Forgot your password?
Document Actions
The empirics of wetland valuation: a comprehensive summary and a meta-analysis of the literature
Ó Springer 2006
Environmental & Resource Economics (2006) 33: 223–250
DOI 10.1007/s10640-005-3104-4
The Empirics of Wetland Valuation:
A Comprehensive Summary and a
Meta-Analysis of the Literature
LUKE M. BRANDER1,*, RAYMOND J. G. M. FLORAX2,3 and
JAN E. VERMAAT1
1
Institute for Environmental Studies (IVM), Vrije Universiteit, De Boelelaan 1087, Amsterdam,
1081 HV, The Netherlands; 2Department of Agricultural Economics, Purdue University, West
Lafayette, USA; 3Department of Spatial Economics, Vrije Universiteit, Amsterdam, The
Netherlands; *Author for correspondence (E-mail: luke.brander@ivm.falw.vu.nl)
Accepted 24 August 2005
Abstract. Wetlands are highly productive ecosystems, providing a number of goods and
services that are of value to people. The open-access nature and the public-good characteristics
of wetlands often result in these regions being undervalued in decisions relating to their use
and conservation. There is now a substantial literature on wetland valuation, including two
meta-analyses that examine subsets of the available wetland valuation literature. We collected
over 190 wetland valuation studies, providing 215 value observations, in order to present a
more comprehensive meta-analysis of the valuation literature that includes tropical wetlands
(e.g., mangroves), estimates from diverse valuation methodologies, and a broader range of
wetland services (e.g., biodiversity value). We also aim for a more comprehensive geographical
coverage. We find that socio-economic variables, such as income and population density, that
are often omitted from such analyses are important in explaining wetland value. We also
assess the prospects for using this analysis for out-of-sample value transfer, and find average
transfer errors of 74%, with just under one-fifth of the transfers showing errors of 10% or less.
Key words: meta-analysis, valuation, value transfer, wetlands
JEL classifications: C53, D62, H23, Q20, Q25
1. Introduction
Wetlands are highly productive and valuable ecosystems. The public-good
characteristics of many of the goods and services they provide often results in
wetlands being undervalued in decisions relating to their use and conserva-
tion. Partly as a response to this situation, there is now substantial literature
on wetland valuation (Barbier et al. 1997; Bardecki 1998; Kazmierczak
This paper has not been submitted elsewhere in identical or similar form, nor will it be
during the first three months after its submission to the Publisher.
224 LUKE M. BRANDER ET AL.
2001). The empirical studies in this literature vary widely in their use of
valuation techniques, the actual products and services being valued, and the
type and geographical location of the wetlands being considered.
The resulting ‘‘flood of numbers’’ and the considerable cost associated
with performing a study that assesses the value of a wetland has stimulated
the use of research synthesis techniques, in particular meta-analysis (Stanley
2001; Smith and Pattanayak 2002; Bateman and Jones 2003). Meta-analysis
is concerned with a quantitative analysis of statistical summary indicators
reported in a series of similar empirical studies. In the case of wetland
valuation, a standardized shadow price can be analyzed, such as the dollar
value per year of 1 ha of wetland area. Meta-analysis extends beyond a state
of the art literature review. Proponents of meta-analysis maintain that the
valuable aspects of narrative reviews can be preserved in meta-analysis, and
are in fact extended with quantitative features (Rosenthal and DiMatteo
2001). Some authors even refer to meta-analysis as a quantitative literature
review (Stanley 2001).
Two wetland valuation meta-analyses already exist (Brouwer et al. 1999;
Woodward and Wui 2001). These meta-analyses examine subsets of the
available wetland valuation literature. They focus on temperate wetlands,
and they consider a limited set of wetland services. Brouwer et al. (1999)
restrict their sample to only contingent valuation studies. In addition, these
studies do not include socio-economic and georeferenced information for the
wetland sites in their respective meta-regression analysis. Consequently, there
is scope for a more comprehensive meta-analysis of the valuation literature
that includes tropical wetlands (e.g., mangroves), estimates from other
valuation methodologies, other wetland services (e.g., biodiversity value),
and estimates from more countries.
In this article, we provide a comprehensive overview of the empirical
wetland valuation literature, reviewing virtually all studies that appeared
over the last 25 years. We categorize the reported value estimates along
several dimensions (such as wetland type, size, services, and valuation
method), which leads to an exploratory synopsis of the determinants of
wetland value. This analysis is complemented by a more rigorous assessment
of the variation in wetland values by means of a meta-regression analysis. In
this analysis we include socio-economic and georeferenced variables in the
form of GDP per capita, population density, and latitude, as well as variables
reflecting wetland and study characteristics. This potentially facilitates the
use of ‘‘value transfer’’ to non-valued wetland sites as an alternative to
primary valuation, although the validity and accuracy of such a value
transfer has been questioned (Downing and Ozuna 1996; Brouwer and
Spaninks 1999; Brouwer 2000). Following up on, among others, Rosenberger
and Loomis (2000), and Bateman and Jones (2003) we explicitly investigate
225
THE EMPIRICS OF WETLAND VALUATION
the validity, the efficiency, and the robustness of value transfers based on the
meta-analysis of wetland values.
The organization of this article is as follows. Section 2 outlines the defi-
nition and typology of wetlands used in this article, the wetland functions
that are utilized by humans, and the valuation methods that are applied to
value various wetland services. This section also discusses the heterogeneity
of the value estimates. Section 3 gives an overview of the empirical wetland
valuation literature, and presents the results of an exploratory analysis. We
show the resulting descriptive statistics and cross-tabulations against, for
instance, type of wetlands, wetland services, and valuation methodology.
Section 4 describes the setup for a meta-regression, specifically the specifi-
cation and functional form of the meta-regression function. This section also
gives the regression output and an interpretation of the results. In Section 5,
we explore the validity, efficiency and robustness of using a meta-valuation
function that includes socio-economic and georeferenced information in a
value transfer exercise. Finally, Section 6 concludes and provides suggestions
for future research and policy.
2. Wetland Types, Functions and Values
A widely agreed upon, precise definition of what constitutes a wetland is not
available. However, in ‘‘The Convention on Wetlands,’’ a UNESCO-based
intergovernmental treaty on wetlands adopted in the Iranian city of Ramsar,
in 1971 (more commonly known as the ‘‘RAMSAR Convention’’) provides a
broad characterization. The RAMSAR convention on wetlands defines
wetlands very broadly as (Article 1.1):
areas of marsh, fen, peatland or water, whether natural or artificial,
permanent or temporary, with water that is static or flowing, fresh,
brackish or salt, including areas of marine water the depth of which at
low tide does not exceed six metres,
and points out (in Article 2.1) that wetlands:
may incorporate riparian and coastal zones adjacent to the wetlands,
and islands or bodies of marine water deeper than six metres at low tide
lying within the wetlands.
Depending on interpretation, this very inclusive definition encompasses a
large number of ecosystem types. As of 2004, the ‘‘RAMSAR Convention’’
includes 1369 wetland sites, located in 139 countries throughout the
world, although the location of the sites is strongly skewed towards Western
Europe (see http://ramsar.org/sitelist.pdf). The RAMSAR-sites cover over
120.5 million hectares of wetland. In this study, we use the same definition
226 LUKE M. BRANDER ET AL.
and we specifically classify wetlands into five types: mangroves, unvegetated
sediment, salt/brackish marsh, freshwater marsh, and freshwater woodland.
Depending partly on wetland type, wetlands provide a number of goods
and services that are of value to humans (Barbier 1991). The services
Table I. Ecological wetland functions, economic goods and services, types of value, and
applicable valuation methods
Ecological function Economic goods Value type Commonly used
valuation method(s)a
and services
Flood and Flood protection Indirect use Replacement cost
flow control Market prices
Opportunity cost
Storm buffering Storm protection Indirect use Replacement cost
Production function
Sediment retention Storm protection Indirect use Replacement cost
Production function
Groundwater Water supply Indirect use Production function, NFI
recharge/discharge Replacement cost
Water quality Improved water Indirect use CVM
maintenance/nutrient quality
retention Waste disposal Direct use Replacement cost
Habitat and nursery Commercial fishing Direct use Market prices, NFI
for plant and animal and hunting
species Recreational fishing Direct use TCM, CVM
and hunting
Harvesting of Direct use Market prices
natural materials
Energy resources Direct use Market prices
Biological diversity Appreciation of Non-use CVM
species existence
Micro-climate Climate stabilization Indirect use Production function
stabilization
Carbon sequestration Reduced global Indirect use Replacement cost
warming
Natural environment Amenity Direct use HP, CVM
Recreational Direct use CVM, TCM
activities
Appreciation of Non-use CVM
uniqueness to
culture/ heritage
Source: with modifications adapted from Barbier (1991, 1997), Brouwer et al. (1999), and
Woodward and Wui (2001).
a
Acronyms refer to the contingent valuation method (CVM), hedonic pricing (HP), net factor
income (NFI), and the travel cost method (TCM).
227
THE EMPIRICS OF WETLAND VALUATION
provided by wetlands are derived from, but should not be confused with,
their ecological and physical functions. Table I lists the main ecological/
physical functions of wetlands, and their associated economic goods and
services.
The range of services provided by wetlands is partly related to direct geo-
physical processes, such as sediment retention and the provision of flood and
storm buffering capacity, but it extends to wider climatologic, biological, and
socio-cultural functions, including impacts on local and global climate change
and stabilization, preservation of biodiversity, and the provision of natural
environmental amenities. In addition, wetlands provide ecological processes
enabling the extraction of goods and services in the form of natural resources
such as water, fish and other edible animals, wood, and energy, and they
provide the natural surroundings for recreational activities (see Larson et al.
1989; Barbier 1991, 1997; Brouwer et al. 1999; Woodward and Wui 2001).
The economic values associated with these wetland goods and services can
be categorized into distinct components of the total economic value
according to the type of use. Direct use values are derived from the uses made
of a wetland’s resources and services, for example wood for energy or
building, water for irrigation and the natural environment for recreation.
Indirect use values are associated with the indirect services provided by a
wetland’s natural functions, such as storm protection or nutrient retention.
Non-use values of wetlands are unrelated to any direct, indirect or future use,
but rather reflect the economic value that can be attached to the mere exis-
tence of a wetland (Pearce and Turner 1990).
These components of the total economic value of wetlands often do not
accrue to the owner of the wetland, and as a result, important wetland values
are often overlooked in decision-making on wetland conversion (see Cummings
and Harrison 1995). Some wetlands goods and services may be traded directly in
well functioning markets and therefore have readily observable (marginal)
values. However, due to market failures resulting from undefined property rights
or the (quasi) public good characteristics of some wetland services, many
valuable wetland services may not be traded directly or even indirectly through
markets. Examples of wetland services that are indirectly traded through
markets may include the amenities associated with housing located near to
wetlands or water supply provided to agriculture. Wetland services that may not
even be indirectly traded through markets include bequest and existence values.
In cases where the values of important wetland services are not observable in
well functioning markets, a number of non-market valuation methods may be
applied to estimate economic values.
A diverse range of valuation methods have been applied to value wetland
services, including the contingent valuation method (e.g., Farber 1988; Bateman
and Langford 1997), hedonic pricing (e.g., Lupi et al. 1991; Doss and Taff 1996),
travel cost method (e.g., Ramdial 1975; Cooper and Loomis 1993), production
228 LUKE M. BRANDER ET AL.
function approach (e.g., Acharya and Barbier 2000; Bell 1997), net factor
income approach (e.g., Amacher et al. 1989; Schuijt 2004), total revenue esti-
mation (e.g., Costanza et al. 1989; Raphael and Jaworski 1979), opportunity
cost (e.g., Leitch and Hovde 1996; Sathirathai and Barbier 2001), and replace-
ment cost (e.g., Breaux et al. 1995; Emerton and Kekulandala 2002). The
applicability of each of these methods depends largely on the wetland service
being valued and the type of value associated with it (Freeman 2003). Table I
lists the valuation methods next to the wetland services and value types that they
are commonly used to value. It must be noted that these valuation methods
differ considerably in terms of the welfare measures that they estimate (see
Freeman 2003; Kopp and Smith 1993; Carson et al. 1996). This source of het-
erogeneity in the meta-data may lead to problems of non-comparability between
estimated values and we need to be wary of comparing inconsistent concepts of
economic value (Brouwer 2000; Smith and Pattanayak 2002).
Table II lists the valuation methods used within the wetland valuation
literature together with a short description of each method and the welfare
measure that it estimates. The contingent valuation method is the only
method capable of estimating non-use values and by directly asking respondents
to state their WTP or WTA for (hypothetical) changes in environmental quality
Table II. Valuation methods and associated welfare measures
Valuation Method Short description Welfare measure
Contingent valuation Hypothetical questions Compensating or
to obtain WTP equivalent surplus
Travel cost Estimate demand (WTP) Consumer surplus
using travel costs to visit site
Hedonic pricing Estimate WTP using price Consumer surplus
differentials and characteristics
of related products
Production function Estimate value as an input Producer and
in production consumer surplus
Net factor income Assign value as revenue of Producer surplus
an associated product(s) net
of costs of other inputs
Replacement cost Cost of replacing the function Value larger thanthe
with an alternative technology current cost of supply
Opportunity cost Value of next best alternative Consumer surplus,
use of resources (e.g., agricultural producer surplus,
use of water and land) or total revenue for
next best alternative
Market prices Assigns value equal to the total Total revenue
market revenue of goods/services
229
THE EMPIRICS OF WETLAND VALUATION
or quantity it provides estimates of the technically precise welfare measures of
compensating and equivalent surplus. The hedonic pricing and travel cost
methods estimate the Marshallian consumer surplus, which approximates, and is
bounded by, the compensating variation (CV) and equivalent variation (EV)
welfare measures. For relatively small price changes, the error of approximation
between consumer surplus, CV and EV is small (Willig 1976). For large price
changes, however, (e.g., when considering a price change large enough to drive
the quantity demanded to zero) the error can be substantial (Freeman 2003). In
response to this potential error there are now numerous travel cost and hedonic
pricing studies that attempt to estimate Hicksian measures of consumer surplus
(see for example Shaw and Ozog 1999). There is empirical evidence to suggest
that revealed preference and contingent valuation methods produce broadly
similar value estimates (Bateman et al. 2004).1 The production function
approach estimates changes in consumer and producer surplus resulting from
quantity or quality changes in an environmental good that is used as an input in
a production process. If the price of output is unaffected by the environmental
change (i.e., if demand for the good is perfectly elastic), only producer surplus is
affected. The net factor income approach also estimates changes in producer
surplus by subtracting the costs of other inputs in production from total reve-
nue, and ascribes the remaining surplus as the value of the environmental input.
The remaining valuation methods do not have a sound basis in welfare
theory and therefore may be expected to over- or underestimate economic
values. The bold line across the middle of Table II indicates the distinction
between methods that have sound underpinnings in welfare economics and
those that do not. The total revenue approach simply estimates values as the
total revenue received from the sale of goods or services derived from the
environmental resource in question. This approach ignores the cost of all
other inputs in the production of these goods and services and will therefore
tend to overestimate producer surplus. The opportunity cost approach takes
the value of the next best alternative use of the resources used to provide the
ecosystem function being valued. This reflects the cost of supplying the good
or service and not the surplus associated with its use. The replacement cost
approach places values on ecosystem services by estimating the cost of
replacing them. This approach is based on the assumption that if individuals
incur costs to replace ecosystem functions, then the lost services must be
worth at least what people are willing to pay to replace them. Replacement
costs are not based on social preferences for ecosystem services, or individuals’
behavior in the absence of those services, and are unlikely to approximate
consumer and producer surpluses. Even with evidence to suggest that society is
willing to pay for an identified least cost replacement for lost ecosystem services
this is not a theoretically valid estimator of ecosystem service value.
The diversity in welfare measures being estimated makes it necessary
to clearly distinguish between the different valuation techniques in the
230 LUKE M. BRANDER ET AL.
meta-analysis. Although we may have a priori expectations as to the direction
of any bias associated with each valuation method,2 it is not possible to make
sensible adjustments to the observed valuation estimates to correct for these
biases. The differences in values estimated through each method are examined
initially in Section 3 and using a meta-regression in Section 4.
3. Overview of the Empirical Wetland Valuation Literature
Responding to the fact that the value of wetland services are often not known
and therefore not included in decisions regarding wetland use and conser-
vation, there is now a large number of studies attempting to value the partial
or total economic value of numerous wetland sites. For the purposes of
conducting a meta-analysis of wetland values, we have attempted to collect as
much of the available literature as possible. The methods employed in lit-
erature retrieval included searching electronic journal databases, libraries,
existing bibliographies and databases of wetland valuation studies, and
contact with authors and relevant agencies.3 In total, 191 studies related to
wetland valuation were collected. The earliest of these studies is a 1969 CVM
estimation of consumer surplus for wildfowl hunting in wetlands of the US
Pacific western flyway (Hammack and Brown 1974). We found eight wetland
valuation studies that were conducted in the 1970s, seven of which are for US
wetlands. Twenty-five studies were found from the 1980s, 124 from the 1990s
and 32 from 2000 or later. As well as an apparent upward trend in the
number of valuation studies being conducted, there has also been a diversi-
fication in the valuation methodologies used and the geographic location of
the wetlands being valued. It should be noted that the apparent trend of
increasing wetland valuation work might in part be due to the enhanced
availability of more recent valuation studies, i.e. through internet publication
and electronic journals.
The studies that have been collected cover various publication outlets,
including journal articles, project reports, dissertations, and book chapters.
This literature is very diverse in terms of the objectives of the research being
presented. Generally, the literature can be categorized into three groups
according to the primary focus of the study. First, some studies merely
estimate one or more values for a specific wetland site (e.g., Costanza et al.
1989; Lant and Roberts 1990; Cooper and Loomis 1991; Barbier and Strand
1998, Emerton et al. 1998; Klein and Bateman 1998; Acharya 2000). Second,
some studies review or compare already existing wetland valuations (e.g.,
Anderson and Rockel 1991; Gren and Soderqvist 1994; Barbier et al. 1997;
Dixon and Lal 1997; Bardeki 1998). Third, some studies develop a specific
methodological innovation for non-market valuation of wetlands (e.g.,
Barbier 1991; Creel and Loomis 1992; Dalecki et al. 1993; Gren et al. 1994;
Bateman and Langford 1997; Ellis and Fisher 1987; Haab and McConnell
231
THE EMPIRICS OF WETLAND VALUATION
1997; Pate and Loomis 1997). Obviously, many studies combine elements of
these three categories to some degree.
In addition, there are some important distinctions to be made between
studies within these three general categories. Within the group of studies that
primarily attempt to estimate values for specific wetland sites, there are some
that estimate values for alternative wetland management strategies (e.g.,
Freeman 1991; Ruitenbeek 1992; Bann 1997; Barbier and Thompson 1998;
Janssen and Padilla 1999; Van den Bergh et al. 2001). Other studies use value
transfer rather than primary valuation techniques to value a specific wetland
(e.g., Farber and Costanza 1987; Dahuri 1991; Farber 1992; Gren 1995;
Dharmaratne and Strand 2002). Within the group of studies that focuses on
methodological issues in wetland valuation, some empirically test different
survey or estimation techniques using real or hypothetical data whereas
others are purely conceptual (e.g., Bergstrom and Stoll 1993; Bystrom et al.
2000; Turner et al. 2000).
Of the 191 collected studies, 80 contained suitable and sufficient infor-
mation for the purposes of comparison in a statistical meta-analysis.4 From
these 80 studies, we were able to extract 215 separate observations of wetland
value. The maximum number of observations taken from one study is 10 and
the average number is approximately 2.7. Care was taken not to double count
value estimates that are reported in more than one study, or to include
estimates that were derived through value transfer from studies also included
in our data set. The main reasons for not including information from a study
are that it either reports already published results or is focused on method-
ological issues rather than primary valuation.
The observations in our data set are from 25 countries and all continents
are represented. Figure 1 presents a map of the spatial distribution of the
wetlands in our database, for which a value estimate is available.
It should be noted that the geographical distribution of observations in
our sample reflects the practice and availability of natural resource valua-
tion studies rather than the distribution of wetlands. North America, for
example, is particularly well represented with half our data set comprising
of observations from the US and Canada. The number of wetlands rep-
resented in our data set, however, is less than the number of observations
because multiple observations are taken from each study, which typically
only consider one wetland, and several studies have valued the same wet-
land. Although we have 16 separate valuation observations for Africa, it
can be seen from Figure 1 that these are for only five wetlands. For Aus-
tralasia, on the other hand, we have seven observations for five different
wetlands.
Figure 2 shows the number of wetland value observations in our sample
categorized according to wetland type, wetland service, and valuation
method. The valuation literature has clearly focused on freshwater wetlands.
232 LUKE M. BRANDER ET AL.
Figure 1. Geographic location of wetlands for which value estimates are available.
In addition to the large number of observations for North American fresh-
water wetlands, there is also a marked focus on mangrove valuation in Asia.
This valuation effort has been prompted by the large-scale conversion of
mangroves to other land uses, such as shrimp ponds, and logging in recent
decades.
A large number of different wetland services have been valued in the
literature, although not all of the services identified in Table I have been
valued (e.g., carbon sequestration and micro-climate stabilization). In order
to reflect the distinctions that are generally made between wetland services in
the valuation literature we have categorized wetland services in our database
slightly differently from the list in Table I. This involved combining some
services (e.g., flood control and storm buffering) and separating others (e.g.,
recreational hunting and fishing). The wetland service categories that we used
are: flood control and storm buffering, water supply, water quality, habitat
and nursery service (specifically support for commercial fisheries and hunt-
ing), recreational hunting, recreational fishing, amenity and other recrea-
tional uses, materials, fuel wood, and biodiversity. The number of observations
for each wetland service is presented in Figure 2. Most studies value only one
particular wetland service rather than all services provided by the wetland in
question. There is, however, also a significant number of studies that value two
or more services of a wetland and a small number that estimate total economic
value, i.e., referring to all important goods and services (for example, Blomquist
and Whitehead 1998; Leitch and Hovde 1996).
A broad range of valuation methodologies has been applied to value
wetlands. The numbers of observations for each method are shown in Figure 2.
The method most commonly used in the literature has been to observe the
233
THE EMPIRICS OF WETLAND VALUATION
Wetland type
52
Woodland
124
Freshwater marsh
57
Salt or brackish marsh
Unvegitated sediment 15
69
Mangrove
0 20 40 60 80 100 120 140
No. of observations
Wetland service
19
Bio diversity
60
Amenity
21
Fuel wood
39
Materials
40
Recreational fishing
50
Recreational hunting
73
Habitat and nursery
31
Water quality
23
Water supply
34
Flood control
0 10 20 30 40 50 60 70 80
No. of observations
Valuation method
11
Opportunity cost
91
Market prices
19
Production function
22
Net factor income
28
Replacement cost
19
TCM
5
HP
38
CVM
0 20 40 60 80 100
No. of observations
Figure 2. Number of observations for each wetland type, wetland service, and valuation
method.
market prices of products related to wetland services and then ascribe the total
revenue from the sale of such products as the value of the wetland. Contingent
valuation has also been widely used. As expected, the different valuation
methodologies have been applied to value different wetland services. CVM,
hedonic pricing and TCM have been applied to value amenity and recreational
values, replacement cost has largely been used to value the role of wetlands in
234 LUKE M. BRANDER ET AL.
improving water quality, and the production function approach has been used
to value the habitat and nursery services of wetlands. The market price approach
has been used to value most wetland services.
Wetland values have been reported in the literature in many different
metrics, currencies and referring to different years (e.g., WTP per household
per year, capitalized values, marginal value per acre, etc). In order to enable
comparison between these values we standardized them to 1995 US dollars
per hectare per year, following Woodward and Wui (2001).5 In standardizing
wetland values we faced the problem of distinguishing between average and
marginal values, both of which can be expressed as a monetary value per
hectare. The majority of wetland valuation studies have estimated total or
average wetland values but there is also a large number of estimates of
marginal wetland values. Small changes in wetlands should be valued using
marginal changes (Batie and Shabman 1979) whereas average values may be
useful for comparing the aggregate value of a wetland area relative to the size
of the area (Bergstrom et al. 1990). Expressing wetland values in per hectare
terms gives the impression that each hectare in a wetland is equally pro-
ductive, i.e., that wetlands exhibit constant returns to scale or equivalently
that the marginal wetland value is equal to average wetland value. Without
being able to convert marginal values to average values or vice versa, we
assume exactly this. This assumption is examined later on in the discussion
on whether wetlands exhibit increasing or decreasing returns to scale.
Standardizing wetland values to WTP per person as in Brouwer et al. (1999)
was not possible because several of the valuation methods used in the literature
(e.g., NFI, opportunity cost, replacement cost and market prices) do not pro-
duce WTP estimates. WTP per person or household on the other hand could be
converted to US$ ha)1 yr)1 given information on the wetland area and the
relevant population size. Using an annual dollar value per unit of area may also
facilitate the use of meta-analysis results for value transfer because in most cases
it is more straightforward to transfer values to a given wetland area than to the
relevant number of people that are willing to pay for wetland conservation.
For our data set the average annual wetland value is just over 2800 US$
per hectare. The median value, however, is 150 US$ ha)1 yr)1, showing that
the distribution of values is skewed with a long tail of high values. As
expected, the mean and median values of wetlands vary considerably by
continent, wetland type, wetland service and valuation method used. Figure 3
presents the mean and median wetland value for each continent, wetland type,
wetland service, and valuation method respectively. The information contained
in this Figure does not account for the variation in wetland values that is
explained by variation in other important variables. In order to examine the
importance of each variable in explaining the variation in wetland values while
accounting for variation in other variables we use a meta-regression as set out in
the following section. This graphical representation of the data, however, helps
235
THE EMPIRICS OF WETLAND VALUATION
Wetland value by continent
Australasia (7)
Africa (16)
Asia (46)
Europe (23)
South America (12)
North America (111)
1 10 100 1,000 10,000 100,000
Wetland value by wetland type
Woodland (52)
Fresh water marsh (124)
Salt/brackish marsh (57)
Unvegetated sediment (15)
Mangrove (69)
1 10 100 1,000 10,000 100,000
Wetland value by wetland service
Biodiversity (13)
Amenity (48)
Fuelwood (18)
Materials (32)
Recreational Fishing (36)
Recreational Hunting (50)
Habitat and Nursery (67)
Water Quality (25)
Water Supply (18)
Flood (26)
1 10 100 1,000 10,000 100,000
Wetland value by valuation method
Opportunity Cost (11)
Market Prices (91)
Production Function (19)
Net Factor Income (22)
Replacement Cost (28)
Travel Cost (19)
Hedonic Pricing (5)
CVM (38)
1 10 100 1,000 10,000 100,000
-1
Wetland value (1995 US $ ha-1yr ; log scale)
Figure 3. Mean and median wetland values for each continent, wetland type, wetland
service, and valuation method. The number of observations for each category is in
parentheses. The bars represent the means, the error bars represent the standard error of
the mean, and the black dots represent the medians.
236 LUKE M. BRANDER ET AL.
to give an initial understanding of the determinants of variation in wetland
values found in the literature.
Figure 3 shows that average wetland values are highest in Europe, followed
by North America, Australasia, Africa, Asia, and finally South America. It
also shows that the wetland type unvegetated sediment has the highest average
value of just over 9000 US$ ha)1 yr)1. Mangroves have the lowest average
value of just over 400 US$ ha)1 yr)1. In terms of median values the variation
is much less, suggesting that different wetland types do not have largely dif-
ferent values. In our sample, the biodiversity service of wetlands has the
highest average value (17,000 US$ ha)1 yr)1), and the use of wetlands for
collecting fuel wood and other raw materials has the lowest values (73 and 300
US$ ha)1 yr)1, respectively). Studies that have used the contingent valuation
method (CVM) have produced the highest estimates of wetland value, fol-
lowed by the replacement cost method and hedonic pricing. The lowest value
estimates are produced by the opportunity cost and production function
methods. These differences in values produced by alternative valuation
methodologies may in part be explained by the application of these methods
to value different wetland services (as described above), and also be due to the
differences and biases in welfare measures that each method estimates.
Another wetland characteristic that we may expect to determine wetland
value is its area. There is no clear a priori expectation of the sign of this
relationship given on the one hand that there may be diminishing marginal
returns to most wetland services as wetland size increases, but on the other
hand some ecological functions require minimum thresholds of habitat area
which suggests that wetland values may increase with size. Figure 4 plots
wetland value by wetland area and reveals a possible negative relationship
between the two. The trend line represents an estimated least squares
regression equation. The coefficient on wetland area is not significant. The
wetlands for which value estimates are available are generally medium to
large size wetland areas. This does have implications for the extent to which
economies of scale can be estimated reliably; see also Woodward and Wui
(2001) who conclude that over a large range of sizeable wetlands constant
returns to scale are apparent.
In addition to the wetland characteristics and valuation methods that are
examined above, we would expect that the value of a wetland is determined by
the socio-economic characteristics of its location. Information regarding per
capita income of the relevant population using each wetland was generally not
available in the valuation studies so we inputted this information from other
sources.6 Figure 4 also plots wetland value per hectare per year by GDP per
capita. As expected there appears to be a positive relationship between the
two. The coefficient on GDP per capita in the simple regression is significant at
the 10% level. We also included population density information for each
wetland site in the database in order to examine the influence of population
237
THE EMPIRICS OF WETLAND VALUATION
1000000.0
y = 1170.4x -0.1996
(1995 US$ ha yr ; log scale)
R2 = 0.0699
100000.0
10000.0
Wetland value
1000.0
-1
-1
100.0
10.0
1.0
0.1
1 10 100 1000 10000 100000 1000000 10000000
Wetland area (ha; log scale)
y = 65.586e6E-05x
1000000.0
(1995 US$ ha yr ; log scale)
R 2 = 0.0824
100000.0
10000.0
Wetland value
-1
1000.0
-1
100.0
10.0
1.0
0.1
0 10000 20000 30000 40000
GDP per capita (1995 US$)
1000000.0
y = 61.483x0.2412
R2 = 0.0204
(1995 US$ ha yr ; log scale)
100000.0
10000.0
Wetland value
-1
1000.0
-1
100.0
10.0
1.0
0.1
1 10 100 1000 10000
Population density ( in 1000s km-2; log scale)
Figure 4. Wetland value per hectare per year plotted against wetland area, GDP per
capita, and population density.
density on wetland values.7 Our expectation is that wetlands have a higher
value in areas with higher population density as most wetland services are
related to direct or indirect human use. The spatial relationship between
wetlands and centers of population is of course important in determining the
use made of wetland services. This spatial relationship will vary with a number
of factors including wetland service, transportation availability, physical
barriers, and cultural norms. For example, in the case of the recreational use
of a wetland we would expect that in the US the distance between a population
238 LUKE M. BRANDER ET AL.
center and the wetland would be of less importance than in a developing
country due to transportation availability and habits. Consequentially, the
‘‘catchment’’ area of population that might use a wetland would be much larger
in the US. We were not able to capture all of these considerations in the data but
use population densities for 50-km radius zones around each wetland site. Fig-
ure 4 plots wetland value by population density. There is an apparent positive
relationship between the two, although the estimated coefficient on population
density in the trend line equation is not significant.
4. Meta-Regression
The above exploratory analysis of the available data in the wetland valuation
literature does of course not allow for interactions between the various
explanatory variables. In order to attain marginal effects – given the inter-
ference of potentially relevant intervening characteristics – we use meta-
regression analysis to assess the relative importance of all potentially relevant
factors simultaneously. The dependent variable in our regression equation is
a vector of values in US$ per hectare per year in 1995 prices, labelled y. The
explanatory variables are grouped in three different matrices that include the
study characteristics in Xs (i.e., valuation method, marginal value), the wet-
land physical and geographical characteristics in Xp (i.e., wetland type, ser-
vices, area, urban, continent, latitude, and RAMSAR proportion), and the
socio-economic characteristics in Xe (i.e., GDP per capita, and population
density). The model fit was considerably improved, and heteroskedasticity
was mitigated, by using the logarithms of the dependent variable, GDP per
capita, population density, and wetland size. The estimated model is, in
matrix notation:
lnðyÞ ¼ a þ Xs bs þ Xp bp þ Xe be þ u
where a is the usual constant term, u a vector of residuals (assuming well
behaved underlying errors), and the vectors b contain the estimated coeffi-
cients on the respective explanatory variables.8 The regression results are
presented in Table III, using White-adjusted standard errors because the
Breusch–Pagan test still indicates that the model is heteroskedastic. Multi-
collinearity was tested for and judged not to be a serious problem.9 The
adjusted R2 value of 0.45 is reasonably high, and indicates that close to half
the variation in wetland value is explained by variation in our explanatory
variables. In this (largely) semi-log model, the coefficients measure the con-
stant proportional or relative change in the dependent variable for a given
absolute change in the value of the explanatory variable. For the explanatory
variables expressed as logarithms, the coefficients should be interpreted as
elasticities, that is, the percentage change in the dependent variable given a
(small) percentage change in the explanatory variable.
239
THE EMPIRICS OF WETLAND VALUATION
Table III. Meta-regression resultsa
Variableb
Category Coefficient Standard error
)6.98
Constant 4.67
1.16**
Socio-economic GDP per capita 0.46
(log)
0.47***
Population 0.12
density (log)
)0.11**
Geographic Wetland size 0.05
characteristics (log)
Latitude 0.03 0.07
(absolute value)
)0.0007
Latitude squared 0.0010
South America 0.23 1.19
Europe 0.84 0.92
Asia 2.01 1.34
3.51**
Africa 1.52
1.75*
Australasia 0.94
1.11**
Urban 0.48
1.49**
Valuation CVM 0.73
)0.71
methods Hedonic pricing 1.54
TCM 0.01 0.65
Replacement cost 0.63 0.81
Net factor income 0.19 0.61
)1.00
Production function 0.75
)0.04
Market prices 0.53
)0.03
Opportunity cost 0.72
0.95*
Type value Marginal 0.48
)0.56
Wetland type Mangrove 0.82
Unvegetated sediment 0.22 1.09
)0.31
Salt/brackish marsh 0.42
)1.46**
Fresh marsh 0.59
0.86**
Woodland 0.42
Wetland service Flood control 0.14 0.55
)0.95
Water supply 0.71
Water quality 0.63 0.74
)0.03
Habitat and nursery 0.35
)1.10**
Hunting 0.43
Fishing 0.06 0.36
)0.83**
Material 0.42
)1.24***
Fuelwood 0.45
Amenity 0.06 0.39
Biodiversity 0.06 0.81
)1.32*
RAMSAR RAMSAR proportion 0.70
202
n
240 LUKE M. BRANDER ET AL.
Table III. Continued
Variableb
Category Coefficient Standard error
R2-adjusted 0.45
5.50***
F
51.46***
Breusch–Pagan
a
OLS results with White-adjusted standard errors. The Breusch–Pagan test concerns heter-
oskedasticity and is v2 distributed with 36 degrees of freedom. Significance is indicated with
***, **, and * for the 1, 5, and 10% level, respectively.
b
The valuation methods, wetland types and wetland services are not strictly non-overlapping
variables. In other words, some wetlands provide more than one service, and comprise smaller
areas of different types. There is also not a one-to-one correspondence between an observed
value and the use of a specific valuation method. Consequently, there is no need for the
omission of one of the categories in order to avoid perfect collinearity.
Regarding the influence of wetland type on the wetland value, differences in
value associated with different wetland types are indicated by the coefficients
on these dummy variables. Two of these coefficients are significantly different
from zero suggesting that fresh marshes have the lowest value as compared to
the average and woodland wetlands have the highest value. This result was not
apparent from Figure 3, but the latter result was merely based on bivariate
comparisons.
On the issue of whether wetlands exhibit increasing or decreasing returns
to scale, the coefficient on the wetland size variable is small and negative, as
well as significant. This suggests that there are significant decreasing returns
to scale. We correct for the fact that marginal values may be significantly
different from average values, which is shown to be the case. Specifically,
marginal values are almost twice as high as compared to average values. The
decreasing return result confirms the findings of Woodward and Wui (2001),
who observe decreasing returns to scale for wetlands at the level of )0.17 and
)0.29 for a comparable function. It should be noted that the double-log
specification induces the returns to scale to decline geometrically with size
(see Woodward and Wui 2001, pp. 267–268), so that the elasticity approaches
zero with increasing size.
The differences in wetland values resulting from the availability of dif-
ferent services were touched on above. Wetland services that involve the
provision of direct use natural resources, such as fuel wood and other
materials tend to have lower than average values. Somewhat surprisingly,
wetlands that provide recreational hunting opportunities also tend to have
lower than average values.
Another unexpected result given the initial analysis of Figure 3 is that
North American wetlands tend to have lower values than wetlands located in
241
THE EMPIRICS OF WETLAND VALUATION
other continents. North America is included in the constant term of our
model and the coefficients on the dummy variables indicating the other
continents are all positive but only significant for Africa and Australasia. We
can only speculate on the reason for this. One possible explanation for
wetlands receiving a lower value in North America is the relative abundance
of substitute natural areas, particularly in comparison with Europe. One
should note that these results are obtained correcting for differences in
latitude. We hypothesized that the value of wetlands might be related non-linearly
(following a parabolic shape) to the absolute distance from the equator. This,
however, is not apparent in the estimation results.
For the socio-economic variables that we were able to include in our
model, the results confirm our expectations. The coefficient on the GDP
per capita variable is positive and highly significant – suggesting a slightly
elastic effect of income on the value of wetland services. The interpretation
of the result is that a 10% increase in GDP per capita results in roughly a
12% increase in wetland value. There is also a positive and significant
relationship between population density and wetland value as described
above. This relationship, however, is inelastic, but this may very well be
due to the dummy variable ‘‘Urban’’ that is also included. On average,
urban wetlands have a value that is significantly higher than rural wet-
lands.
The results for the valuation methodology dummy variables show that
value estimates from contingent valuation, replacement cost, the travel cost
method and NFI methods are higher than estimates from other valuation
methods. The only statistically significant result, however, pertains to CVM,
which show the highest values as compared to the other valuation methods.
This result is in contrast with the findings of Woodward and Wui (2001), who
observed that the hedonic pricing and the replacement cost method produce
higher values than CVM.
Finally, the variable referring to a comparison between RAMSAR and
other sites shows an interesting outcome. We use a variable operational-
ized as the proportion of the wetland that is designated a RAMSAR site,
and it shows up indicating significantly lower values for RAMSAR sites.
Possible explanations for this result might be that certain uses of these
wetlands are restricted and therefore not valued, or that WTP bids for the
conservation of these sites are affected by respondents’ knowledge that
they are already protected.
5. Value Transfer
There remains the question of whether the results from this meta-analysis can
be used for value transfer, that is, the prediction or estimation of the value of
a wetland given knowledge of its physical and socio-economic characteristics.
242 LUKE M. BRANDER ET AL.
There is substantial academic and policy interest in the potential for and
validity of value transfer as it offers a means of estimating monetary values
for environmental resources without performing relatively time consuming
and expensive primary valuation studies (see Florax et al. 2002).
There are two general approaches to value transfer: direct value transfer
and function value transfer. The first involves simply transferring the value(s)
estimated in one or more primary studies to the policy site in question.
Ideally, the study site and policy site should be similar in their characteristics
or adjustments should be made to the transferred value to reflect differences
in site characteristics (Brouwer 2000). The second approach involves transferring
values to a policy site based on its known characteristics using a value transfer
function, possibly estimated through a meta-regression. Rosenberger and Phipps
(2001) identify the important assumptions underlying the use of a value function
for value transfer:
(1) there exists a valuation function that links the values of a resource with
the characteristics of the relevant markets and sites across space and
time, and from which values for specific characteristics can be inferred,
(2) differences between sites can be captured through a price vector,
(3) values are stable over time, or vary in a systematic way, and
(4) the sampled primary valuation studies provide ‘‘correct’’ estimates of
resource value.
It is generally accepted that function transfers perform better than direct
value transfers for a number of reasons. Firstly, information from a larger
number of studies is used. Secondly, methodological differences between
primary valuation studies can be controlled for. And thirdly, explanatory
variables can be adjusted to represent the policy site (Bateman and Jones
2003). Rosenberger and Phipps (2001) review a number of studies that test
the relative performance of direct value transfer and function value transfer
(see for example Loomis 1992; Parsons and Kealy 1994; Brouwer and
Spaninks 1999). The general conclusion is that meta-analysis value transfer
functions perform better than other approaches (see also Engel, 2002).
For a number of reasons value transfer may result in substantial
‘transfer errors’, particularly when the characteristics of the site to which
values are being transferred are not well represented in the data underlying
the estimated value function (Brouwer 2000). Another reason for error
might be that the characterization of wetland types and wetland services is
oftentimes rather crude. The use of dummy variables to characterize types
and services does not capture the true variation in these characteristics.
Similarly, it is difficult to capture important quality and quantity differ-
ences in provision of services across sites. Unfortunately, we cannot
overcome this in our value transfer experiment, but we have instead
focused on including non-sample information such as GDP per capita and
243
THE EMPIRICS OF WETLAND VALUATION
population density. To the extent that such information is relevant, it may
increase the accuracy of value transfer estimates. We have also attempted
to be as comprehensive as possible in our collection of valuation studies
but clearly due to the limited availability of studies, our sample focuses on
certain wetland types, services and locations. This raises the question of
the validity of value transfer to policy sites in countries that are not
represented in the data (Shrestha and Loomis 2001). Another important
source of transfer error is related to errors in the primary valuation studies
from which the value transfer function is derived (Woodward and Wui
2001). As described in Section 2, there are a number of biases associated
with each valuation method that may result in mis-estimation of ‘‘true’’
resource values and thereby introduce a source of error in estimating a
value function.
As a first step, before actually performing a value transfer, we looked at
the in-sample forecast performance of our model. As an indicator we used the
Mean Absolute Percentage Error (MAPE), which is defined as |(yobs)yest)/
yobs|. For our sample of 201 observations10 the average MAPE equals 58%,
which is a rather high forecast error. However, if we look at the average
MAPE for different quartiles of the data series ordering them by magnitude
of the wetland value, in ascending order, we find an average MAPE of 173,
26, 16, and 19, respectively. This indicates that the fit for low wetland values
is particularly poor and the fit for medium to high-valued wetlands is much
more acceptable.
Subsequently, we use an n)1 data splitting technique to estimate 200 value
transfer functions by applying the estimated parameters generated with n)1
observations to the omitted observation.11 The upper panel in Figure 5
presents the observed and predicted wetland values in ascending order of
observed value. This shows that for a number of observations there is a
considerable difference between observed and predicted values. It also indi-
cates that our value transfer function systematically over-predicts very low
wetland values and slightly under-predicts high values.12 The lower panel in
Figure 5 presents the transfer error (defined as MAPE) associated with each
observation ranked in ascending order. The overall average transfer error is
74%, which is slightly higher than the in-sample forecast error – as can be
expected. The average transfer error for different quartiles of the data series
ordered by magnitude of the wetland value, in ascending order, is 213, 34, 19,
and 33, respectively. Slightly less than 20% of the sample has transfer errors
of 10% or less, and roughly 15% of observations show transfer errors over
100%. In comparison to other value transfer exercises (reviewed in Brouwer
2000; Rosenberger and Phipps 2001) our results appear to be similar despite
the relative diversity of our data in terms of the activities and services being
valued, valuation techniques employed, geographic locations and socio-
economic characteristics. One advantage, however, of our value transfer is
244 LUKE M. BRANDER ET AL.
16
Yobs
Ypred
12
lnvalue (US$ha-1y-1)
8
4
0
-4
Observations
1500
Transfer error
1000
500
0
Observations
Figure 5. Observed and predicted wetland values and transfer errors, ranked in
ascending order of observed wetland value.
that we have been able to include important population density and income
variables that have not been included in most other studies.
An important proviso for the validity of value transfer is that sites for
which a transfer is being conducted, and the method on which the valuation
is based, are adequately represented in the meta-dataset (see Rosenberger and
Phipps 2001). It is, however, not enough to merely count sites with specific
characteristics and the number of studies using particular methods. In this
case a multivariate analysis is preferable as well. We therefore regressed the
exogenous variables distinguished in the meta-analysis on the transfer errors,
defined as the mean absolute percentage error. This analysis shows that the
transfer errors are significantly negatively correlated with the dummies for
Africa, Asia, and Australasia. This result contradicts the suggestion of
Rosenberger and Phipps (2001) that the accuracy of value transfer is directly
related to the incidence of specific characteristics in the meta-database,
because most of our observations are from North America. There is also a
significant positive correlation for the replacement cost method and a similar
negative correlation with net factor income methods. Finally, the transfer
245
THE EMPIRICS OF WETLAND VALUATION
errors are also significantly positively correlated with the variable measuring
the proportion of the wetland that is under the RAMSAR convention.
6. Conclusions
This article provides a comprehensive overview of the wetland valuation lit-
erature and has attempted to identify the important physical, socio-economic
and study characteristics that determine wetland value. The wetland valuation
literature has been shown to be extremely diverse in terms of values estimated,
wetland types considered and valuation methods used. The value estimates
produced by different valuation methodologies are not necessarily directly
comparable and need to be explicitly modeled in our meta-regression. One of
the key results from our meta-regression analysis is the importance of the
GDP per capita and population density variables in explaining variation in
wetland value. Both variables were shown to have a positive relationship with
wetland value. Although such information is often not available in primary
valuation studies it is suggested that future valuation meta-analyses attempt
to include relevant socio-economic information from other sources in order to
represent important determinants of value. Another interesting result is that
CVM studies have tended, ceteris paribus, to produce higher value estimates
than other valuation methods. This contrasts with our expectations and with
the findings of Woodward and Wui (2001). In terms of the ecological and
physical characteristics of wetlands, we found freshwater marshes to be valued
less than other wetland types and a negative relationship between wetland size
and value. Of the various wetland services that we identified, water quality
improvement was found to be valued the highest. Two unexpected results
from this meta-analysis were that North American wetlands and RAMSAR
sites were found to be valued lower than other wetlands.
Using an n)1 data splitting technique we examined the robustness of using
our meta-regression for out-of-sample value transfer. The resulting average
transfer error is 74%, which is comparable to the transfer errors associated
with other value transfer exercises in the literature. Given the high costs of
performing primary valuation studies, this level of transfer error may be
acceptable in considering transferred values as input in wetland conservation
decisions. However, our value transfer function systematically over-predicts
very low wetland values and slightly under-predicts high values. Remarkably
enough, the value transfer performs better for wetlands that are located in
countries not well represented in our data (Africa, Asia, and Australasia).
The value transfer error is positively correlated with transfers based on the
replacement cost method. The same result holds for the degree to which
wetlands are RAMSAR sites. We therefore urge caution in using the
results of such a meta-analysis for value transfer, particularly to policy sites
for which their characteristics are not well represented in the underlying
246 LUKE M. BRANDER ET AL.
valuation studies. There is clearly still a need for more (and higher quality)
primary valuation studies, particularly in developing countries.
Acknowledgements
This research has, in part, been carried out within the project DINAS-
COAST (Dynamic and Interactive Assessment of National, Regional and
Global Vulnerability of Coastal Zones to Climate Change and Sea-Level
Rise), which is funded by the EU Directorate-General Research under pro-
ject number EVK2-CT-2000–00084. The authors would like to thank the
participants of the International Colloquium ‘‘Meta-Analysis in Economics,’’
held in Amsterdam in December 2002, for useful comments and suggestions.
Notes
1. Carson et al. (1996) review 83 valuation studies for quasi-public goods from which 616
comparisons of contingent valuation (CV) and revealed preference (RP) estimates are
made. The sample mean CV/RP ratio is 0.89 with a 95% confidence interval of 0.81–0.96
and a median of 0.75. While the results from this study show that RP methods produces
higher value estimates than CV, it also shows that estimates from these two methods are
within the same range.
2. Bateman and Jones (2003) set out expectations for the ordering of values estimated
through different analytical modes of the contingent valuation and travel cost methods.
3. One particularly useful source was an annotated review of the wetland valuation literature
for the period 1988–1998 by Bardecki (1998).
4. In order to compare observations in a statistical meta-analysis we required sufficient
information on a number of key variables. These are: wetland value, area, type,
function(s) being valued, location, year of valuation and valuation method used.
5. We used GDP deflators and purchasing power parity converters from the World Bank
World Development Indicators 2002 to standardize values estimated in different years and
different currencies.
6. The 1995 national per capita income level data were taken from World Bank World
Development Indicators 2000, and US state level data were taken from the US Census
2000 for the US states.
7. The population densities included in our analysis represent an area of 50-km radius around
each wetland site. Population and population density information was derived from CIESIN
data. The process by which this data was converted to represent each wetland site in our data
set is described in Wagtendonk and Omtzigt (2003).
8. A multi-level modelling (MLM) approach such as used in Brouwer et al. (1999), and
Bateman and Jones (2003) was considered but not adopted. This approach incorporates
natural hierarchies or levels within the data, e.g., study sites, author, method and study,
allowing the (somewhat unrealistic) assumption of independence between estimates to be
relaxed. MLM is, however, problematic in that it requires the use of dummy variables
for each group within a level, e.g., for each author or study site. This may be feasible in
reasonably limited or homogeneous data sets but less so for very diverse data (such as
ours).
247
THE EMPIRICS OF WETLAND VALUATION
9. The presence of multicollinearity was tested by examining the correlation coefficients on
pairs of explanatory variables and by regressing selected explanatory variables on the
remaining variables.
10. For this exercise we used a slightly restricted data set as one observation for which the log
value was very close to zero would otherwise have a disproportionate influence.
11. This resembles the jackknife resampling technique (see Efron 1982).
12. Again it should be noted that the primary valuation studies are highly unlikely to have
produced ‘‘true’’ estimates of wetland value and therefore we would not necessarily expect
(or want) transferred values to be exactly equal to primary valuation study results.
References
Acharya, G. (2000), ‘Approaches to Valuing the Hidden Hydrological Services of Wetland
Ecosystems’, Ecological Economics 35, 63–74.
Acharya, G. and E. B. Barbier (2000), ‘Valuing Ground Water Recharge through Agricultural
Production in the Hadejia-Nguru Wetlands in Northern Nigeria’, Agricultural Economics
22, 247–259.
Amacher, G. S., R. J. Brazee, J. W. Bulkley and R. A. Moll (1989), Application of Wetland
Valuation Techniques: Examples from Great Lakes Coastal Wetlands. Ann Arbor, MI:
University of Michigan, School of Natural Resources.
Anderson, R. and M. Rockel (1991), ‘Economic Valuation of Wetlands’, American Petroleum
Institute, Discussion Paper No. 65, Washington D.C.
Bann, C. (1997), An Economic Analysis of Alternative Mangrove Management Strategies in Koh
Kong Province, Cambodia. Singapore: EEPSEA research report series.
Batie, S. S. and L. Shabman (1979), ‘Valuing Non-Market Goods: Conceptual and Empirical
Issues’, American Journal of Agricultural Economics 61, 931–932.
Bateman, I. and I. H. Langford (1997), ‘Non-Users Willingness to Pay for a National Park:
An Application of the Contingent Valuation Method’, Regional Studies 31, 571–582.
Bateman, I. and L. P. Jones (2003), ‘Contrasting Conventional with Multi-level Modeling
Approaches to Meta-analysis: Expectation Consistency in U.K. Woodland Recreation
Values’, Land Economics 79(2), 235–258.
Bateman, I.J., A.A. Lovett and J.S. Brainard (2004), Applied Environmental Economics: A GIS
Approach to Cost-Benefit Analysis. Cambridge University Press.
Barbier, E. B., (1991), ‘An Approach to Economic Evaluation of Tropical Wetlands: With
Examples from Guatemala and Nicaragua’, N. P. Girvan and D. Simons, eds., Caribbean
Ecology and Economics, Caribbean Conservation Association, St. Michael, Barbados, 207–231.
Barbier, E. B., M. Acreman and D. Knowler (1997), Economic Valuation of Wetlands: A Guide
for Policy Makers and Planners. Gland, Switzerland.: Ramsar Convention Bureau.
Barbier, E. B. and I. Strand (1998), ‘Valuing Mangrove-Fishery Linkages’, Environmental and
Resource Economics 12, 151–166.
Barbier, E. B. and J. R. Thompson (1998), ‘The Value of Water: Floodplain versus Large-scale
Irrigation Benefits in Northern Nigeria’, Ambio 26, 434–440.
Bardecki, M. J. (1998), Wetlands and Economics: An Annotated Review of the Literature, 1988–
1998. Ontario: Environment Canada.
Bell, F. W. (1997), ‘The Economic Valuation of Saltwater Marsh Supporting Marine
Recreational Fishing in the Southeastern United States’, Ecological Economics 21, 243–254.
Bergstrom, J. C., J. R. Stoll, J. P. Titre and V. L. Wright (1990), ‘Economic Valuation of
Wetlands-based Recreation’, Ecological Economics 2, 129–147.
Bergstrom, J. C. and J. R. Stoll (1993), ‘Value Estimator Models for Wetlands-based
Recreational use Values’, Land Economics 69, 132–137.
248 LUKE M. BRANDER ET AL.
Blomquist, G. C. and J. C. Whitehead (1998), ‘Resource Quality Information and Validity of
Willingness to Pay in Contingent Valuation’, Resource and Energy Economics 20, 179–196.
Breaux, A., S. C. Farber and J. Day (1995), ‘Using Natural Coastal Wetlands Systems for
Wastewater Treatment: An Economic Benefit Analysis’, Journal of Environmental
Management 44, 285–291.
Brouwer, R. (2000), ‘Environmental Value Transfer: State of the Art and Future Prospects’,
Ecological Economics 32, 137–152.
Brouwer, R., I. H. Langford, I. J. Bateman, T. C. Crowards and R. K. Turner (1999), ‘A Meta-
Analysis of Wetland Contingent Valuation Studies’, Regional Environmental Change 1, 47–57.
Brouwer, R. and F. A. Spaninks (1999), ‘The Validity of Environmental Benefits Transfer:
Further Empirical Testing’, Environmental and Resource Economics 14, 95–117.
Bystrom, O., H. Andersson and I.-M. Gren (2000), ‘Economic Criteria for Using Wetlands as
Nitrogen Sinks under Uncertainty’, Ecological Economics 35, 35–45.
Carson, R. T., N. E. Flores, K. M. Martin and J. L. Wright (1996), ‘Contingent Valuation and
Revealed Preference Methodologies: Comparing the Estimates for Quasi-Public Goods’,
Land Economics 72(Feb), 80–99.
Cooper, J. and J. Loomis (1991), ÔEconomic Value of Wildlife Resources in the San Joaquin Valley:
Hunting and Fishing ValuesÕ, in A. Dinar and D. Zilberman, eds., The Economics and
Management of Water and Drainage in Agriculture. Norwell Mass: Kluwer Academic
Publishers.
Cooper, J. and J. Loomis (1993), ‘Testing whether Waterfowl Hunting Benefits Increase with
Greater Water Deliveries to Wetlands’, Environmental and Resource Economics 3, 545–561.
Costanza, R., S. C. Farber and J. Maxwell (1989), ‘Valuation and Management of Wetland
Ecosystems’, Ecological Economics 1, 335–361.
Creel, M. and J. Loomis (1992), ‘Recreation Value of Water to Wetlands in the San Joaquin
Valley: Linked Multinomial Logit and Count Data Trip Frequency Models’, Water
Resources Research 28, 2597–2606.
Cummings, R. G. and G. W. Harrison (1995), ‘The Measurement and Decomposition of Non-
use Values: A Critical Review’, Environmental and Resource Economics 5, 225–247.
Dahuri, R. (1991), An Approach to Coastal Resource Utilisation: The Nature and Role of
Sustainable Development in East Kalimantan Coastal Zone, Indonesia. Halifax, Canada:
Dalhousie University PhD Thesis.
Dharmaratne, G. and I. Strand (2002), Adaptation to Climate Change in the Caribbean: The
Role of Economic Valuation. Report to the CPACC, London.
Dalecki, M. G., J. C. Whitehead and G. C. Blomquist (1993), ‘Sample Non-Response Bias and
Aggregate Benefits in Contingent Valuation: An Examination of Early, Late and Non-
Respondents’, Journal of Environmental Management 38, 133–143.
Dixon, J. A. and P. N. Lal (1997). ‘The Management of Coastal Wetlands: Economic Analysis
of Combined Ecologic-Economic Systems’, in P. Dasgupta and K.-G. Maler, eds., The
¨
Environment and Emerging Development Issues, Vol 2. Oxford University Press.
Doss, C. R. and S. J. Taff (1996), ‘The Influence of Wetland Type and Wetland Proximity on
Residential Property Values’, Journal of Agricultural and Resource Economics 21, 120–129.
Downing, M. and T. Ozuna (1996), ‘Testing the Reliability of the Benefit Function Transfer
Approach’, Journal of Environmental Economics and Management 30, 316–322.
Efron, B. (1982), The Jackknife, the Bootstrap and other Resampling Plans. Philadelphia:
Society for Industrial and Applied Mathematics.
Ellis, G. M. and A. C. Fisher (1987), ‘Valuing the Environment as Input’, Journal of
Environmental Management 25, 149–156.
Emerton, L., L. Iyango, P. Luwum and A. Malinga (1998), The Present Economic Value of
Nakivubo Urban Wetland, Uganda. Nairobi, Kenya: IUCN Report.
249
THE EMPIRICS OF WETLAND VALUATION
Emerton, L. and B. Kekulandala (2002). ‘Assessment of the Economic Value of Muthura-
jawela Wetland’, IUCN – The World Conservation Union, Sri Lanka Country Office.
Engel, S. (2002), ÔBenefit Function Transfer versus Meta-Analysis as Policy-making Tools: A
ComparisonÕ, in R. J. G. M. Florax, P. Nijkamp and K. G. Willis, eds., Comparative
Environmental Economic Assessment. (pp. 133–153). Edward Elgar: Cheltenham.
Farber, S. (1988), ‘Non-user’s WTP for a National Park: An Application and Critique of the
Contingent Valuation Method’, Regional Studies 31, 571–582.
Farber, S. (1992), ‘The Economic Cost of Residential Environmental Risk: A Case Study of
Louisiana’, Journal of Environmental Management 36, 1–16.
Farber, S. and R. Costanza (1987), ‘The Economic Value of Wetland Systems’, Journal of
Environmental Management 24, 41–51.
Florax, R. J. G. M., P. Nijkamp and K. G. Willis (eds.) (2002), Comparative Environmental
Economic Assessment. Edward Elgar: Cheltenham.
Freeman III, A. M. (1991), ‘Valuing Environmental Resources under Alternative Manage-
ment Regimes’, Ecological Economics 3, 247–256.
Freeman III, A. M. (2003), The Measurement of Environmental and Resource Values: Theories
and Methods. Washington D.C.: Resources for the Future.
Gren, I.-M. and T. Soderqvist (1994), Economic Valuation of Wetlands: A Survey. Stockholm:
Sweden, Beijer Discussion Paper Series No. 54.
Gren, I.-M., C. Folke, K. Turner and I. Bateman (1994), ‘Primary and Secondary Values of
Wetland Ecosystems’, Environmental and Resource Economics 4, 55–74.
Gren, I.-M. (1995), ‘The Value of Investing in Wetlands for Nitrogen Abatement’, European
Review of Agricultural Economics 22, 157–172.
Haab, T.C. and K.E. McConnell (1997). ‘A Simple Method for Bounding WTP using a Probit
or Logit Model’. Economic Working Paper 9713, East Carolina University.
Hammack, J. and G. M. Brown (1974), Waterfowl and Wetlands: Toward A Bio-Economic
Analysis. Washington, D.C.: Resources for the Future, John Hopkins University Press.
Janssen, R. and J. E. Padilla (1999), ‘Preservation of Conversion? Valuation and Evaluation
of a Mangrove Forest in the Philippines’, Environmental and Resource Economics 14,
297–331.
Kazmierczak, R. F. (2001), Economic Linkages between Coastal Wetlands and Hunting and
Fishing: A Review of Value Estimates Reported in the Published Literature. Baton Rouge:
Louisiana State University Agricultural Center, Staff Paper 2001 – 03.
Klein, R. J. T. and I. J. Bateman (1998), ‘The Recreational Value of Cley Marshes Nature
Reserve: An Argument Against Managed Retreat?’, Journal of the Chartered Institution of
Water and Environmental Management 12, 280–285.
Kopp, R. J., V. K. Smith (eds.) (1993), Valuing Natural Assets, The Economics of Natural
Resource Damage Assessment. Washington, D.C.: Resources for the Future.
Lant, C. L. and R. S. Roberts (1990), ‘Greenbelts in the Cornbelt: Riparian Wetlands,
Intrinsic Values and Market Failure’, Environment and Planning 22, 1375–1388.
Larson, J. S., P. R. Adamus and E. J. Clairain Jr (1989), ‘Functional Assessment of
Freshwater Wetlands: A Manual and Training Outline’, Glaud, Switzerland: WWF
Publication 89–6:62 pp.
Leitch, J. A. and B. Hovde (1996), ‘Empirical Valuation of Prairie Potholes: Five Case
Studies’, Great Plains Research 6, 25–39.
Loomis, J. B. (1992), ‘The Evolution of a More Rigorous Approach to Benefit Transfer:
Benefit Function Transfer’, Water Resources Research 28(3), 701–705.
Lupi, F., T. Graham-Tomasi and S. J. Taff (1991), ‘A Hedonic Approach to Urban Wetland
Valuation’. Staff Paper P91-8, Department of Agricultural and Applied Economics,
University of Minnesota.
250 LUKE M. BRANDER ET AL.
Parsons, G. R. and M. J. Kealy (1994), ‘Benefit Transfer in a Random Utility Model of
Recreation’, Water Resources Research 30(8), 2477–2484.
Pate, J. and J. Loomis (1997), ‘The Effect of Distance on Willingness to Pay Values: A Case
Study of Wetlands and Salmon in California’, Ecological Economics 20, 199–207.
Pearce, D. and R. K. Turner (1990), Economics of Natural Resources and the Environment.
Hemel Hempstead, UK: Harvester Wheatsheaf.
Ramdial, B. S. (1975), ‘The Social and Economic Importance of the Caroni Swamp in
Trinidad and Tobago’. PhD thesis, University of Michigan.
Raphael, C. N. and E. Jaworski (1979), ‘Economic Value of Fish, Wildlife, and Recreation in
Michigan’s Coastal Wetlands’, Coastal Zone Management Journal 5, 181–194.
Rosenberger, R. and J. B. Loomis (2000), ‘Using Meta-Analysis For Benefit Transfer: In
Sample Convergent Validity Tests of an Outdoor Recreation Database’, Water Resources
Research 36, 1097–1107.
Rosenberger, R. and T. T. Phipps (2001). ‘Site correspondence effects in benefit transfers: A
meta-analysis transfer function’. Research Paper 2001–6, West Virginia University.
Rosenthal, R. and M. R. DiMatteo (2001), ‘Meta-Analysis: Recent Developments in
Quantitative Methods for Literature Reviews’, Annual Review of Psychology 52, 59–82.
Ruitenbeek, H. J. (1992), ‘Mangrove Management: An Economic Analysis of Management
Options with a Focus on Bintuni Bay, Irian Jaya’. Environmental Management
Development in Indonesia Project, EMDI Environmental Report 8, Jakarta, Indonesia.
Sathirathai, S. and E. B. Barbier (2001), ‘Valuing Mangrove Conservation in Southern
Thailand’, Contemporary Economic Policy 19, 109–122.
Schuijt, K. (2004), Land and Water Use of Wetlands in Africa: Economic Values of African
Wetlands. Laxenburg, Austria: International Institute for Applied Systems Analysis,
Report No. IR-02 – 063.
Shaw, W. D. and M. T. Ozog (1999), ‘Modeling Overnight Recreation Trip Choice:
Application of a Repeated Nested Multinomial Logit Model’, Environmental and Resource
Economics 13, 397–414.
Shrestha, R. K. and J. B. Loomis (2001), ‘Testing a Meta-Analysis Model for Benefit Transfer
in International Outdoor Recreation’, Ecological Economics 39, 67–83.
Smith, V. K. and K. Pattanayak (2002), ‘Is Meta-Analysis A Noah’s Ark for Non-Market
Valuation?’, Environmental and Resource Economics 22, 271–96.
Stanley, T. D. (2001), ‘Wheat from Chaff: Meta-Analysis as Quantitative Literature Review’,
Journal of Economic Perspectives 15, 131–50.
Turner, R. K., J. C. J. M. van den Bergh, T. Soderqvist, A. Barendregt, J. van der Straaten,
E. Maltby and E. C. van Ierland (2000), ‘Ecological-Economic Analysis of Wetlands:
Scientific Integration for Management and Policy’, Ecological Economics 35, 7–23.
Van den Bergh, J., A. Barendregt, A. Gilbert, M. van Herwijnedn, P. van Horssen,
P. Kandelaars and C. Lorenz (2001), ‘Spatial Economic-Hydrological Modeling and
Evaluation of Land Use Impacts in the Vecht Wetlands Area’, Environmental Modeling
and Assessment 6, 87–100.
Wagtendonk, A. and N. Omtzigt (2003), Analysis of Population Density Around Wetland Areas
in GPW (Gridded Population of the World) Files. Amsterdam: SPINlab document,
Institute for Environmental Studies.
Willig, R. D. (1976), ‘Consumers’ Surplus without An Apology’, American Economic Review
66, 589–597.
Woodward, R. T. and Y.-S. Wui (2001), ‘The Economic Value of Wetland Services: A
Meta-Analysis’, Ecological Economics 37, 257–270.
Environmental & Resource Economics (2006) 33: 223–250
DOI 10.1007/s10640-005-3104-4
The Empirics of Wetland Valuation:
A Comprehensive Summary and a
Meta-Analysis of the Literature
LUKE M. BRANDER1,*, RAYMOND J. G. M. FLORAX2,3 and
JAN E. VERMAAT1
1
Institute for Environmental Studies (IVM), Vrije Universiteit, De Boelelaan 1087, Amsterdam,
1081 HV, The Netherlands; 2Department of Agricultural Economics, Purdue University, West
Lafayette, USA; 3Department of Spatial Economics, Vrije Universiteit, Amsterdam, The
Netherlands; *Author for correspondence (E-mail: luke.brander@ivm.falw.vu.nl)
Accepted 24 August 2005
Abstract. Wetlands are highly productive ecosystems, providing a number of goods and
services that are of value to people. The open-access nature and the public-good characteristics
of wetlands often result in these regions being undervalued in decisions relating to their use
and conservation. There is now a substantial literature on wetland valuation, including two
meta-analyses that examine subsets of the available wetland valuation literature. We collected
over 190 wetland valuation studies, providing 215 value observations, in order to present a
more comprehensive meta-analysis of the valuation literature that includes tropical wetlands
(e.g., mangroves), estimates from diverse valuation methodologies, and a broader range of
wetland services (e.g., biodiversity value). We also aim for a more comprehensive geographical
coverage. We find that socio-economic variables, such as income and population density, that
are often omitted from such analyses are important in explaining wetland value. We also
assess the prospects for using this analysis for out-of-sample value transfer, and find average
transfer errors of 74%, with just under one-fifth of the transfers showing errors of 10% or less.
Key words: meta-analysis, valuation, value transfer, wetlands
JEL classifications: C53, D62, H23, Q20, Q25
1. Introduction
Wetlands are highly productive and valuable ecosystems. The public-good
characteristics of many of the goods and services they provide often results in
wetlands being undervalued in decisions relating to their use and conserva-
tion. Partly as a response to this situation, there is now substantial literature
on wetland valuation (Barbier et al. 1997; Bardecki 1998; Kazmierczak
This paper has not been submitted elsewhere in identical or similar form, nor will it be
during the first three months after its submission to the Publisher.
224 LUKE M. BRANDER ET AL.
2001). The empirical studies in this literature vary widely in their use of
valuation techniques, the actual products and services being valued, and the
type and geographical location of the wetlands being considered.
The resulting ‘‘flood of numbers’’ and the considerable cost associated
with performing a study that assesses the value of a wetland has stimulated
the use of research synthesis techniques, in particular meta-analysis (Stanley
2001; Smith and Pattanayak 2002; Bateman and Jones 2003). Meta-analysis
is concerned with a quantitative analysis of statistical summary indicators
reported in a series of similar empirical studies. In the case of wetland
valuation, a standardized shadow price can be analyzed, such as the dollar
value per year of 1 ha of wetland area. Meta-analysis extends beyond a state
of the art literature review. Proponents of meta-analysis maintain that the
valuable aspects of narrative reviews can be preserved in meta-analysis, and
are in fact extended with quantitative features (Rosenthal and DiMatteo
2001). Some authors even refer to meta-analysis as a quantitative literature
review (Stanley 2001).
Two wetland valuation meta-analyses already exist (Brouwer et al. 1999;
Woodward and Wui 2001). These meta-analyses examine subsets of the
available wetland valuation literature. They focus on temperate wetlands,
and they consider a limited set of wetland services. Brouwer et al. (1999)
restrict their sample to only contingent valuation studies. In addition, these
studies do not include socio-economic and georeferenced information for the
wetland sites in their respective meta-regression analysis. Consequently, there
is scope for a more comprehensive meta-analysis of the valuation literature
that includes tropical wetlands (e.g., mangroves), estimates from other
valuation methodologies, other wetland services (e.g., biodiversity value),
and estimates from more countries.
In this article, we provide a comprehensive overview of the empirical
wetland valuation literature, reviewing virtually all studies that appeared
over the last 25 years. We categorize the reported value estimates along
several dimensions (such as wetland type, size, services, and valuation
method), which leads to an exploratory synopsis of the determinants of
wetland value. This analysis is complemented by a more rigorous assessment
of the variation in wetland values by means of a meta-regression analysis. In
this analysis we include socio-economic and georeferenced variables in the
form of GDP per capita, population density, and latitude, as well as variables
reflecting wetland and study characteristics. This potentially facilitates the
use of ‘‘value transfer’’ to non-valued wetland sites as an alternative to
primary valuation, although the validity and accuracy of such a value
transfer has been questioned (Downing and Ozuna 1996; Brouwer and
Spaninks 1999; Brouwer 2000). Following up on, among others, Rosenberger
and Loomis (2000), and Bateman and Jones (2003) we explicitly investigate
225
THE EMPIRICS OF WETLAND VALUATION
the validity, the efficiency, and the robustness of value transfers based on the
meta-analysis of wetland values.
The organization of this article is as follows. Section 2 outlines the defi-
nition and typology of wetlands used in this article, the wetland functions
that are utilized by humans, and the valuation methods that are applied to
value various wetland services. This section also discusses the heterogeneity
of the value estimates. Section 3 gives an overview of the empirical wetland
valuation literature, and presents the results of an exploratory analysis. We
show the resulting descriptive statistics and cross-tabulations against, for
instance, type of wetlands, wetland services, and valuation methodology.
Section 4 describes the setup for a meta-regression, specifically the specifi-
cation and functional form of the meta-regression function. This section also
gives the regression output and an interpretation of the results. In Section 5,
we explore the validity, efficiency and robustness of using a meta-valuation
function that includes socio-economic and georeferenced information in a
value transfer exercise. Finally, Section 6 concludes and provides suggestions
for future research and policy.
2. Wetland Types, Functions and Values
A widely agreed upon, precise definition of what constitutes a wetland is not
available. However, in ‘‘The Convention on Wetlands,’’ a UNESCO-based
intergovernmental treaty on wetlands adopted in the Iranian city of Ramsar,
in 1971 (more commonly known as the ‘‘RAMSAR Convention’’) provides a
broad characterization. The RAMSAR convention on wetlands defines
wetlands very broadly as (Article 1.1):
areas of marsh, fen, peatland or water, whether natural or artificial,
permanent or temporary, with water that is static or flowing, fresh,
brackish or salt, including areas of marine water the depth of which at
low tide does not exceed six metres,
and points out (in Article 2.1) that wetlands:
may incorporate riparian and coastal zones adjacent to the wetlands,
and islands or bodies of marine water deeper than six metres at low tide
lying within the wetlands.
Depending on interpretation, this very inclusive definition encompasses a
large number of ecosystem types. As of 2004, the ‘‘RAMSAR Convention’’
includes 1369 wetland sites, located in 139 countries throughout the
world, although the location of the sites is strongly skewed towards Western
Europe (see http://ramsar.org/sitelist.pdf). The RAMSAR-sites cover over
120.5 million hectares of wetland. In this study, we use the same definition
226 LUKE M. BRANDER ET AL.
and we specifically classify wetlands into five types: mangroves, unvegetated
sediment, salt/brackish marsh, freshwater marsh, and freshwater woodland.
Depending partly on wetland type, wetlands provide a number of goods
and services that are of value to humans (Barbier 1991). The services
Table I. Ecological wetland functions, economic goods and services, types of value, and
applicable valuation methods
Ecological function Economic goods Value type Commonly used
valuation method(s)a
and services
Flood and Flood protection Indirect use Replacement cost
flow control Market prices
Opportunity cost
Storm buffering Storm protection Indirect use Replacement cost
Production function
Sediment retention Storm protection Indirect use Replacement cost
Production function
Groundwater Water supply Indirect use Production function, NFI
recharge/discharge Replacement cost
Water quality Improved water Indirect use CVM
maintenance/nutrient quality
retention Waste disposal Direct use Replacement cost
Habitat and nursery Commercial fishing Direct use Market prices, NFI
for plant and animal and hunting
species Recreational fishing Direct use TCM, CVM
and hunting
Harvesting of Direct use Market prices
natural materials
Energy resources Direct use Market prices
Biological diversity Appreciation of Non-use CVM
species existence
Micro-climate Climate stabilization Indirect use Production function
stabilization
Carbon sequestration Reduced global Indirect use Replacement cost
warming
Natural environment Amenity Direct use HP, CVM
Recreational Direct use CVM, TCM
activities
Appreciation of Non-use CVM
uniqueness to
culture/ heritage
Source: with modifications adapted from Barbier (1991, 1997), Brouwer et al. (1999), and
Woodward and Wui (2001).
a
Acronyms refer to the contingent valuation method (CVM), hedonic pricing (HP), net factor
income (NFI), and the travel cost method (TCM).
227
THE EMPIRICS OF WETLAND VALUATION
provided by wetlands are derived from, but should not be confused with,
their ecological and physical functions. Table I lists the main ecological/
physical functions of wetlands, and their associated economic goods and
services.
The range of services provided by wetlands is partly related to direct geo-
physical processes, such as sediment retention and the provision of flood and
storm buffering capacity, but it extends to wider climatologic, biological, and
socio-cultural functions, including impacts on local and global climate change
and stabilization, preservation of biodiversity, and the provision of natural
environmental amenities. In addition, wetlands provide ecological processes
enabling the extraction of goods and services in the form of natural resources
such as water, fish and other edible animals, wood, and energy, and they
provide the natural surroundings for recreational activities (see Larson et al.
1989; Barbier 1991, 1997; Brouwer et al. 1999; Woodward and Wui 2001).
The economic values associated with these wetland goods and services can
be categorized into distinct components of the total economic value
according to the type of use. Direct use values are derived from the uses made
of a wetland’s resources and services, for example wood for energy or
building, water for irrigation and the natural environment for recreation.
Indirect use values are associated with the indirect services provided by a
wetland’s natural functions, such as storm protection or nutrient retention.
Non-use values of wetlands are unrelated to any direct, indirect or future use,
but rather reflect the economic value that can be attached to the mere exis-
tence of a wetland (Pearce and Turner 1990).
These components of the total economic value of wetlands often do not
accrue to the owner of the wetland, and as a result, important wetland values
are often overlooked in decision-making on wetland conversion (see Cummings
and Harrison 1995). Some wetlands goods and services may be traded directly in
well functioning markets and therefore have readily observable (marginal)
values. However, due to market failures resulting from undefined property rights
or the (quasi) public good characteristics of some wetland services, many
valuable wetland services may not be traded directly or even indirectly through
markets. Examples of wetland services that are indirectly traded through
markets may include the amenities associated with housing located near to
wetlands or water supply provided to agriculture. Wetland services that may not
even be indirectly traded through markets include bequest and existence values.
In cases where the values of important wetland services are not observable in
well functioning markets, a number of non-market valuation methods may be
applied to estimate economic values.
A diverse range of valuation methods have been applied to value wetland
services, including the contingent valuation method (e.g., Farber 1988; Bateman
and Langford 1997), hedonic pricing (e.g., Lupi et al. 1991; Doss and Taff 1996),
travel cost method (e.g., Ramdial 1975; Cooper and Loomis 1993), production
228 LUKE M. BRANDER ET AL.
function approach (e.g., Acharya and Barbier 2000; Bell 1997), net factor
income approach (e.g., Amacher et al. 1989; Schuijt 2004), total revenue esti-
mation (e.g., Costanza et al. 1989; Raphael and Jaworski 1979), opportunity
cost (e.g., Leitch and Hovde 1996; Sathirathai and Barbier 2001), and replace-
ment cost (e.g., Breaux et al. 1995; Emerton and Kekulandala 2002). The
applicability of each of these methods depends largely on the wetland service
being valued and the type of value associated with it (Freeman 2003). Table I
lists the valuation methods next to the wetland services and value types that they
are commonly used to value. It must be noted that these valuation methods
differ considerably in terms of the welfare measures that they estimate (see
Freeman 2003; Kopp and Smith 1993; Carson et al. 1996). This source of het-
erogeneity in the meta-data may lead to problems of non-comparability between
estimated values and we need to be wary of comparing inconsistent concepts of
economic value (Brouwer 2000; Smith and Pattanayak 2002).
Table II lists the valuation methods used within the wetland valuation
literature together with a short description of each method and the welfare
measure that it estimates. The contingent valuation method is the only
method capable of estimating non-use values and by directly asking respondents
to state their WTP or WTA for (hypothetical) changes in environmental quality
Table II. Valuation methods and associated welfare measures
Valuation Method Short description Welfare measure
Contingent valuation Hypothetical questions Compensating or
to obtain WTP equivalent surplus
Travel cost Estimate demand (WTP) Consumer surplus
using travel costs to visit site
Hedonic pricing Estimate WTP using price Consumer surplus
differentials and characteristics
of related products
Production function Estimate value as an input Producer and
in production consumer surplus
Net factor income Assign value as revenue of Producer surplus
an associated product(s) net
of costs of other inputs
Replacement cost Cost of replacing the function Value larger thanthe
with an alternative technology current cost of supply
Opportunity cost Value of next best alternative Consumer surplus,
use of resources (e.g., agricultural producer surplus,
use of water and land) or total revenue for
next best alternative
Market prices Assigns value equal to the total Total revenue
market revenue of goods/services
229
THE EMPIRICS OF WETLAND VALUATION
or quantity it provides estimates of the technically precise welfare measures of
compensating and equivalent surplus. The hedonic pricing and travel cost
methods estimate the Marshallian consumer surplus, which approximates, and is
bounded by, the compensating variation (CV) and equivalent variation (EV)
welfare measures. For relatively small price changes, the error of approximation
between consumer surplus, CV and EV is small (Willig 1976). For large price
changes, however, (e.g., when considering a price change large enough to drive
the quantity demanded to zero) the error can be substantial (Freeman 2003). In
response to this potential error there are now numerous travel cost and hedonic
pricing studies that attempt to estimate Hicksian measures of consumer surplus
(see for example Shaw and Ozog 1999). There is empirical evidence to suggest
that revealed preference and contingent valuation methods produce broadly
similar value estimates (Bateman et al. 2004).1 The production function
approach estimates changes in consumer and producer surplus resulting from
quantity or quality changes in an environmental good that is used as an input in
a production process. If the price of output is unaffected by the environmental
change (i.e., if demand for the good is perfectly elastic), only producer surplus is
affected. The net factor income approach also estimates changes in producer
surplus by subtracting the costs of other inputs in production from total reve-
nue, and ascribes the remaining surplus as the value of the environmental input.
The remaining valuation methods do not have a sound basis in welfare
theory and therefore may be expected to over- or underestimate economic
values. The bold line across the middle of Table II indicates the distinction
between methods that have sound underpinnings in welfare economics and
those that do not. The total revenue approach simply estimates values as the
total revenue received from the sale of goods or services derived from the
environmental resource in question. This approach ignores the cost of all
other inputs in the production of these goods and services and will therefore
tend to overestimate producer surplus. The opportunity cost approach takes
the value of the next best alternative use of the resources used to provide the
ecosystem function being valued. This reflects the cost of supplying the good
or service and not the surplus associated with its use. The replacement cost
approach places values on ecosystem services by estimating the cost of
replacing them. This approach is based on the assumption that if individuals
incur costs to replace ecosystem functions, then the lost services must be
worth at least what people are willing to pay to replace them. Replacement
costs are not based on social preferences for ecosystem services, or individuals’
behavior in the absence of those services, and are unlikely to approximate
consumer and producer surpluses. Even with evidence to suggest that society is
willing to pay for an identified least cost replacement for lost ecosystem services
this is not a theoretically valid estimator of ecosystem service value.
The diversity in welfare measures being estimated makes it necessary
to clearly distinguish between the different valuation techniques in the
230 LUKE M. BRANDER ET AL.
meta-analysis. Although we may have a priori expectations as to the direction
of any bias associated with each valuation method,2 it is not possible to make
sensible adjustments to the observed valuation estimates to correct for these
biases. The differences in values estimated through each method are examined
initially in Section 3 and using a meta-regression in Section 4.
3. Overview of the Empirical Wetland Valuation Literature
Responding to the fact that the value of wetland services are often not known
and therefore not included in decisions regarding wetland use and conser-
vation, there is now a large number of studies attempting to value the partial
or total economic value of numerous wetland sites. For the purposes of
conducting a meta-analysis of wetland values, we have attempted to collect as
much of the available literature as possible. The methods employed in lit-
erature retrieval included searching electronic journal databases, libraries,
existing bibliographies and databases of wetland valuation studies, and
contact with authors and relevant agencies.3 In total, 191 studies related to
wetland valuation were collected. The earliest of these studies is a 1969 CVM
estimation of consumer surplus for wildfowl hunting in wetlands of the US
Pacific western flyway (Hammack and Brown 1974). We found eight wetland
valuation studies that were conducted in the 1970s, seven of which are for US
wetlands. Twenty-five studies were found from the 1980s, 124 from the 1990s
and 32 from 2000 or later. As well as an apparent upward trend in the
number of valuation studies being conducted, there has also been a diversi-
fication in the valuation methodologies used and the geographic location of
the wetlands being valued. It should be noted that the apparent trend of
increasing wetland valuation work might in part be due to the enhanced
availability of more recent valuation studies, i.e. through internet publication
and electronic journals.
The studies that have been collected cover various publication outlets,
including journal articles, project reports, dissertations, and book chapters.
This literature is very diverse in terms of the objectives of the research being
presented. Generally, the literature can be categorized into three groups
according to the primary focus of the study. First, some studies merely
estimate one or more values for a specific wetland site (e.g., Costanza et al.
1989; Lant and Roberts 1990; Cooper and Loomis 1991; Barbier and Strand
1998, Emerton et al. 1998; Klein and Bateman 1998; Acharya 2000). Second,
some studies review or compare already existing wetland valuations (e.g.,
Anderson and Rockel 1991; Gren and Soderqvist 1994; Barbier et al. 1997;
Dixon and Lal 1997; Bardeki 1998). Third, some studies develop a specific
methodological innovation for non-market valuation of wetlands (e.g.,
Barbier 1991; Creel and Loomis 1992; Dalecki et al. 1993; Gren et al. 1994;
Bateman and Langford 1997; Ellis and Fisher 1987; Haab and McConnell
231
THE EMPIRICS OF WETLAND VALUATION
1997; Pate and Loomis 1997). Obviously, many studies combine elements of
these three categories to some degree.
In addition, there are some important distinctions to be made between
studies within these three general categories. Within the group of studies that
primarily attempt to estimate values for specific wetland sites, there are some
that estimate values for alternative wetland management strategies (e.g.,
Freeman 1991; Ruitenbeek 1992; Bann 1997; Barbier and Thompson 1998;
Janssen and Padilla 1999; Van den Bergh et al. 2001). Other studies use value
transfer rather than primary valuation techniques to value a specific wetland
(e.g., Farber and Costanza 1987; Dahuri 1991; Farber 1992; Gren 1995;
Dharmaratne and Strand 2002). Within the group of studies that focuses on
methodological issues in wetland valuation, some empirically test different
survey or estimation techniques using real or hypothetical data whereas
others are purely conceptual (e.g., Bergstrom and Stoll 1993; Bystrom et al.
2000; Turner et al. 2000).
Of the 191 collected studies, 80 contained suitable and sufficient infor-
mation for the purposes of comparison in a statistical meta-analysis.4 From
these 80 studies, we were able to extract 215 separate observations of wetland
value. The maximum number of observations taken from one study is 10 and
the average number is approximately 2.7. Care was taken not to double count
value estimates that are reported in more than one study, or to include
estimates that were derived through value transfer from studies also included
in our data set. The main reasons for not including information from a study
are that it either reports already published results or is focused on method-
ological issues rather than primary valuation.
The observations in our data set are from 25 countries and all continents
are represented. Figure 1 presents a map of the spatial distribution of the
wetlands in our database, for which a value estimate is available.
It should be noted that the geographical distribution of observations in
our sample reflects the practice and availability of natural resource valua-
tion studies rather than the distribution of wetlands. North America, for
example, is particularly well represented with half our data set comprising
of observations from the US and Canada. The number of wetlands rep-
resented in our data set, however, is less than the number of observations
because multiple observations are taken from each study, which typically
only consider one wetland, and several studies have valued the same wet-
land. Although we have 16 separate valuation observations for Africa, it
can be seen from Figure 1 that these are for only five wetlands. For Aus-
tralasia, on the other hand, we have seven observations for five different
wetlands.
Figure 2 shows the number of wetland value observations in our sample
categorized according to wetland type, wetland service, and valuation
method. The valuation literature has clearly focused on freshwater wetlands.
232 LUKE M. BRANDER ET AL.
Figure 1. Geographic location of wetlands for which value estimates are available.
In addition to the large number of observations for North American fresh-
water wetlands, there is also a marked focus on mangrove valuation in Asia.
This valuation effort has been prompted by the large-scale conversion of
mangroves to other land uses, such as shrimp ponds, and logging in recent
decades.
A large number of different wetland services have been valued in the
literature, although not all of the services identified in Table I have been
valued (e.g., carbon sequestration and micro-climate stabilization). In order
to reflect the distinctions that are generally made between wetland services in
the valuation literature we have categorized wetland services in our database
slightly differently from the list in Table I. This involved combining some
services (e.g., flood control and storm buffering) and separating others (e.g.,
recreational hunting and fishing). The wetland service categories that we used
are: flood control and storm buffering, water supply, water quality, habitat
and nursery service (specifically support for commercial fisheries and hunt-
ing), recreational hunting, recreational fishing, amenity and other recrea-
tional uses, materials, fuel wood, and biodiversity. The number of observations
for each wetland service is presented in Figure 2. Most studies value only one
particular wetland service rather than all services provided by the wetland in
question. There is, however, also a significant number of studies that value two
or more services of a wetland and a small number that estimate total economic
value, i.e., referring to all important goods and services (for example, Blomquist
and Whitehead 1998; Leitch and Hovde 1996).
A broad range of valuation methodologies has been applied to value
wetlands. The numbers of observations for each method are shown in Figure 2.
The method most commonly used in the literature has been to observe the
233
THE EMPIRICS OF WETLAND VALUATION
Wetland type
52
Woodland
124
Freshwater marsh
57
Salt or brackish marsh
Unvegitated sediment 15
69
Mangrove
0 20 40 60 80 100 120 140
No. of observations
Wetland service
19
Bio diversity
60
Amenity
21
Fuel wood
39
Materials
40
Recreational fishing
50
Recreational hunting
73
Habitat and nursery
31
Water quality
23
Water supply
34
Flood control
0 10 20 30 40 50 60 70 80
No. of observations
Valuation method
11
Opportunity cost
91
Market prices
19
Production function
22
Net factor income
28
Replacement cost
19
TCM
5
HP
38
CVM
0 20 40 60 80 100
No. of observations
Figure 2. Number of observations for each wetland type, wetland service, and valuation
method.
market prices of products related to wetland services and then ascribe the total
revenue from the sale of such products as the value of the wetland. Contingent
valuation has also been widely used. As expected, the different valuation
methodologies have been applied to value different wetland services. CVM,
hedonic pricing and TCM have been applied to value amenity and recreational
values, replacement cost has largely been used to value the role of wetlands in
234 LUKE M. BRANDER ET AL.
improving water quality, and the production function approach has been used
to value the habitat and nursery services of wetlands. The market price approach
has been used to value most wetland services.
Wetland values have been reported in the literature in many different
metrics, currencies and referring to different years (e.g., WTP per household
per year, capitalized values, marginal value per acre, etc). In order to enable
comparison between these values we standardized them to 1995 US dollars
per hectare per year, following Woodward and Wui (2001).5 In standardizing
wetland values we faced the problem of distinguishing between average and
marginal values, both of which can be expressed as a monetary value per
hectare. The majority of wetland valuation studies have estimated total or
average wetland values but there is also a large number of estimates of
marginal wetland values. Small changes in wetlands should be valued using
marginal changes (Batie and Shabman 1979) whereas average values may be
useful for comparing the aggregate value of a wetland area relative to the size
of the area (Bergstrom et al. 1990). Expressing wetland values in per hectare
terms gives the impression that each hectare in a wetland is equally pro-
ductive, i.e., that wetlands exhibit constant returns to scale or equivalently
that the marginal wetland value is equal to average wetland value. Without
being able to convert marginal values to average values or vice versa, we
assume exactly this. This assumption is examined later on in the discussion
on whether wetlands exhibit increasing or decreasing returns to scale.
Standardizing wetland values to WTP per person as in Brouwer et al. (1999)
was not possible because several of the valuation methods used in the literature
(e.g., NFI, opportunity cost, replacement cost and market prices) do not pro-
duce WTP estimates. WTP per person or household on the other hand could be
converted to US$ ha)1 yr)1 given information on the wetland area and the
relevant population size. Using an annual dollar value per unit of area may also
facilitate the use of meta-analysis results for value transfer because in most cases
it is more straightforward to transfer values to a given wetland area than to the
relevant number of people that are willing to pay for wetland conservation.
For our data set the average annual wetland value is just over 2800 US$
per hectare. The median value, however, is 150 US$ ha)1 yr)1, showing that
the distribution of values is skewed with a long tail of high values. As
expected, the mean and median values of wetlands vary considerably by
continent, wetland type, wetland service and valuation method used. Figure 3
presents the mean and median wetland value for each continent, wetland type,
wetland service, and valuation method respectively. The information contained
in this Figure does not account for the variation in wetland values that is
explained by variation in other important variables. In order to examine the
importance of each variable in explaining the variation in wetland values while
accounting for variation in other variables we use a meta-regression as set out in
the following section. This graphical representation of the data, however, helps
235
THE EMPIRICS OF WETLAND VALUATION
Wetland value by continent
Australasia (7)
Africa (16)
Asia (46)
Europe (23)
South America (12)
North America (111)
1 10 100 1,000 10,000 100,000
Wetland value by wetland type
Woodland (52)
Fresh water marsh (124)
Salt/brackish marsh (57)
Unvegetated sediment (15)
Mangrove (69)
1 10 100 1,000 10,000 100,000
Wetland value by wetland service
Biodiversity (13)
Amenity (48)
Fuelwood (18)
Materials (32)
Recreational Fishing (36)
Recreational Hunting (50)
Habitat and Nursery (67)
Water Quality (25)
Water Supply (18)
Flood (26)
1 10 100 1,000 10,000 100,000
Wetland value by valuation method
Opportunity Cost (11)
Market Prices (91)
Production Function (19)
Net Factor Income (22)
Replacement Cost (28)
Travel Cost (19)
Hedonic Pricing (5)
CVM (38)
1 10 100 1,000 10,000 100,000
-1
Wetland value (1995 US $ ha-1yr ; log scale)
Figure 3. Mean and median wetland values for each continent, wetland type, wetland
service, and valuation method. The number of observations for each category is in
parentheses. The bars represent the means, the error bars represent the standard error of
the mean, and the black dots represent the medians.
236 LUKE M. BRANDER ET AL.
to give an initial understanding of the determinants of variation in wetland
values found in the literature.
Figure 3 shows that average wetland values are highest in Europe, followed
by North America, Australasia, Africa, Asia, and finally South America. It
also shows that the wetland type unvegetated sediment has the highest average
value of just over 9000 US$ ha)1 yr)1. Mangroves have the lowest average
value of just over 400 US$ ha)1 yr)1. In terms of median values the variation
is much less, suggesting that different wetland types do not have largely dif-
ferent values. In our sample, the biodiversity service of wetlands has the
highest average value (17,000 US$ ha)1 yr)1), and the use of wetlands for
collecting fuel wood and other raw materials has the lowest values (73 and 300
US$ ha)1 yr)1, respectively). Studies that have used the contingent valuation
method (CVM) have produced the highest estimates of wetland value, fol-
lowed by the replacement cost method and hedonic pricing. The lowest value
estimates are produced by the opportunity cost and production function
methods. These differences in values produced by alternative valuation
methodologies may in part be explained by the application of these methods
to value different wetland services (as described above), and also be due to the
differences and biases in welfare measures that each method estimates.
Another wetland characteristic that we may expect to determine wetland
value is its area. There is no clear a priori expectation of the sign of this
relationship given on the one hand that there may be diminishing marginal
returns to most wetland services as wetland size increases, but on the other
hand some ecological functions require minimum thresholds of habitat area
which suggests that wetland values may increase with size. Figure 4 plots
wetland value by wetland area and reveals a possible negative relationship
between the two. The trend line represents an estimated least squares
regression equation. The coefficient on wetland area is not significant. The
wetlands for which value estimates are available are generally medium to
large size wetland areas. This does have implications for the extent to which
economies of scale can be estimated reliably; see also Woodward and Wui
(2001) who conclude that over a large range of sizeable wetlands constant
returns to scale are apparent.
In addition to the wetland characteristics and valuation methods that are
examined above, we would expect that the value of a wetland is determined by
the socio-economic characteristics of its location. Information regarding per
capita income of the relevant population using each wetland was generally not
available in the valuation studies so we inputted this information from other
sources.6 Figure 4 also plots wetland value per hectare per year by GDP per
capita. As expected there appears to be a positive relationship between the
two. The coefficient on GDP per capita in the simple regression is significant at
the 10% level. We also included population density information for each
wetland site in the database in order to examine the influence of population
237
THE EMPIRICS OF WETLAND VALUATION
1000000.0
y = 1170.4x -0.1996
(1995 US$ ha yr ; log scale)
R2 = 0.0699
100000.0
10000.0
Wetland value
1000.0
-1
-1
100.0
10.0
1.0
0.1
1 10 100 1000 10000 100000 1000000 10000000
Wetland area (ha; log scale)
y = 65.586e6E-05x
1000000.0
(1995 US$ ha yr ; log scale)
R 2 = 0.0824
100000.0
10000.0
Wetland value
-1
1000.0
-1
100.0
10.0
1.0
0.1
0 10000 20000 30000 40000
GDP per capita (1995 US$)
1000000.0
y = 61.483x0.2412
R2 = 0.0204
(1995 US$ ha yr ; log scale)
100000.0
10000.0
Wetland value
-1
1000.0
-1
100.0
10.0
1.0
0.1
1 10 100 1000 10000
Population density ( in 1000s km-2; log scale)
Figure 4. Wetland value per hectare per year plotted against wetland area, GDP per
capita, and population density.
density on wetland values.7 Our expectation is that wetlands have a higher
value in areas with higher population density as most wetland services are
related to direct or indirect human use. The spatial relationship between
wetlands and centers of population is of course important in determining the
use made of wetland services. This spatial relationship will vary with a number
of factors including wetland service, transportation availability, physical
barriers, and cultural norms. For example, in the case of the recreational use
of a wetland we would expect that in the US the distance between a population
238 LUKE M. BRANDER ET AL.
center and the wetland would be of less importance than in a developing
country due to transportation availability and habits. Consequentially, the
‘‘catchment’’ area of population that might use a wetland would be much larger
in the US. We were not able to capture all of these considerations in the data but
use population densities for 50-km radius zones around each wetland site. Fig-
ure 4 plots wetland value by population density. There is an apparent positive
relationship between the two, although the estimated coefficient on population
density in the trend line equation is not significant.
4. Meta-Regression
The above exploratory analysis of the available data in the wetland valuation
literature does of course not allow for interactions between the various
explanatory variables. In order to attain marginal effects – given the inter-
ference of potentially relevant intervening characteristics – we use meta-
regression analysis to assess the relative importance of all potentially relevant
factors simultaneously. The dependent variable in our regression equation is
a vector of values in US$ per hectare per year in 1995 prices, labelled y. The
explanatory variables are grouped in three different matrices that include the
study characteristics in Xs (i.e., valuation method, marginal value), the wet-
land physical and geographical characteristics in Xp (i.e., wetland type, ser-
vices, area, urban, continent, latitude, and RAMSAR proportion), and the
socio-economic characteristics in Xe (i.e., GDP per capita, and population
density). The model fit was considerably improved, and heteroskedasticity
was mitigated, by using the logarithms of the dependent variable, GDP per
capita, population density, and wetland size. The estimated model is, in
matrix notation:
lnðyÞ ¼ a þ Xs bs þ Xp bp þ Xe be þ u
where a is the usual constant term, u a vector of residuals (assuming well
behaved underlying errors), and the vectors b contain the estimated coeffi-
cients on the respective explanatory variables.8 The regression results are
presented in Table III, using White-adjusted standard errors because the
Breusch–Pagan test still indicates that the model is heteroskedastic. Multi-
collinearity was tested for and judged not to be a serious problem.9 The
adjusted R2 value of 0.45 is reasonably high, and indicates that close to half
the variation in wetland value is explained by variation in our explanatory
variables. In this (largely) semi-log model, the coefficients measure the con-
stant proportional or relative change in the dependent variable for a given
absolute change in the value of the explanatory variable. For the explanatory
variables expressed as logarithms, the coefficients should be interpreted as
elasticities, that is, the percentage change in the dependent variable given a
(small) percentage change in the explanatory variable.
239
THE EMPIRICS OF WETLAND VALUATION
Table III. Meta-regression resultsa
Variableb
Category Coefficient Standard error
)6.98
Constant 4.67
1.16**
Socio-economic GDP per capita 0.46
(log)
0.47***
Population 0.12
density (log)
)0.11**
Geographic Wetland size 0.05
characteristics (log)
Latitude 0.03 0.07
(absolute value)
)0.0007
Latitude squared 0.0010
South America 0.23 1.19
Europe 0.84 0.92
Asia 2.01 1.34
3.51**
Africa 1.52
1.75*
Australasia 0.94
1.11**
Urban 0.48
1.49**
Valuation CVM 0.73
)0.71
methods Hedonic pricing 1.54
TCM 0.01 0.65
Replacement cost 0.63 0.81
Net factor income 0.19 0.61
)1.00
Production function 0.75
)0.04
Market prices 0.53
)0.03
Opportunity cost 0.72
0.95*
Type value Marginal 0.48
)0.56
Wetland type Mangrove 0.82
Unvegetated sediment 0.22 1.09
)0.31
Salt/brackish marsh 0.42
)1.46**
Fresh marsh 0.59
0.86**
Woodland 0.42
Wetland service Flood control 0.14 0.55
)0.95
Water supply 0.71
Water quality 0.63 0.74
)0.03
Habitat and nursery 0.35
)1.10**
Hunting 0.43
Fishing 0.06 0.36
)0.83**
Material 0.42
)1.24***
Fuelwood 0.45
Amenity 0.06 0.39
Biodiversity 0.06 0.81
)1.32*
RAMSAR RAMSAR proportion 0.70
202
n
240 LUKE M. BRANDER ET AL.
Table III. Continued
Variableb
Category Coefficient Standard error
R2-adjusted 0.45
5.50***
F
51.46***
Breusch–Pagan
a
OLS results with White-adjusted standard errors. The Breusch–Pagan test concerns heter-
oskedasticity and is v2 distributed with 36 degrees of freedom. Significance is indicated with
***, **, and * for the 1, 5, and 10% level, respectively.
b
The valuation methods, wetland types and wetland services are not strictly non-overlapping
variables. In other words, some wetlands provide more than one service, and comprise smaller
areas of different types. There is also not a one-to-one correspondence between an observed
value and the use of a specific valuation method. Consequently, there is no need for the
omission of one of the categories in order to avoid perfect collinearity.
Regarding the influence of wetland type on the wetland value, differences in
value associated with different wetland types are indicated by the coefficients
on these dummy variables. Two of these coefficients are significantly different
from zero suggesting that fresh marshes have the lowest value as compared to
the average and woodland wetlands have the highest value. This result was not
apparent from Figure 3, but the latter result was merely based on bivariate
comparisons.
On the issue of whether wetlands exhibit increasing or decreasing returns
to scale, the coefficient on the wetland size variable is small and negative, as
well as significant. This suggests that there are significant decreasing returns
to scale. We correct for the fact that marginal values may be significantly
different from average values, which is shown to be the case. Specifically,
marginal values are almost twice as high as compared to average values. The
decreasing return result confirms the findings of Woodward and Wui (2001),
who observe decreasing returns to scale for wetlands at the level of )0.17 and
)0.29 for a comparable function. It should be noted that the double-log
specification induces the returns to scale to decline geometrically with size
(see Woodward and Wui 2001, pp. 267–268), so that the elasticity approaches
zero with increasing size.
The differences in wetland values resulting from the availability of dif-
ferent services were touched on above. Wetland services that involve the
provision of direct use natural resources, such as fuel wood and other
materials tend to have lower than average values. Somewhat surprisingly,
wetlands that provide recreational hunting opportunities also tend to have
lower than average values.
Another unexpected result given the initial analysis of Figure 3 is that
North American wetlands tend to have lower values than wetlands located in
241
THE EMPIRICS OF WETLAND VALUATION
other continents. North America is included in the constant term of our
model and the coefficients on the dummy variables indicating the other
continents are all positive but only significant for Africa and Australasia. We
can only speculate on the reason for this. One possible explanation for
wetlands receiving a lower value in North America is the relative abundance
of substitute natural areas, particularly in comparison with Europe. One
should note that these results are obtained correcting for differences in
latitude. We hypothesized that the value of wetlands might be related non-linearly
(following a parabolic shape) to the absolute distance from the equator. This,
however, is not apparent in the estimation results.
For the socio-economic variables that we were able to include in our
model, the results confirm our expectations. The coefficient on the GDP
per capita variable is positive and highly significant – suggesting a slightly
elastic effect of income on the value of wetland services. The interpretation
of the result is that a 10% increase in GDP per capita results in roughly a
12% increase in wetland value. There is also a positive and significant
relationship between population density and wetland value as described
above. This relationship, however, is inelastic, but this may very well be
due to the dummy variable ‘‘Urban’’ that is also included. On average,
urban wetlands have a value that is significantly higher than rural wet-
lands.
The results for the valuation methodology dummy variables show that
value estimates from contingent valuation, replacement cost, the travel cost
method and NFI methods are higher than estimates from other valuation
methods. The only statistically significant result, however, pertains to CVM,
which show the highest values as compared to the other valuation methods.
This result is in contrast with the findings of Woodward and Wui (2001), who
observed that the hedonic pricing and the replacement cost method produce
higher values than CVM.
Finally, the variable referring to a comparison between RAMSAR and
other sites shows an interesting outcome. We use a variable operational-
ized as the proportion of the wetland that is designated a RAMSAR site,
and it shows up indicating significantly lower values for RAMSAR sites.
Possible explanations for this result might be that certain uses of these
wetlands are restricted and therefore not valued, or that WTP bids for the
conservation of these sites are affected by respondents’ knowledge that
they are already protected.
5. Value Transfer
There remains the question of whether the results from this meta-analysis can
be used for value transfer, that is, the prediction or estimation of the value of
a wetland given knowledge of its physical and socio-economic characteristics.
242 LUKE M. BRANDER ET AL.
There is substantial academic and policy interest in the potential for and
validity of value transfer as it offers a means of estimating monetary values
for environmental resources without performing relatively time consuming
and expensive primary valuation studies (see Florax et al. 2002).
There are two general approaches to value transfer: direct value transfer
and function value transfer. The first involves simply transferring the value(s)
estimated in one or more primary studies to the policy site in question.
Ideally, the study site and policy site should be similar in their characteristics
or adjustments should be made to the transferred value to reflect differences
in site characteristics (Brouwer 2000). The second approach involves transferring
values to a policy site based on its known characteristics using a value transfer
function, possibly estimated through a meta-regression. Rosenberger and Phipps
(2001) identify the important assumptions underlying the use of a value function
for value transfer:
(1) there exists a valuation function that links the values of a resource with
the characteristics of the relevant markets and sites across space and
time, and from which values for specific characteristics can be inferred,
(2) differences between sites can be captured through a price vector,
(3) values are stable over time, or vary in a systematic way, and
(4) the sampled primary valuation studies provide ‘‘correct’’ estimates of
resource value.
It is generally accepted that function transfers perform better than direct
value transfers for a number of reasons. Firstly, information from a larger
number of studies is used. Secondly, methodological differences between
primary valuation studies can be controlled for. And thirdly, explanatory
variables can be adjusted to represent the policy site (Bateman and Jones
2003). Rosenberger and Phipps (2001) review a number of studies that test
the relative performance of direct value transfer and function value transfer
(see for example Loomis 1992; Parsons and Kealy 1994; Brouwer and
Spaninks 1999). The general conclusion is that meta-analysis value transfer
functions perform better than other approaches (see also Engel, 2002).
For a number of reasons value transfer may result in substantial
‘transfer errors’, particularly when the characteristics of the site to which
values are being transferred are not well represented in the data underlying
the estimated value function (Brouwer 2000). Another reason for error
might be that the characterization of wetland types and wetland services is
oftentimes rather crude. The use of dummy variables to characterize types
and services does not capture the true variation in these characteristics.
Similarly, it is difficult to capture important quality and quantity differ-
ences in provision of services across sites. Unfortunately, we cannot
overcome this in our value transfer experiment, but we have instead
focused on including non-sample information such as GDP per capita and
243
THE EMPIRICS OF WETLAND VALUATION
population density. To the extent that such information is relevant, it may
increase the accuracy of value transfer estimates. We have also attempted
to be as comprehensive as possible in our collection of valuation studies
but clearly due to the limited availability of studies, our sample focuses on
certain wetland types, services and locations. This raises the question of
the validity of value transfer to policy sites in countries that are not
represented in the data (Shrestha and Loomis 2001). Another important
source of transfer error is related to errors in the primary valuation studies
from which the value transfer function is derived (Woodward and Wui
2001). As described in Section 2, there are a number of biases associated
with each valuation method that may result in mis-estimation of ‘‘true’’
resource values and thereby introduce a source of error in estimating a
value function.
As a first step, before actually performing a value transfer, we looked at
the in-sample forecast performance of our model. As an indicator we used the
Mean Absolute Percentage Error (MAPE), which is defined as |(yobs)yest)/
yobs|. For our sample of 201 observations10 the average MAPE equals 58%,
which is a rather high forecast error. However, if we look at the average
MAPE for different quartiles of the data series ordering them by magnitude
of the wetland value, in ascending order, we find an average MAPE of 173,
26, 16, and 19, respectively. This indicates that the fit for low wetland values
is particularly poor and the fit for medium to high-valued wetlands is much
more acceptable.
Subsequently, we use an n)1 data splitting technique to estimate 200 value
transfer functions by applying the estimated parameters generated with n)1
observations to the omitted observation.11 The upper panel in Figure 5
presents the observed and predicted wetland values in ascending order of
observed value. This shows that for a number of observations there is a
considerable difference between observed and predicted values. It also indi-
cates that our value transfer function systematically over-predicts very low
wetland values and slightly under-predicts high values.12 The lower panel in
Figure 5 presents the transfer error (defined as MAPE) associated with each
observation ranked in ascending order. The overall average transfer error is
74%, which is slightly higher than the in-sample forecast error – as can be
expected. The average transfer error for different quartiles of the data series
ordered by magnitude of the wetland value, in ascending order, is 213, 34, 19,
and 33, respectively. Slightly less than 20% of the sample has transfer errors
of 10% or less, and roughly 15% of observations show transfer errors over
100%. In comparison to other value transfer exercises (reviewed in Brouwer
2000; Rosenberger and Phipps 2001) our results appear to be similar despite
the relative diversity of our data in terms of the activities and services being
valued, valuation techniques employed, geographic locations and socio-
economic characteristics. One advantage, however, of our value transfer is
244 LUKE M. BRANDER ET AL.
16
Yobs
Ypred
12
lnvalue (US$ha-1y-1)
8
4
0
-4
Observations
1500
Transfer error
1000
500
0
Observations
Figure 5. Observed and predicted wetland values and transfer errors, ranked in
ascending order of observed wetland value.
that we have been able to include important population density and income
variables that have not been included in most other studies.
An important proviso for the validity of value transfer is that sites for
which a transfer is being conducted, and the method on which the valuation
is based, are adequately represented in the meta-dataset (see Rosenberger and
Phipps 2001). It is, however, not enough to merely count sites with specific
characteristics and the number of studies using particular methods. In this
case a multivariate analysis is preferable as well. We therefore regressed the
exogenous variables distinguished in the meta-analysis on the transfer errors,
defined as the mean absolute percentage error. This analysis shows that the
transfer errors are significantly negatively correlated with the dummies for
Africa, Asia, and Australasia. This result contradicts the suggestion of
Rosenberger and Phipps (2001) that the accuracy of value transfer is directly
related to the incidence of specific characteristics in the meta-database,
because most of our observations are from North America. There is also a
significant positive correlation for the replacement cost method and a similar
negative correlation with net factor income methods. Finally, the transfer
245
THE EMPIRICS OF WETLAND VALUATION
errors are also significantly positively correlated with the variable measuring
the proportion of the wetland that is under the RAMSAR convention.
6. Conclusions
This article provides a comprehensive overview of the wetland valuation lit-
erature and has attempted to identify the important physical, socio-economic
and study characteristics that determine wetland value. The wetland valuation
literature has been shown to be extremely diverse in terms of values estimated,
wetland types considered and valuation methods used. The value estimates
produced by different valuation methodologies are not necessarily directly
comparable and need to be explicitly modeled in our meta-regression. One of
the key results from our meta-regression analysis is the importance of the
GDP per capita and population density variables in explaining variation in
wetland value. Both variables were shown to have a positive relationship with
wetland value. Although such information is often not available in primary
valuation studies it is suggested that future valuation meta-analyses attempt
to include relevant socio-economic information from other sources in order to
represent important determinants of value. Another interesting result is that
CVM studies have tended, ceteris paribus, to produce higher value estimates
than other valuation methods. This contrasts with our expectations and with
the findings of Woodward and Wui (2001). In terms of the ecological and
physical characteristics of wetlands, we found freshwater marshes to be valued
less than other wetland types and a negative relationship between wetland size
and value. Of the various wetland services that we identified, water quality
improvement was found to be valued the highest. Two unexpected results
from this meta-analysis were that North American wetlands and RAMSAR
sites were found to be valued lower than other wetlands.
Using an n)1 data splitting technique we examined the robustness of using
our meta-regression for out-of-sample value transfer. The resulting average
transfer error is 74%, which is comparable to the transfer errors associated
with other value transfer exercises in the literature. Given the high costs of
performing primary valuation studies, this level of transfer error may be
acceptable in considering transferred values as input in wetland conservation
decisions. However, our value transfer function systematically over-predicts
very low wetland values and slightly under-predicts high values. Remarkably
enough, the value transfer performs better for wetlands that are located in
countries not well represented in our data (Africa, Asia, and Australasia).
The value transfer error is positively correlated with transfers based on the
replacement cost method. The same result holds for the degree to which
wetlands are RAMSAR sites. We therefore urge caution in using the
results of such a meta-analysis for value transfer, particularly to policy sites
for which their characteristics are not well represented in the underlying
246 LUKE M. BRANDER ET AL.
valuation studies. There is clearly still a need for more (and higher quality)
primary valuation studies, particularly in developing countries.
Acknowledgements
This research has, in part, been carried out within the project DINAS-
COAST (Dynamic and Interactive Assessment of National, Regional and
Global Vulnerability of Coastal Zones to Climate Change and Sea-Level
Rise), which is funded by the EU Directorate-General Research under pro-
ject number EVK2-CT-2000–00084. The authors would like to thank the
participants of the International Colloquium ‘‘Meta-Analysis in Economics,’’
held in Amsterdam in December 2002, for useful comments and suggestions.
Notes
1. Carson et al. (1996) review 83 valuation studies for quasi-public goods from which 616
comparisons of contingent valuation (CV) and revealed preference (RP) estimates are
made. The sample mean CV/RP ratio is 0.89 with a 95% confidence interval of 0.81–0.96
and a median of 0.75. While the results from this study show that RP methods produces
higher value estimates than CV, it also shows that estimates from these two methods are
within the same range.
2. Bateman and Jones (2003) set out expectations for the ordering of values estimated
through different analytical modes of the contingent valuation and travel cost methods.
3. One particularly useful source was an annotated review of the wetland valuation literature
for the period 1988–1998 by Bardecki (1998).
4. In order to compare observations in a statistical meta-analysis we required sufficient
information on a number of key variables. These are: wetland value, area, type,
function(s) being valued, location, year of valuation and valuation method used.
5. We used GDP deflators and purchasing power parity converters from the World Bank
World Development Indicators 2002 to standardize values estimated in different years and
different currencies.
6. The 1995 national per capita income level data were taken from World Bank World
Development Indicators 2000, and US state level data were taken from the US Census
2000 for the US states.
7. The population densities included in our analysis represent an area of 50-km radius around
each wetland site. Population and population density information was derived from CIESIN
data. The process by which this data was converted to represent each wetland site in our data
set is described in Wagtendonk and Omtzigt (2003).
8. A multi-level modelling (MLM) approach such as used in Brouwer et al. (1999), and
Bateman and Jones (2003) was considered but not adopted. This approach incorporates
natural hierarchies or levels within the data, e.g., study sites, author, method and study,
allowing the (somewhat unrealistic) assumption of independence between estimates to be
relaxed. MLM is, however, problematic in that it requires the use of dummy variables
for each group within a level, e.g., for each author or study site. This may be feasible in
reasonably limited or homogeneous data sets but less so for very diverse data (such as
ours).
247
THE EMPIRICS OF WETLAND VALUATION
9. The presence of multicollinearity was tested by examining the correlation coefficients on
pairs of explanatory variables and by regressing selected explanatory variables on the
remaining variables.
10. For this exercise we used a slightly restricted data set as one observation for which the log
value was very close to zero would otherwise have a disproportionate influence.
11. This resembles the jackknife resampling technique (see Efron 1982).
12. Again it should be noted that the primary valuation studies are highly unlikely to have
produced ‘‘true’’ estimates of wetland value and therefore we would not necessarily expect
(or want) transferred values to be exactly equal to primary valuation study results.
References
Acharya, G. (2000), ‘Approaches to Valuing the Hidden Hydrological Services of Wetland
Ecosystems’, Ecological Economics 35, 63–74.
Acharya, G. and E. B. Barbier (2000), ‘Valuing Ground Water Recharge through Agricultural
Production in the Hadejia-Nguru Wetlands in Northern Nigeria’, Agricultural Economics
22, 247–259.
Amacher, G. S., R. J. Brazee, J. W. Bulkley and R. A. Moll (1989), Application of Wetland
Valuation Techniques: Examples from Great Lakes Coastal Wetlands. Ann Arbor, MI:
University of Michigan, School of Natural Resources.
Anderson, R. and M. Rockel (1991), ‘Economic Valuation of Wetlands’, American Petroleum
Institute, Discussion Paper No. 65, Washington D.C.
Bann, C. (1997), An Economic Analysis of Alternative Mangrove Management Strategies in Koh
Kong Province, Cambodia. Singapore: EEPSEA research report series.
Batie, S. S. and L. Shabman (1979), ‘Valuing Non-Market Goods: Conceptual and Empirical
Issues’, American Journal of Agricultural Economics 61, 931–932.
Bateman, I. and I. H. Langford (1997), ‘Non-Users Willingness to Pay for a National Park:
An Application of the Contingent Valuation Method’, Regional Studies 31, 571–582.
Bateman, I. and L. P. Jones (2003), ‘Contrasting Conventional with Multi-level Modeling
Approaches to Meta-analysis: Expectation Consistency in U.K. Woodland Recreation
Values’, Land Economics 79(2), 235–258.
Bateman, I.J., A.A. Lovett and J.S. Brainard (2004), Applied Environmental Economics: A GIS
Approach to Cost-Benefit Analysis. Cambridge University Press.
Barbier, E. B., (1991), ‘An Approach to Economic Evaluation of Tropical Wetlands: With
Examples from Guatemala and Nicaragua’, N. P. Girvan and D. Simons, eds., Caribbean
Ecology and Economics, Caribbean Conservation Association, St. Michael, Barbados, 207–231.
Barbier, E. B., M. Acreman and D. Knowler (1997), Economic Valuation of Wetlands: A Guide
for Policy Makers and Planners. Gland, Switzerland.: Ramsar Convention Bureau.
Barbier, E. B. and I. Strand (1998), ‘Valuing Mangrove-Fishery Linkages’, Environmental and
Resource Economics 12, 151–166.
Barbier, E. B. and J. R. Thompson (1998), ‘The Value of Water: Floodplain versus Large-scale
Irrigation Benefits in Northern Nigeria’, Ambio 26, 434–440.
Bardecki, M. J. (1998), Wetlands and Economics: An Annotated Review of the Literature, 1988–
1998. Ontario: Environment Canada.
Bell, F. W. (1997), ‘The Economic Valuation of Saltwater Marsh Supporting Marine
Recreational Fishing in the Southeastern United States’, Ecological Economics 21, 243–254.
Bergstrom, J. C., J. R. Stoll, J. P. Titre and V. L. Wright (1990), ‘Economic Valuation of
Wetlands-based Recreation’, Ecological Economics 2, 129–147.
Bergstrom, J. C. and J. R. Stoll (1993), ‘Value Estimator Models for Wetlands-based
Recreational use Values’, Land Economics 69, 132–137.
248 LUKE M. BRANDER ET AL.
Blomquist, G. C. and J. C. Whitehead (1998), ‘Resource Quality Information and Validity of
Willingness to Pay in Contingent Valuation’, Resource and Energy Economics 20, 179–196.
Breaux, A., S. C. Farber and J. Day (1995), ‘Using Natural Coastal Wetlands Systems for
Wastewater Treatment: An Economic Benefit Analysis’, Journal of Environmental
Management 44, 285–291.
Brouwer, R. (2000), ‘Environmental Value Transfer: State of the Art and Future Prospects’,
Ecological Economics 32, 137–152.
Brouwer, R., I. H. Langford, I. J. Bateman, T. C. Crowards and R. K. Turner (1999), ‘A Meta-
Analysis of Wetland Contingent Valuation Studies’, Regional Environmental Change 1, 47–57.
Brouwer, R. and F. A. Spaninks (1999), ‘The Validity of Environmental Benefits Transfer:
Further Empirical Testing’, Environmental and Resource Economics 14, 95–117.
Bystrom, O., H. Andersson and I.-M. Gren (2000), ‘Economic Criteria for Using Wetlands as
Nitrogen Sinks under Uncertainty’, Ecological Economics 35, 35–45.
Carson, R. T., N. E. Flores, K. M. Martin and J. L. Wright (1996), ‘Contingent Valuation and
Revealed Preference Methodologies: Comparing the Estimates for Quasi-Public Goods’,
Land Economics 72(Feb), 80–99.
Cooper, J. and J. Loomis (1991), ÔEconomic Value of Wildlife Resources in the San Joaquin Valley:
Hunting and Fishing ValuesÕ, in A. Dinar and D. Zilberman, eds., The Economics and
Management of Water and Drainage in Agriculture. Norwell Mass: Kluwer Academic
Publishers.
Cooper, J. and J. Loomis (1993), ‘Testing whether Waterfowl Hunting Benefits Increase with
Greater Water Deliveries to Wetlands’, Environmental and Resource Economics 3, 545–561.
Costanza, R., S. C. Farber and J. Maxwell (1989), ‘Valuation and Management of Wetland
Ecosystems’, Ecological Economics 1, 335–361.
Creel, M. and J. Loomis (1992), ‘Recreation Value of Water to Wetlands in the San Joaquin
Valley: Linked Multinomial Logit and Count Data Trip Frequency Models’, Water
Resources Research 28, 2597–2606.
Cummings, R. G. and G. W. Harrison (1995), ‘The Measurement and Decomposition of Non-
use Values: A Critical Review’, Environmental and Resource Economics 5, 225–247.
Dahuri, R. (1991), An Approach to Coastal Resource Utilisation: The Nature and Role of
Sustainable Development in East Kalimantan Coastal Zone, Indonesia. Halifax, Canada:
Dalhousie University PhD Thesis.
Dharmaratne, G. and I. Strand (2002), Adaptation to Climate Change in the Caribbean: The
Role of Economic Valuation. Report to the CPACC, London.
Dalecki, M. G., J. C. Whitehead and G. C. Blomquist (1993), ‘Sample Non-Response Bias and
Aggregate Benefits in Contingent Valuation: An Examination of Early, Late and Non-
Respondents’, Journal of Environmental Management 38, 133–143.
Dixon, J. A. and P. N. Lal (1997). ‘The Management of Coastal Wetlands: Economic Analysis
of Combined Ecologic-Economic Systems’, in P. Dasgupta and K.-G. Maler, eds., The
¨
Environment and Emerging Development Issues, Vol 2. Oxford University Press.
Doss, C. R. and S. J. Taff (1996), ‘The Influence of Wetland Type and Wetland Proximity on
Residential Property Values’, Journal of Agricultural and Resource Economics 21, 120–129.
Downing, M. and T. Ozuna (1996), ‘Testing the Reliability of the Benefit Function Transfer
Approach’, Journal of Environmental Economics and Management 30, 316–322.
Efron, B. (1982), The Jackknife, the Bootstrap and other Resampling Plans. Philadelphia:
Society for Industrial and Applied Mathematics.
Ellis, G. M. and A. C. Fisher (1987), ‘Valuing the Environment as Input’, Journal of
Environmental Management 25, 149–156.
Emerton, L., L. Iyango, P. Luwum and A. Malinga (1998), The Present Economic Value of
Nakivubo Urban Wetland, Uganda. Nairobi, Kenya: IUCN Report.
249
THE EMPIRICS OF WETLAND VALUATION
Emerton, L. and B. Kekulandala (2002). ‘Assessment of the Economic Value of Muthura-
jawela Wetland’, IUCN – The World Conservation Union, Sri Lanka Country Office.
Engel, S. (2002), ÔBenefit Function Transfer versus Meta-Analysis as Policy-making Tools: A
ComparisonÕ, in R. J. G. M. Florax, P. Nijkamp and K. G. Willis, eds., Comparative
Environmental Economic Assessment. (pp. 133–153). Edward Elgar: Cheltenham.
Farber, S. (1988), ‘Non-user’s WTP for a National Park: An Application and Critique of the
Contingent Valuation Method’, Regional Studies 31, 571–582.
Farber, S. (1992), ‘The Economic Cost of Residential Environmental Risk: A Case Study of
Louisiana’, Journal of Environmental Management 36, 1–16.
Farber, S. and R. Costanza (1987), ‘The Economic Value of Wetland Systems’, Journal of
Environmental Management 24, 41–51.
Florax, R. J. G. M., P. Nijkamp and K. G. Willis (eds.) (2002), Comparative Environmental
Economic Assessment. Edward Elgar: Cheltenham.
Freeman III, A. M. (1991), ‘Valuing Environmental Resources under Alternative Manage-
ment Regimes’, Ecological Economics 3, 247–256.
Freeman III, A. M. (2003), The Measurement of Environmental and Resource Values: Theories
and Methods. Washington D.C.: Resources for the Future.
Gren, I.-M. and T. Soderqvist (1994), Economic Valuation of Wetlands: A Survey. Stockholm:
Sweden, Beijer Discussion Paper Series No. 54.
Gren, I.-M., C. Folke, K. Turner and I. Bateman (1994), ‘Primary and Secondary Values of
Wetland Ecosystems’, Environmental and Resource Economics 4, 55–74.
Gren, I.-M. (1995), ‘The Value of Investing in Wetlands for Nitrogen Abatement’, European
Review of Agricultural Economics 22, 157–172.
Haab, T.C. and K.E. McConnell (1997). ‘A Simple Method for Bounding WTP using a Probit
or Logit Model’. Economic Working Paper 9713, East Carolina University.
Hammack, J. and G. M. Brown (1974), Waterfowl and Wetlands: Toward A Bio-Economic
Analysis. Washington, D.C.: Resources for the Future, John Hopkins University Press.
Janssen, R. and J. E. Padilla (1999), ‘Preservation of Conversion? Valuation and Evaluation
of a Mangrove Forest in the Philippines’, Environmental and Resource Economics 14,
297–331.
Kazmierczak, R. F. (2001), Economic Linkages between Coastal Wetlands and Hunting and
Fishing: A Review of Value Estimates Reported in the Published Literature. Baton Rouge:
Louisiana State University Agricultural Center, Staff Paper 2001 – 03.
Klein, R. J. T. and I. J. Bateman (1998), ‘The Recreational Value of Cley Marshes Nature
Reserve: An Argument Against Managed Retreat?’, Journal of the Chartered Institution of
Water and Environmental Management 12, 280–285.
Kopp, R. J., V. K. Smith (eds.) (1993), Valuing Natural Assets, The Economics of Natural
Resource Damage Assessment. Washington, D.C.: Resources for the Future.
Lant, C. L. and R. S. Roberts (1990), ‘Greenbelts in the Cornbelt: Riparian Wetlands,
Intrinsic Values and Market Failure’, Environment and Planning 22, 1375–1388.
Larson, J. S., P. R. Adamus and E. J. Clairain Jr (1989), ‘Functional Assessment of
Freshwater Wetlands: A Manual and Training Outline’, Glaud, Switzerland: WWF
Publication 89–6:62 pp.
Leitch, J. A. and B. Hovde (1996), ‘Empirical Valuation of Prairie Potholes: Five Case
Studies’, Great Plains Research 6, 25–39.
Loomis, J. B. (1992), ‘The Evolution of a More Rigorous Approach to Benefit Transfer:
Benefit Function Transfer’, Water Resources Research 28(3), 701–705.
Lupi, F., T. Graham-Tomasi and S. J. Taff (1991), ‘A Hedonic Approach to Urban Wetland
Valuation’. Staff Paper P91-8, Department of Agricultural and Applied Economics,
University of Minnesota.
250 LUKE M. BRANDER ET AL.
Parsons, G. R. and M. J. Kealy (1994), ‘Benefit Transfer in a Random Utility Model of
Recreation’, Water Resources Research 30(8), 2477–2484.
Pate, J. and J. Loomis (1997), ‘The Effect of Distance on Willingness to Pay Values: A Case
Study of Wetlands and Salmon in California’, Ecological Economics 20, 199–207.
Pearce, D. and R. K. Turner (1990), Economics of Natural Resources and the Environment.
Hemel Hempstead, UK: Harvester Wheatsheaf.
Ramdial, B. S. (1975), ‘The Social and Economic Importance of the Caroni Swamp in
Trinidad and Tobago’. PhD thesis, University of Michigan.
Raphael, C. N. and E. Jaworski (1979), ‘Economic Value of Fish, Wildlife, and Recreation in
Michigan’s Coastal Wetlands’, Coastal Zone Management Journal 5, 181–194.
Rosenberger, R. and J. B. Loomis (2000), ‘Using Meta-Analysis For Benefit Transfer: In
Sample Convergent Validity Tests of an Outdoor Recreation Database’, Water Resources
Research 36, 1097–1107.
Rosenberger, R. and T. T. Phipps (2001). ‘Site correspondence effects in benefit transfers: A
meta-analysis transfer function’. Research Paper 2001–6, West Virginia University.
Rosenthal, R. and M. R. DiMatteo (2001), ‘Meta-Analysis: Recent Developments in
Quantitative Methods for Literature Reviews’, Annual Review of Psychology 52, 59–82.
Ruitenbeek, H. J. (1992), ‘Mangrove Management: An Economic Analysis of Management
Options with a Focus on Bintuni Bay, Irian Jaya’. Environmental Management
Development in Indonesia Project, EMDI Environmental Report 8, Jakarta, Indonesia.
Sathirathai, S. and E. B. Barbier (2001), ‘Valuing Mangrove Conservation in Southern
Thailand’, Contemporary Economic Policy 19, 109–122.
Schuijt, K. (2004), Land and Water Use of Wetlands in Africa: Economic Values of African
Wetlands. Laxenburg, Austria: International Institute for Applied Systems Analysis,
Report No. IR-02 – 063.
Shaw, W. D. and M. T. Ozog (1999), ‘Modeling Overnight Recreation Trip Choice:
Application of a Repeated Nested Multinomial Logit Model’, Environmental and Resource
Economics 13, 397–414.
Shrestha, R. K. and J. B. Loomis (2001), ‘Testing a Meta-Analysis Model for Benefit Transfer
in International Outdoor Recreation’, Ecological Economics 39, 67–83.
Smith, V. K. and K. Pattanayak (2002), ‘Is Meta-Analysis A Noah’s Ark for Non-Market
Valuation?’, Environmental and Resource Economics 22, 271–96.
Stanley, T. D. (2001), ‘Wheat from Chaff: Meta-Analysis as Quantitative Literature Review’,
Journal of Economic Perspectives 15, 131–50.
Turner, R. K., J. C. J. M. van den Bergh, T. Soderqvist, A. Barendregt, J. van der Straaten,
E. Maltby and E. C. van Ierland (2000), ‘Ecological-Economic Analysis of Wetlands:
Scientific Integration for Management and Policy’, Ecological Economics 35, 7–23.
Van den Bergh, J., A. Barendregt, A. Gilbert, M. van Herwijnedn, P. van Horssen,
P. Kandelaars and C. Lorenz (2001), ‘Spatial Economic-Hydrological Modeling and
Evaluation of Land Use Impacts in the Vecht Wetlands Area’, Environmental Modeling
and Assessment 6, 87–100.
Wagtendonk, A. and N. Omtzigt (2003), Analysis of Population Density Around Wetland Areas
in GPW (Gridded Population of the World) Files. Amsterdam: SPINlab document,
Institute for Environmental Studies.
Willig, R. D. (1976), ‘Consumers’ Surplus without An Apology’, American Economic Review
66, 589–597.
Woodward, R. T. and Y.-S. Wui (2001), ‘The Economic Value of Wetland Services: A
Meta-Analysis’, Ecological Economics 37, 257–270.