Benefit Transfer of Non-Market Values – Understanding the concepts

1. Benefit Transfer of Non-Market Values – Understanding the concepts John Rolfe Central Queensland University

2. Outline of this talk • How does benefit transfer work? • Types of benefit transfer • Applications of benefit transfer • Improving the accuracy of transferred values

3. 1 - Benefit transfer • The transfer of values from one case study to another policy situation • Attractive because of cost and time advantages over the separate conduct of non-market valuation experiments • Can be complex because source and target sites may not be identical • Benefit transfer may involve some adjustment of values • BT may be associated with increased uncertainty about values

4. How a benefit transfer function works • An indirect conditional utility function: • Utility = Constant + attributes + individual factors • Vih =  + 1Z1 + 2Z2 + .... nZn + aS1 + bS2 + .... zSn • Vih = -0.828 - 0.023*cost + 0.031*Vegetation + 0.002*waterways - 0.015* people leaving + 0.149* reserve + 0.265*education + 0.0001*income • In this example, it would be possible to adjust this benefit transfer function to: • Different site characteristics (e.g. vegetation, waterways) • Different respondent characteristics (e.g. education, income)

5. Stages in BT process

6. 2. Key mechanisms for benefit transfer • Point – total value • Total value from a previous study • Point – marginal value • Value per unit transferred • Benefit function transfer • Function allows adjustments for site and population differences • Integrations across multiple studies • Meta analysis • Bayesian methods

7. Total point value transfer • Takes the lump sum value from a source study and applies it to the target • Simple in theory but does not allow …. • Variations in the quantity of the environmental impact • Variations in site differences • Variations in population differences • Variations in framing differences

8. Marginal point value transfer • Takes the marginal value from a source study and applies it to the target • Allows some adjustment according to the variation in the quantity of the environmental impact • May also allow some variation in the size of the population affected • Still easy to do but does not allow …. • Most variations in site differences • Variations in population differences • Variations in framing differences

9. Benefit function transfer • The transfer of a benefit function from a source study to a target application • Allows for adjustments in site and population differences • May still not account for differences in the way that tradeoffs are framed, or for temporal differences • Techniques such as Choice Modelling, which produce value functions, are suitable for this

10. Meta analysis • Meta-analysis for use in benefit transfer involves the summarizing of results for several existing source studies in a regression function, • This function is then used to predict value estimates for a target site • Often difficult to do in practice because of methodological and framing differences between studies

11. Bayesian transfer • Provides a mechanism for the results of several studies to be combined together with attitudinal data • Begins with a set of prior beliefs about the parameters of interest, by drawing on existing data and/or experience and beliefs from a range of stakeholders. • This prior can then be modified as new data (e.g. additional source studies) are incorporated. • Modified priors are normally referred to as ‘posterior’ beliefs. • Both priors and posteriors are usually presented in the form of distribution functions

12. 3 - Three main approaches to benefit transfer • ‘The Prospector’ – searches for suitable previous studies and transfers results across to target site • ‘The Systematic’ – designs a database of values suitable for benefit transfer • ‘The Bayesian’ – combines both a review of previous studies with potential data gathering

13. The ‘Prospector’ in practice • Once suitable source studies have been identified, then decision is taken about the best way of transferring values • Point – total value • Point – marginal value • Benefit function transfer • Meta analysis • Reflects most examples of benefit transfer

14. Examples of the Prospector • A number of studies conducted in the Fitzroy dealing with water allocation and riparian development issues • Results have been transferred to other policy issues dealing with vegetation, water development, protection of cultural heritage

16. Values for vegetation and waterways over time

17. Examples of the Systematic • Van Beuren and Bennett (2004) explored values for NRM protection across Australia • Morrison and Bennett (2004) explored values for protecting river systems across NSW • Windle and Rolfe (2007) developed a broad data base of NRM values in Qld relating to soils, vegetation and waterways

18. The Windle and Rolfe database • Identified the values for improvements in 3 key areas of the investment plans for NRM regional groups • Healthy vegetation • Healthy waterways • Healthy soils • Identified sensitivity to regional issues • Identified sensitivity to framing issues

19. Regional choice set example

20. Summary of values

21. Issues with the systematic approach • Usually designed to be very broadscale and inclusive so as to cover large range of different potential applications • Not always easy to transfer values from a very general application back to the specific • Normally needs to be some adjustment involved for the scope and scale changes • But it is not clear what the theoretical basis for the adjustment factor is and how they should be calculated

22. The Bayesian approach • Provides a mechanism to combine data from a variety of sources and update it with attitudinal data from target population • Complex and difficult to perform • Not always clear how to weight the contribution of different data sources

23. 4 - How to make Benefit transfer more accurate • There are four main strategies • Increase the pool of non-market valuation studies • Increase the accuracy and understanding of the conducted studies • Develop better systematic BT case studies • Improve the use of BT tools and databases

24. Increasing the pool of studies • Very limited pool of studies available to source BT values • Limits the potential use of BT • Means many BT applications involve substantial differences in site and populations, which may lead to inaccuracies • Makes it difficult to identify patterns of values, and identify source studies that may be outliers

25. Increasing the accuracy • Currently there are high levels of uncertainty around many non-market valuation estimates • Unclear how differences in framing and methodology affect outcomes • Unclear how variations in scope and scale can affect value estimates • Unclear how values are set when elements of risk and uncertainty are present

26. Dealing with the ‘specific to general’ tradeoffs • A benefit transfer application will rarely satisfy ‘ideal’ conditions • Identical site characteristics • Identical population characteristics • Identical policy and tradeoff situations • Better to think of a BT application as a filtering mechanism • Identify if there are major differences between benefits and costs, or • Identify if more detailed analysis needs to be applied.

27. More systematic approaches • It is cost-effective to develop data-bases of values rather than a large number of ad-hoc valuations • Critical that future attempts to develop systematic databases address the issue of adjustment factors • Required to transfer general values to specific situations with scale and scope differences

28. Develop better BT tools and databases • Better development of BT guides • More examples and applications of meta studies and bayesian approaches • More use of BT databases • Environmental Valuation Reference Inventory in Canada (http://www.evri.ca/) • Older ENVALUE site in NSW

29. 5 - Conclusions • Benefit transfer applications likely to rise • Increased demand for valuations coupled with time and cost constraints • Increased supply of source studies, whether on an ‘ad hoc’ or systematic basis • Caution is needed to make BT accurate • Work needed to improve the potential for BT in a number of key areas