e.g. if the policy site has an average income equivalent to 45% of the policy site’s income, values can be reduced to that level, given the income elasticity of the valued good or service).

Overall, the unit transfer approach requires suspension of the possibility that preferences in two different sites may vary. Direct transfers neglect site-specific information such as: socio-economic and demographic characteristics, physical and environmental attributes of a site and differences in ‘market conditions’ (such as availability of substitutes). When adjusting for income, enough data must be available to determine income elasticity for the good being valued.

Value function transfer

The value function transfer converts a relationship for the same good in two different sites by transferring a value function responsive to specific characteristics at each site. From a conceptual point of view this is a more rigorous procedure as more information is used for the value transfer. In this way the valuation function used as the study-site is applied at the policy-site by introducing information and parameters from the area under study. Navrud and Ready propose the following value function for calculating willingness to pay:

where Gj are the characteristics of the environmental good of interest at the policy site and Hi are the characteristics of households at the study site. The function attempts to delineate the value for environmental goods given specific household characteristics and calculate this value for a different set of households. To avoid statistical errors from transferring values from a single site, practitioners often rely on meta-analyses to yield significant results for G values in the value function. OECD (2006) suggest the following stylised function:

Meta-analysis function transfer

Meta-analysis is “the statistical analysis of a large collection of analysis results from individual studies for the purpose of integrating the findings” (Glass, 1976: 3). The difference with the previous approach relies on the fact that this function is built based on multiple values from different studies. That is, the value for the study-site estimated with this approach is not obtained from one single study but from a compilation of values obtained from a meta-analysis (Galarraga et al., 2004; TEEB, 2010).

Many authors advocate that meta-analysis (MA) transfer functions represent a more robust approach to benefit transfer than using alternative transfer methods (see, for example, Moeltner et al., 2007; Stapler and Johnston, 2008; Johnston and Rosenberger, 2009; Brander et al., 2012). Despite MA outperforming unit and value function transfers, Rosenberg and Stanley (2006: 377) remind us that the final quality of benefit transfers depends on the data on which they are built. Furthermore, Nelson and Kennedy (2009: 370) state that “it is easy to do a meta-analysis, but it is difficult to do a good one”.

Bergstrom and Taylor (2006) caution against the widespread application of meta-analysis approaches to benefits transfer, calling for improved methodological checks for this approach.  In order to strengthen meta-analysis benefits transfer, they stress the importance of ‘commodity consistency’ of environment goods included in studies that form the meta-analysis. This can be proxied by substituting the services provided by an environmental good for the good itself. Limiting meta-analyses to sites that provide similar services from environmental goods improves the validity of the values for transfer, but may reduce the pool of available data for a desired benefits transfer.

If commodity consistency cannot be achieved, heterogeneity between commodities must be controlled for in the value function (i.e., if various-sized wetlands are included in a meta-analysis, the size of wetland can be included as a controlling factor in a value function).

Additionally, Bergstrom and Taylor argue that any meta-analysis must satisfy ‘welfare change measure consistency.’ For this condition to hold, the methods of estimating value for a given service from an environmental good must be equivalent. Either the same method (i.e., travel cost method, contingent value method, hedonic pricing, etc.) must be applied in each study, or an adjustment reflecting the theoretical differences in valuation methods must be included in calculations for the meta-analysis benefits transfer.

A major issue with using meta-analysis for benefit transfer relates to correlation, or the interdependency of multiple records that are based on the same primary study. The simplest way to avoid correlation is to use a single estimate per primary study; however this produces an extremely small sample. Other adjustments, like econometric methods, can be used to adjust for dependency.

Transfer errors in benefit transfer

The ‘transfer error’ is a value calculated by comparing benefits transfer from a study site to real data measured in a policy site. This measure can be calculated, where data is available, with the following formula:

Challenges related to value transfer include: a paucity of quality existing studies to transfer from, a mismatch in a change in a good from a proposed policy and the observed change in an existing study, differences in study sites and policy sites unaccounted for in the transfer formula, an incomplete or overgenerous determination of the market at a policy site and potential bias from aggregating individual components of value for a good. Use values—preferences measured by real expenditures such as fee payments, travel costs, lost wages, etc.—are likely to be more accurate than non-use values, which can suffer from a hypothetical bias.

Rosenberger and Stanley (2006) define three sources of transfer error present in benefits transfer applications: measurement, generalisation and publication, as outlined in Table 1. Measurement errors can occur from the inclusion of biases in collecting primary information for an initial test site. This type of error may result from the valuation method used, flawed samples or any other problems with initial collection methodology. Generalisation errors result from applying values from one site to another without appropriately correcting for differences between sites. Finally, a publication selection bias results from careful selection of which studies are published, excluding those that may not support theory or that lack statistical significance.

Table 1. Types of error in value transfers

                                                    Transfer Error Elements                   

 

Measurement Error

Generalisation Error

Publication Selection Bias

Includes

Valuation errors, sampling errors, excluded populations

Differences from income, preferences, types of environmental goods

Favouring studies that support existing theory, only publishing statistically significant findings

Strategies for reduction

Careful data collection, uniform preference measurement methods (Contingent Valuation, hedonic pricing, etc.) 

Adjust for differences between sites, choose similar sites and environmental services

Pull studies from open-access/grey literature inventories such as the Environmental Valuation Resource Inventory

Spash and Vatn (2006) illustrate the various dimensions that must be accounted for in any value transfer. Figure 1 demonstrates the various factors in a value function that can differ between sites. When comparing two sites, differences between natural science and social groupings shown in Figure 1 should be accounted for in a value function. Parallels in economic preferences can be established through previously published studies, if available, but valuation methods should be conducted with care as to reduce any error in the study site, as discussed below. Site-specific risk preferences may be adjusted for under ‘Attitudes, norms and beliefs’ in this framework.

Figure 1. Site-specific factors in benefits transfer

Source: Spash and Vatn, 2006.

 

Main implications and recommendations 

There are some general challenges and limitations of using benefits transfer in economic appraisals. Firstly, primary data heterogeneity refers to the general lack of uniformity in the data on costs and benefits of adaptation options. This can make it particularly difficult to perform a benefit transfer. Other heterogeneity issues were noted related to commodity consistency and methodologies used. Some suggested solutions to these issues include calibration, adjusting the dependant variable, including new explanatory variables or dividing the sample into smaller and more homogeneous sub-samples, however, these approaches are not without limitations.

There is also the issue known as the generalisation error, which can be one of the main sources of error in benefit transfers. It arises from the adaptation of estimates from study sites to policy sites – an important component of benefit transfers. While meta-analysis transfer functions tend to be a preferred approach to minimise this issue, it seems that this improved performance stems from the similarity between sites, and not the function itself.

A number of good practice can be followed to minimise transfer error. Goods at a study site must be similar to the policy site values are transferred from in the definition of the good, the population affected and the degree of change in the stock of the good by a given policy. To combat omitted variable bias, maximum information be controlled for in a value transfer function. Best practices for reducing transfer errors include defining a level of acceptable error before calculating variance between transferred and real values and performing preference calibrations between sites.

While additional research will continue to improve accuracy and understanding around benefits transfer, some level of transfer error must be accepted for current policy analysis. Ultimately, transfer errors represent the acceptable level of risk a decision maker is willing to incur to avoid the costs of conducting full studies at policy sites.

Bibliography 

Bergstrom, J. C., Taylor, L.O. (2006), Using meta-analysis for benefits transfer. Ecological Economics , 60, 351-360.

Brander, L.M., Bräuer, I., Gerdes, H., Ghermandi, A., Kuik, O., Markandya, A., Navrud, S., Nunes, P.A.L.D., Schaafsma, M., Vos, H. (2012), Using meta-analysis and GIS for value transfer and scaling up: Valuing climate change induced losses of European wetlands. Environ. Resour. Econ. 52, 1–19.

Brookshire, D. S. (1992), Issues regarding benefits transfer. Association of Environmental and Resource Economists Workshop (pp. 1-13). Snowbird, Utah: AERE.

Galarraga, I., Beristain Etxabe, I., Martín Landa, I., Boto Bastegieta, A. (2004), El método de transferencia de valor (benefit transfer), una segunda opción para la evaluación de impactos económicos: el caso del Prestige. Ekonomiaz 57, 30–45.

Glass, G.V. (1976), Primary, Secondary, and Meta-Analysis of Research. Educ. Res. 5, 3–8, http://dx.doi.org/10.2307/1174772.

Johnston, R.J., Rosenberger, R.S. (2009), Methods, trends and controversies in contemporary benefit transfer. J. Econ. Surv. http://dx.doi.org/10.1111/j.1467-6419.2009.00592.x.

Kirshen, P., Knee, K., Ruth, M. (2008), Climate change and coastal flooding in Metro Boston: impacts and adaptation strategies. Climatic Change, 90(4), 453–473, http://dx.doi.org/10.1007/s10584-008-9398-9.

Moeltner, K., Boyle, K.J., Paterson, R.W. (2007), Meta-analysis and benefit transfer for resource valuation-addressing classical challenges with Bayesian modeling. J. Environ. Econ. Manag. 53, 250–269, http://dx.doi.org/10.1016/j.jeem.2006.08.004.

Navrud, S., Ready, R. (2006), Review of methods for value transfer. In S. Navrud, & R. Ready, Environmental value transfer: issues and methods (pp. 1-25). The Netherlands: Springer.

OECD (2006), Benefits transfer. In OECD, Cost-benefit analysis and the environment: recent developments. Paris: OECD Publishing.

Rosenberger, R.S., Stanley, T.D. (2006), Measurement, generalization, and publication: Sources of error in benefit transfers and their management. Ecol. Econ., Environmental Benefits Transfer: Methods, Applications and New Directions Benefits Transfer S.I. 60, 372–378, http://dx.doi.org/10.1016/j.ecolecon.2006.03.018.

Spash, C. L., Vatn, A. (2006), Transferring environmental value estimates: issues and alternatives. Ecological Economics, 60, 379-388.

Stapler, R.W., Johnston, R.J. (2008), Meta-Analysis, Benefit Transfer, and Methodological Covariates: Implications for Transfer Error. Environ. Resour. Econ. 42, 227–246, http://dx.doi.org/10.1007/s10640-008-9230-z.

TEEB (The Economics of Ecosystems & Biodiversity) (2010), The Economics of Ecosystems and Biodiversity for Local and Regional Policy Makers, Progress Press, Malta.

Tol R. S. J. (2007), The double trade-off between adaptation and mitigation for sea level rise: an application of FUND. Mitigation and Adaptation Strategies for Global Change 12(5): 741-753.