A Systematic Database for Benefit Transfer of NRM Values in Queensland

Introduction

Policy makers often need to assess the community demands for environmental or social improvements to evaluate particular funding, policy or development proposals. Key examples of benefit requirements include investment proposals by Regional Natural Resource Management Groups, where some estimates of the potential benefits generated is needed to justify the investment.

Typically the evaluation is performed by comparing the desirable outcomes (benefits) with the costs involved, often with the aid of cost benefit analysis. A key task is to assess values for environmental and social benefits involved so that they can be compared to the costs. A range of non-market valuation techniques can be used for this purpose.

Benefit transfer (BT) is a pragmatic way of estimating values for environmental or social tradeoffs when there is limited time or funding available. It operates by transferring values in some way from existing valuation studies to a target study of interest. Benefit transfer can be used at a range of levels and accuracies, from technical analysis through to broad policy overviews. An overview of the benefit transfer approach is available here.

The availability of source studies for a benefit transfer exercise is often limited. An alternative source is the development of a systematic database designed specifically to underpin benefit transfer in a particular field. This document details the development of a systematic database for benefit transfer of natural resource management values in Queensland. Details on the background to the study are available here.

The database of benefit transfer values were developed for potential improvements in the condition of soils, waterways and vegetation in different regions of Queensland. This database is freely available for access and use. Using the database can involve several steps and adjustment factors. A practical example is provided to demonstrate the potential use of the database to evaluate investment proposals.

An Overview of Benefit Transfer

Benefit transfer (BT) is a pragmatic way of estimating values for environmental or social tradeoffs when there is limited time or funding available. It operates by transferring values in some way from existing valuation studies to a target study of interest. The concept of benefit transfer is routinely applied in the commercial world, where, for example, prices of real estate from sales data are routinely extrapolated to estimate market values for other real estate areas. However, the term ‘benefit transfer' is normally used to identify the transfer of non-market values from source studies to a target site.

Benefit transfer can be applied to extrapolate values of different non-market valuation techniques, including those involving:

Hedonic pricing,
Travel cost method,
Contingent valuation,
Choice modelling.

The types of benefit transfer applications can range from simple and pragmatic applications through to more technical and complex operations. Examples of applications include:

Broadscale policy issues where benefit transfer is used to give some indication of values involved,
Policy advice issues, where benefit transfer might be used to assess values for particular policy outcomes,
Specific case study or cost-benefit evaluations, where benefit transfer is used to provide technical inputs.

There are four main ways of performing a benefit transfer process for this purpose, as shown in the following table.

Table 1. Types of transfer methods

Transfer method	Description	Example
Single point value transfer	A single value is transferred without adjustment from source study to target site	A forest protection value of $50/person is transferred from Case Study A to Site B
Marginal point value transfer	A single value that allows for site differences is transferred	A forest protection value of $2/hectare/person is transferred from Case Study A to Site B. The values are adjusted for the size of the area.
Benefit function transfer	A valuation function is transferred, allowing adjustment for variety of site differences	A forest valuation function that involves several attributes is transferred Case Study A to Site B
Meta value analysis	Results of several studies are combined to generate a pooled model	Results from studies A, X, Y and Z are pooled to estimate a value for Site B

Transfer processes can be complex to avoid potential sources of error in the extrapolation of values to sites or issues of interest. Benefit transfer is uncontroversial when:

the sites are almost identical,
the same population is involved,
the extent of the change being considered is the same in both source and target cases,
the ‘frame' of the source study matches the ‘frame' of the target study, and
the initial valuation study has been performed rigorously and accurately.

However, most applications of benefit transfer involve greater differences between site and target studies, indicating that more analysis, adjustment and testing of the transfer process needs to take place.

Table 2. Stages in a benefit transfer exercise

#	Stage	Notes
1	Assess target situation
2	Identify source studies available and select benefit transfer type	Transfer type largely dependent on source studies available
3	Assess site differences	identify if BT possible identify basis for BT adjustment
4	Assess population differences	identify if BT possible identify basis for BT adjustment
5	Assess scale of change in both cases	identify if BT possible identify basis for BT adjustment
6	Assess framing issues (Differences between the issues involved and the ways the tradeoffs were framed in the source study)	test if source study is appropriate for BT identify any basis for BT adjustment
7	Assess statistical modelling issues	identify appropriateness of model in source study identify any basis for BT adjustment
8	Perform benefit transfer process

1. Assessing the Target Situation
This stage involves the assessment of the site characteristics, the policy change or development being considered, and the population likely to be affected. The assessment is used to identify the key issues and attributes of interest, and to provide a template for comparative assessment in the following stages.

2. Identifying Source Studies
There are two main ways of sourcing studies for a benefit transfer exercise. The first (and most common) approach has been to ‘forage' for relevant primary studies in the available literature. Several data bases have to been developed to assist in the sourcing of suitable case studies. Some of the most useful are the ENVALUE site in Australia (www.epa.nsw.gov.au/envalue), and the Environmental Valuation Reference Inventory in Canada (http://www.evri.ca/).

The criteria for selecting suitable source studies might include factors such as:

Similarities between source and target sites, populations and scale of changes involved,
The type of source study available,
The age of the source study,
The type of statistical modelling performed in a source study, and
The rigour of the source study (including sample size, validity tests performed and the strength of statistical relationships observed).

The second main way of performing a benefit transfer exercise is to develop a database explicitly for the purpose of providing values into benefit transfer applications. This systematic approach provides a database of estimated values or value functions that can then be accessed by analysts searching for source values. The Queensland NRM values database provides an example of this type of approach.

3. Assessing Site Differences
Key characteristics of a study site might include:

the physical characteristics,
the type of policy or development changes being considered,
the types of impacts being generated, including physical, environmental, social and economic impacts, and
the size of the changes involved.

The pattern of site differences between source and target sites can be grouped into four main categories, as shown in the table below.

Table 3. Types of site differences

#	Extent of difference	Application	Comment
1	Little site difference exists	All transfer methods are suitable	Benefit transfer function is likely to be more accurate where sites are very similar
2	Key attributes are the same, but levels vary between sites	Single value point transfer not suitable; others are appropriate	Marginal values or benefit functions need to be adjusted for levels at target site.
3	Target site does not have all the key attributes of source site	Single value point transfer not suitable; others may be appropriate	Levels for unnecessary attributes may be set at 0 in benefit transfer function
4	Target site has extra key attributes over source site	None of the transfer methods may be fully appropriate	May be handled by a series of marginal point value transfers

4. Assessing Population Differences
Different populations may hold different preferences for the benefits involved in a case study situation. This is particularly the case where use or indirect use values are involved, and different populations have different levels of involvement with, or use of a particular resource. Different populations may also hold different non-use values, particularly where the populations have very different characteristics and/or attitudes.

The pattern of possible population differences between source and target sites can be grouped into three main categories, as shown in the table below.

Table 4. Types of population differences

#	Extent of difference	Application	Comment
1	None - same population for source and target sites	All transfer methods are suitable
2	Some - Substantial overlaps between populations for source and target sites	All transfer methods are suitable	Benefit transfer function is preferred because it allows adjustment for population differences
3	Large - Different populations for source and target sites	Benefit transfer not always appropriate	Where different populations have similar preference structures, then benefit transfer still appropriate

5. Assess Framing Issues
Framing issues in stated preference studies can be categorised as framing differences and framing problems. Framing differences occur when the tradeoffs in stated preference studies are presented in different ways, while framing problems occur when the respondent to a survey is unduly sensitive to the context in which a particular tradeoff is offered.

There are a large number of framing issues which can affect a valuation process. These are important in a benefit transfer process because:

Framing problems in a source study can make it invalid for benefit transfer;
Differences in framing between source and target sites might invalidate benefit transfer; and
Differences in framing between source and target sites may generate requirements for benefit adjustment.

There are a number of areas where framing issues can be identified, including:

Willingness to pay versus willingness to accept formats;
The scope and scale of the issue presented;
The policy mechanisms involved;
The payment vehicle used;
The survey instrument and collection method used; and
Context factors such as wealth conditions, the political debate, and other factors.

6. Assess Statistical Modelling Issues
There are a number of different ways of modelling data from various case studies using different non-market valaution techniques. Caution has to be taken that the context in which values are transferred between source and target sites matches the way that the results have been modelled. For example, it may be inappropriate to transfer a part-worth value estimated for a particular group under a nested model to a general population.

Background: Benefit Transfer Database for NRM values in Queensland

In 2005 a non-market valuation study was conducted in Queensland to determine community values (public benefits) for improved NRM outcomes in different Queensland catchments. The project was funded by the Australian and State Governments under the AGSIP program. The study and subsequent database of values were developed by Dr Jill Windle and Professor John Rolfe at CQUniversity.

The detailed results of the study are available under the second of the AgSIP non market valuation research reports. The study was designed to compile a systematic set of indicative values that could then be applied directly in some regions as well as used for benefit transfer in other areas. The values developed in the data base should have some level of applicability in all areas of Queensland.

The benefit transfer database is focused on the use of marginal values, with some variations between regional and state populations. More specialized applications would be possible that involved either:

application of benefit transfer functions
other case studies that could be sourced for primary values,
a meta-analysis that involved an overview of several studies,
a dedicated valuation study to collect data for the case study of interest.

Database of Benefit Transfer Values for NRM outcomes in Queensland

To develop the systematic database, values were calculated for improvements in the condition of:

soil,
waterways,
vegetation.

Because values can vary across regions and populations, values have been derived at both:

individual regional levels
an average across regions

The database shows the expected value to the average household of a 1% improvement in the condition of each broad asset.

Table 5. Marginal values (per household) for 1% improvement in NRM outcomes

	Soils in good condition	Waterways in good health	Vegetation in good health
Brisbane households - values for S.E. Queensland Area = 23,000sq km; river length:2,000km Sample response rate = 50%
Range	45%-60%	35%-55%	25%-45%
Marginal Value	$3.05	$3.42	$3.01
Toowoomba households - values for the Murray Darling Area = 314,000sq km; river length:20,000km Sample response rate = 50%
Range	50%-65%	40%-60%	25%-45%
Marginal Value	$4.02	$6.28	$2.35
Mackay households - values Mackay Whitsunday Area = 9,000sq km; river length:700km Sample response rate = 61%
Range	50%-65%	40%-60%	45%-65%
Marginal Value	$4.60	$7.82	$2.42
Rockhampton households - values for the Fitzroy Area = 143,000sq km; river length:15,000km Sample response rate = 72%
Range	50%-65%	30%-50%	25%-45%
Marginal Value	$3.70	$6.69	$4.48
All households /regions combined
Marginal Value - average of regions	$3.72	$5.80	$2.88

Application process:

Select the marginal value that appears most applicable to the case study of interest,
Identify the proportional improvement by the relevant area for the case study of interest (e.g. a project might generate a 0.2% increase in healthy waterways),
Identify the relevant population base that is likely to benefit from an improvement in the case study of interest (e.g. the number of households in a regional area),
Identify any adjustment factors that might be involved (e.g. proportion of non-interested households),
Multiple the marginal value x the proportional improvement x the number of households x adjustment factors, and
Results give indicative annual value of the benefits of actions in the case study of interest.

Points to consider:

The values reported are annual values showing the amount that households would pay to get the estimated outcomes.
These values are marginal values which means they should be used to assess a change in environmental condition and should not be used to determine overall values. For example, the value for a 1% improvement in the area of soils in good condition should not be used to determine the value of all the soils in good condition in an area.
The stock of natural resource assets varies across regions, so the value of a change per hectare of area or kilometer of waterways will be quite different across regions and populations.
The range of asset condition changes that is relevant to the database is reported in the table. It will be less accurate to apportion values to larger changes in condition outside of these ranges.
Only a percentage of households in each region have participated in the surveys relevant to the database. The participation rates can be used to identify the proportions of each population likely to support the value estimates. The ‘proportion of support' can be reflected in an adjustment factor.
Many actions will only generate incremental improvements in the condition of a resource stock. The ‘proportion of improvement' might be another adjustment factor.

Using a database value for benefit transfer

There are a number of issues to consider when selecting a variable from the database table and applying it.

If the target location and population is the same or similar to one of the study locations and populations, these values can be applied.
- For example, if there is a requirement to estimate the values that a regional population in the Murray-Darling basin in Queensland might hold for a 1% improvement in vegetation in the region, then the estimate of $2.35 per household can be applied from the database.
If the required target locations/populations do not match the four regional case studies in the database, the analyst can choose to either:
- Apply the results from a similar case study area/population, or
- Use the average values for all residents/regions
Local-scale issues may only be relevant to local populations. However, there are some environmental protection issues where there may be benefits to the regional and state populations. The selection of the relevant population base is likely to be related to the scale of the issue involved and any iconic status of the environmental assets.
- For larger scale issues, it might be appropriate to extrapolate values across a regional population and to the capital city population (Brisbane). This might be a conservative population estimate.
- For larger scale issues, a wider population estimate might include all households in the State. This would be a less conservative population estimate.
Only a percentage of households in each region have participated in the surveys relevant to the database. Care needs to be taken in assuming that all households in a regional area hold an indicative value for an environmental improvement. There is no golden rule about how far sample values can be extrapolated and what the adjustment factor for non-participation should be.
- The most conservative option is to extrapolate the same survey response rate in the sample to the population. For example, if there was a 72% survey response rate in the Rockhampton community, then values can be extrapolated to 72% of the Rockhampton population.
- Some studies have assumed that 30% of the proportion of the population that did not respond to the survey hold the same values as survey respondents. In this case a higher rate of support can be assumed. This is a less conservative approach.
In the survey, soil, water, and vegetation resources were described in very general terms and are therefore suitable to be used in a wide range of situations.
- For example, improvements in healthy vegetation could include projects that eradicate weeds, revegetate areas or improve ground cover.
- For specific actions, the generic database values will only be a general guide to values of benefits.
Some projects or investment proposals might involve benefits to more than one category (e.g. healthy soils and healthy waterways). In these cases estimates of value can be made for each category and added.
Many actions will only generate incremental improvements in the condition of a resource stock. This might occur because the resource stock is already in good condition, and/or because the actions of interest only make a small improvement.
- An adjustment factor would need to be developed to represent the ‘proportion of improvement' generated.

Example of a database application: Vegetation improvement in the Fitzroy

The Fitzroy Basin Association in 2006 allocated approximately $180,000 to an incentive scheme to protect and improve biodiversity values in 13,647 ha of remnant vegetation. An example of how the public benefits of this scheme may be calculated is outlined below.

Selecting a value

Values for ‘healthy vegetation' at the state and Fitzroy regional level are available from the benefit transfer database.
These values were for a 1% improvement in condition.
Approximately 45% or 64,500 sq km of the vegetation in the Fitzroy Basin is currently in good condition. A one percent improvement would mean an increase of 645 sq km or 64,500ha.
An improvement in 13,647ha equates to 21% of 64,500ha.

The valuation results suggested that households in Rockhampton were willing to pay $4.48 per year to improve the condition of 64,500ha of vegetation in the Fitzroy Basin. This equates to a value of $0.94 per household for an improvement in 13,647ha.

In addition, households in Brisbane also hold values for improvements in the Fitzroy (see previous section). There is no specific value for this in the database and so it is appropriate to apply the average value from all households for all regions. This means that households in Brisbane were willing to pay $2.88 per year to improve the condition of 64,500ha of vegetation in the Fitzroy Basin. This equates to a value of $0.60 per household for an improvement in 13,647ha.

These household values can be extrapolated to estimate an overall value for the environmental improvements gained under the incentive scheme, as shown in the following table.

Benefits of improving vegetation condition in the Fitzroy Basin

	Rockhampton and Fitzroy Basin¹	Brisbane
Population²	193,722	1,508,161
Average household size	2.5	3.3
No of households	77,489	457,018
Survey response rate	72%	50%
Valid households	55,792	228,509
Household value for an improvement in 13,647ha	$0.94	$0.60
Total value - $/year	$52,445	$137,105

¹ Responses were elicited from households in the Rockhampton area and it is assumed households in the rest of the region have the same values.
² ABS 2001 Census

The individual household values for Brisbane were extrapolated to 50% of Brisbane households. This was the same as the response rate to the survey and is a conservative estimate. It assumes that people who did not respond to the survey had no value for improvements in vegetation condition in the Fitzroy Basin, which is most unlikely. There was a 72% response rate for the survey in Rockhampton and this was extrapolated to households in the entire Fitzroy Basin.

Results from the valuation survey also indicated the households in the rest of Queensland hold values for improving vegetation condition in the Fitzroy, but again, to take a conservative approach, these values were not included in this extrapolation exercise.

Overall, the general public would be willing to pay $189,550 per year (for up to 15 years) to improve the condition of 13,647ha of vegetation in the Fitzroy Basin. However, no adjustment factor was applied to take account of the proportion of vegetation involved that was already in good condition. This may mean that the estimated benefits are overstated. At the same time, values have only been estimated for the Brisbane and Fitzroy region populations, and values from the rest of Queensland have not been included.

The estimated benefit of $189,550 per year is considerably more than the $90,000 per year ($180,000 for two years) it cost FBA. The analysis demonstrates that the investment in vegetation improvement was beneficial.