Nature: Volume 387 no 6230

The Value of the World's Ecosystem
Services and Natural Capital

Robert Costanza*†, Ralph d'Arge‡, Rudolf de Groot, Stephen Farber¶,
Monica Grasso†, Bruce Hannon, Karin Limburg#, Shahid Naeem**,
Robert V. O'Neill††, Jose Paruelo‡‡, Robert G. Raskin§§,
Paul Sutton¶¶ & Marjan van den Belt

* Center for Environmental and Estuarine Studies, Zoology Department, and † Insitute for
Ecological Economics, University of Maryland, Box 38, Solomons, Maryland 20688, USA
‡ Economics Department (emeritus), University of Wyoming, Laramie, Wyoming 82070,
§ Center for Environment and Climate Studies, Wageningen Agricultural University, PO
Box 9101, 6700 HB Wageninengen, The Netherlands
¶ Graduate School of Public and International Affairs, University of Pittsburgh, Pittsburgh,
Pennsylvania 15260, USA
Geography Department and NCSA, University of Illinois, Urbana, Illinois 61801, USA
# Institute of Ecosystem Studies, Millbrook, New York, USA
** Department of Ecology, Evolution and Behavior, University of Minnesota, St Paul,
Minnesota 55108, USA
†† Environmental Sciences Division, Oak Ridge National Laboratory, Oak Ridge,
Tennessee 37831, USA
‡‡ Department of Ecology, Faculty of Agronomy, University of Buenos Aires, Av. San
Martin 4453, 1417 Buenos Aires, Argentina
§§ Jet Propulsion Laboratory, Pasadena, California 91109, USA
¶¶ National Center for Geographic Information and Analysis, Department of Geography,
University of California at Santa Barbara, Santa Barbara, California 93106, USA
Ecological Economics Research and Applications Inc., PO Box 1589, Solomons,
Maryland 20688, USA

The services of ecological systems and the natural capital stocks that
produce them are critical to the functioning of the Earth's life-support
system. They contribute to human welfare, both directly and indirectly,
and therefore represent part of the total economic value of the planet.
We have estimated the current economic value of 17 ecosystem services
for 16 biomes, based on published studies and a few original calculations.
For the entire biosphere, the value (most of which is outside the market)
is estimated to be in the range of US$16--54 trillion (1012) per year, with
an average of US$33 trillion per year. Because of the nature of the
uncertainties, this must be considered a minimum estimate. Global gross
national product total is around US$18 trillion per year.

Because ecosystem services are not fully 'captured' in commercial markets or
adequately quantified in terms comparable with economic services and
manufactured capital, they are often given too little weight in policy decisions.
This neglect may ultimately compromise the sustainability of humans in the
biosphere. The economies of the Earth would grind to a halt without the services
of ecological life- support systems, so in one sense their total value to the
economy is infinite. However, it can be instructive to estimate the 'incremental' or
'marginal' value of ecosystem services (the estimated rate of change of value
compared with changes in ecosystem services from their current levels). There
have been many studies in the past few decades aimed at estimating the value of
a wide variety of ecosystem services. We have gathered together this large (but
scattered) amount of information and present it here in a form useful for
ecologists, economists, policy makers and the general public. From this
synthesis, we have estimated values for ecosystem services per unit area by
biome, and then multiplied by the total area of each biome and summed over all
services and biomes.

Although we acknowledge that there are many conceptual and empirical
problems inherent in producing such an estimate, we think this exercise is
essential in order to: (1) make the range of potential values of the services of
ecosystems more apparent (2) establish at least a first approximation of the
relative magnitude of global ecosystem services; (3) set up a framework for their
further analysis; (4) point out those areas most in need of additional research;
and (5) stimulate additional research and debate. Most of the problems and
uncertainties we encountered indicate that our estimate represents a minimum
value, which would probably increase: (1) with additional effort in studying and
valuing a broader range of ecosystem services; (2) with the incorporation of
more realistic representations of ecosystem dynamics and interdependence; and
(3) as ecosystem services become more stressed and 'scarce' in the future.

Ecosystem functions and ecosystem services

Ecosystem functions refer variously to the habitat, biological or system properties
or processes of ecosystems. Ecosystem goods (such as food) and services (such
as waste assimilation) represent the benefits human populations derive, directly
or indirectly, from ecosystem functions. For simplicity, we will refer to ecosystem
goods and services together as ecosystem services. A large number of functions
and services can be identified1, 2, 3 4. Reference 5 provides a recent, detailed
compendium on describing, measuring and valuing ecosystem services. For the
purposes of this analysis we grouped ecosystem services into 17 major
categories. These groups are listed in Table 1. We included only renewable
ecosystem services, excluding non-renewable fuels and minerals and the
atmosphere. Note that ecosystem services and functions do not necessarily show
a one-to-one correspondence. In some cases a single ecosystem service is the
product of two or more ecosystem functions whereas in other cases a single
ecosystem function contributes to two or more ecosystem services. It is also
important to emphasize the interdependent nature of many ecosystem functions.
For example, some of the net primary production in an ecosystem ends up as
food, the consumption of which generates respiratory products necessary for
primary production. Even though these functions and services are
interdependent, in many cases they can be added because they represent 'joint
products' of the ecosystem, which support human welfare. To the extent
possible, we have attempted to distinguish joint and 'addable' products from
products that would represent 'double counting' (because they represent different
aspects of the same service) if they were added. It is also important to recognize
that a minimum level of ecosystem 'infrastructure' is necessary in order to allow
production of the range of services shown in Table 1. Several authors have
stressed the importance of this 'infrastructure' of the ecosystem itself as a
contributor to its total value6, 7. This component of the value is not included in the
current analysis.

Natural capital and ecosystem services

In general, capital is considered to be a stock of materials or information that
exists at a point in time. Each form of capital stock generates, either
autonomously or in conjunction with services from other capital stocks, a flow of
services that may be used to transform materials, or the spatial configuration of
materials, to enhance the welfare of humans. The human use of this flow of
services may or may not leave the original capital stock intact. Capital stock
takes different identifiable forms, most notably in physical forms including natural
capital, such as trees, minerals, ecosystems, the atmosphere and so on;
manufactured capital, such as machines and buildings; and the human capital of
physical bodies. In addition, capital stocks can take intangible forms, especially
as information such as that stored in computers and in individual human brains, as
well as that stored in species and ecosystems.

Ecosystem services consist of flows of materials, energy, and information from
natural capital stocks which combine with manufactured and human capital
services to produce human welfare. Although it is possible to imagine generating
human welfare without natural capital and ecosystem services in artificial 'space
colonies', this possibility is too remote and unlikely to be of much current interest.
In fact, one additional way to think about the value of ecosystem services is to
determine what it would cost to replicate them in a technologically produced,
artificial biosphere. Experience with manned space missions and with Biosphere
II in Arizona indicates that this is an exceedingly complex and expensive
proposition. Biosphere I (the Earth) is a very efficient, least-cost provider of
human life- support services.

Thus we can consider the general class of natural capital as essential to human
welfare. Zero natural capital implies zero human welfare because it is not feasible
to substitute, in total, purely 'non- natural' capital for natural capital.
Manufactured and human capital require natural capital for their construction7.
Therefore, it is not very meaningful to ask the total value of natural capital to
human welfare, nor to ask the value of massive, particular forms of natural
capital. It is trivial to ask what is the value of the atmosphere to humankind, or
what is the value of rocks and soil infrastructure as support systems. Their value
is infinite in total.

However, it is meaningful to ask how changes in the quantity or quality of various
types of natural capital and ecosystem services may have an impact on human
welfare. Such changes include both small changes at large scales and large
changes at small scales. For example, changing the gaseous composition of the
global atmosphere by a small amount may have large-scale climate change
effects that will affect the viability and welfare of global human populations.
Large changes at small scales include, for example, dramatically changing local
forest composition. These changes may dramatically alter terrestrial and aquatic
ecosystems, having an impact on the benefits and costs of local human activities.
In general, changes in particular forms of natural capital and ecosystem services
will alter the costs or benefits of maintaining human welfare.

Valuation of ecosystem services

The issue of valuation is inseparable from the choices and decisions we have to
make about ecological systems6,8. Some argue that valuation of ecosystems is
either impossible or unwise, that we cannot place a value on such 'intangibles' as
human life, environmental aesthetics, or long-term ecological benefits. But, in
fact, we do so every day. When we set construction standards for highways,
bridges and the like, we value human life (acknowledged or not) because
spending more money on construction would save lives. Another frequent
argument is that we should protect ecosystems for purely moral or aesthetic
reasons, and we do not need valuations of ecosystems for this purpose. But
there are equally compelling moral arguments that may be in direct conflict with
the moral argument to protect ecosystems; for example, the moral argument that
no one should go hungry. Moral arguments translate the valuation and decision
problem into a different set of dimensions and a different language of discourse6;
one that, in our view, makes the problem of valuation and choice more difficult
and less explicit. But moral and economic arguments are certainly not mutually
exclusive. Both discussions can and should go on in parallel.

So, although ecosystem valuation is certainly difficult and fraught with
uncertainties, one choice we do not have is whether or not to do it. Rather, the
decisions we make as a society about ecosystems imply valuations (although not
necessarily expressed in monetary terms). We can choose to make these
valuations explicit or not; we can do them with an explicit acknowledgement of
the huge uncertainties involved or not; but as long as we are forced to make
choices, we are going through the process of valuation.

The exercise of valuing the services of natural capital 'at the margin' consists of
determining the differences that relatively small changes in these services make to
human welfare. Changes in quality or quantity of ecosystem services have value
insofar as they either change the benefits associated with human activities or
change the costs of those activities. These changes in benefits and costs either
have an impact on human welfare through established markets or through
non-market activities. For example, coral reefs provide habitats for fish. One
aspect of their value is to increase and concentrate fish stocks. One effect of
changes in coral reef quality or quantity would be discernible in commercial
fisheries markets, or in recreational fisheries. But other aspects of the value of
coral reefs, such as recreational diving and biodiversity conservation, do not
show up completely in markets. Forests provide timber materials through well
established markets, but the associated habitat values of forests are also felt
through unmarketed recreational activities. The chains of effects from ecosystem
services to human welfare can range from extremely simple to exceedingly
complex. Forests provide timber, but also hold soils and moisture, and create
microclimates, all of which contribute to human welfare in complex, and generally
non-marketed ways.

Valuation methods

Various methods have been used to estimate both the market and non-market
components of the value of ecosystem services9, 10, 11, 12, 13, 14, 15, 16 . In this
analysis, we synthesized previous studies based on a wide variety of methods,
noting the limitations and assumptions underlying each.

Many of the valuation techniques used in the studies covered in our synthesis are
based, either directly or indirectly, on attempts to estimate the
'willingness-to-pay' of individuals for ecosystem services. For example, if
ecological services provided a $50 increment to the timber productivity of a
forest, then the beneficiaries of this service should be willing to pay up to $50 for
it. In addition to timber production, if the forest offered non-marketed, aesthetic,
existence, and conservation values of $70, those receiving this non-market
benefit should be willing to pay up to $70 for it. The total value of ecological
services would be $120, but the contribution to the money economy of
ecological services would be $50, the amount that actually passes through
markets. In this study we have tried to estimate the total value of ecological
services, regardless of whether they are currently marketed.

Figure 1 shows some of these concepts diagrammatically. Figure 1a shows
conventional supply (marginal cost) and demand (marginal benefit) curves for a
typical marketed good or service. The value that would show up in gross
national product (GNP) is the market price p times the quantity q, or the area
pbqc. There are three other relevant areas represented on the diagram, however.
The cost of production is the area under the supply curve, cbq. The 'producer
surplus' or 'net rent' for a resource is the area between the market price and the
supply curve, pbc. The 'consumer surplus' or the amount of welfare the
consumer receives over and above the price paid in the market is the area
between the demand curve and the market price, abp. The total economic value
of the resource is the sum of the producer and consumer surplus (excluding the
cost of production), or the area abc on the diagram. Note that total economic
value can be greater or less than the price times quantity estimates used in GNP.

Figure 1a refers to a human- made, substitutable good. Many ecosystem
services are only substitutable up to a point, and their demand curves probably
look more like Fig. 1b. Here the demand approaches infinity as the quantity
available approaches zero (or some minimum necessary level of services), and
the consumer surplus (as well as the total economic value) approaches infinity.
Demand curves for ecosystem services are very difficult, if not impossible, to
estimate in practice. In addition, to the extent that ecosystem services cannot be
increased or decreased by actions of the economic system, their supply curves
are more nearly vertical, as shown in Fig. 1b.

In this study we estimated the value per unit area of each ecosystem service for
each ecosystem type. To estimate this 'unit value' we used (in order of
preference) either: (1) the sum of consumer and producer surplus; or (2) the net
rent (or producer surplus); or (3) price times quantity as a proxy for the
economic value of the service, assuming that the demand curve for ecosystem
services looks more like Fig. 1b than Fig. 1a, and that therefore the area pbqc is
a conservative underestimate of the area abc. We then multiplied the unit values
times the surface area of each ecosystem to arrive at global totals.

Ecosystem values, markets and GNP

As we have noted, the value of many types of natural capital and ecosystem
services may not be easily traceable through well functioning markets, or may not
show up in markets at all. For example, the aesthetic enhancement of a forest
may alter recreational expenditures at that site, but this change in expenditure
bears no necessary relation to the value of the enhancement. Recreationists may
value the improvement at $100, but transfer only $20 in spending from other
recreational areas to the improved site. Enhanced wetlands quality may improve
waste treatment, saving on potential treatment costs. For example, tertiary
treatment by wetlands may save $100 in alternative treatment. Existing treatment
may cost only $30. The treatment cost savings does not show up in any market.
There is very little relation between the value of services and observable current
spending behaviour in many cases.

There is also no necessary relationship between the valuation of natural capital
service flows, even on the margin, and aggregate spending, or GNP, in the
economy. This is true even if all capital service flows had an impact on well
functioning markets. A large part of the contributions to human welfare by
ecosystem services are of a purely public goods nature. They accrue directly to
humans without passing through the money economy at all. In many cases people
are not even aware of them. Examples include clean air and water, soil
formation, climate regulation, waste treatment, aesthetic values and good health,
as mentioned above.

Global land use and land cover

In order to estimate the total value of ecosystem services, we needed estimates
of the total global extent of the ecosystems themselves. We devised an
aggregated classification scheme with 16 primary categories as shown in Table 2
to represent current global land use. The major division is between marine and
terrestrial systems. Marine was further subdivided into open ocean and coastal,
which itself includes estuaries, seagrass/algae beds, coral reefs, and shelf
systems. Terrestrial systems were broken down into two types of forest (tropical
and temperate/boreal), grasslands/rangelands, wetlands, lakes/rivers, desert,
tundra, ice/rock, cropland, and urban. Primary data were from ref. 17 as
summarized in ref. 4 with additional information from a number of sources18, 19,
20, 21, 22,. We also used data from ref. 23, as a cross-check on the terrestrial
estimates and ref. 24 and ref. 25 as a check on the marine estimates. The 32
landcover types of ref. 17 were recategorized for Table 2 and Fig. 2. The major
assumptions were: (1) chaparral and steppe were considered rangeland and
combined with grasslands; and (2) a variety of tropical forest and woodland
types were combined into 'tropical forests'.


We conducted a thorough literature review and synthesized the information,
along with a few original calculations, during a one-week intensive workshop at
the new National Center for Ecological Analysis and Synthesis (NCEAS) at the
University of California at Santa Barbara. Supplementary Information lists the
primary results for each ecosystem service and biome. Supplementary
Information includes all the estimates we could identify from the literature (from
over 100 studies), their valuation methods, location and stated value. We
converted each estimate into 1994 US$ ha-1 yr-1 using the USA consumer price
index and other conversion factors as needed. These are listed in the notes to the
Supplementary Information. For some estimates we also converted the service
estimate into US$ equivalents using the ratio of purchasing power GNP per
capita for the country of origin to that of the USA. This was intended to adjust
for income effects. Where possible the estimates are stated as a range, based on
the high and low values found in the literature, and an average value, with
annotated comments as to methods and assumptions. We also included in the
Supplementary Information some estimates from the literature on 'total
ecosystem value', mainly using energy analysis techniques10. We did not include
these estimates in any of the totals or averages given below, but only for
comparison with the totals from the other techniques. Interestingly, these different
methods showed fairly close agreement in the final results.

Each biome and each ecosystem service had its special considerations. Detailed
notes explaining each biome and each entry in Supplementary Information are
given in notes following the table. More detailed descriptions of some of the
ecosystems, their services, and general valuation issues can be found in ref. 5.
Below we briefly discuss some general considerations that apply across the

Sources of error, limitations and caveats

Our attempt to estimate the total current economic value of ecosystem services is
limited for a number of reasons, including:

(1) Although we have attempted to include as much as possible, our estimate
leaves out many categories of services, which have not yet been adequately
studied for many ecosystems. In addition, we could identify no valuation studies
for some major biomes (desert, tundra, ice/rock, and cropland). As more and
better information becomes available we expect the total estimated value to

(2) Current prices, which form the basis (either directly or indirectly) of many of
the valuation estimates, are distorted for a number of reasons, including the fact
that they exclude the value of ecosystem services, household labour and the
informal economy. In addition to this, there are differences between total value,
consumer surplus, net rent (or producer surplus) and p × q, all of which are used
to estimate unit values (see Fig. 1).

(3) In many cases the values are based on the current willingness-to-pay of
individuals for ecosystem services, even though these individuals may be
ill-informed and their preferences may not adequately incorporate social fairness,
ecological sustainability and other important goals16. In other words, if we
actually lived in a world that was ecologically sustainable, socially fair and where
everyone had perfect knowledge of their connection to ecosystem services, both
market prices and surveys of willingness-to-pay would yield very different results
than they currently do, and the value of ecosystem services would probably

(4) In calculating the current value, we generally assumed that the demand and
supply curves look something like Fig. 1a. In reality, supply curves for many
ecosystem services are more nearly inelastic vertical lines, and the demand
curves probably look more like Fig. 1b, approaching infinity as quantity goes to
zero. Thus the consumer and producer surplus and thereby the total value of
ecosystem services would also approach infinity.

(5) The valuation approach taken here assumes that there are no sharp
thresholds, discontinuities or irreversibilities in the ecosystem response functions.
This is almost certainly not the case. Therefore this valuation yields an
underestimate of the total value.

(6) Extrapolation from point estimates to global totals introduces error. In
general, we estimated unit area values for the ecosystem services (in $ ha-1 yr-1)
and then multiplied by the total area of each biome. This can only be considered
a crude first approximation and can introduce errors depending on the type of
ecosystem service and its spatial heterogeneity.

(7) To avoid double counting, a general equilibrium framework that could
directly incorporate the interdependence between ecosystem functions and
services would be preferred to the partial equilibrium framework used in this
study (see below).

(8) Values for individual ecosystem functions should be based on sustainable use
levels, taking account of both the carrying capacity for individual functions (such
as food-production or waste recycling) and the combined effect of simultaneous
use of more functions. Ecosystems should be able to provide all the functions
listed in Table 1 simultaneously and indefinitely. This is certainly not the case for
some current ecosystem services because of overuse at existing prices.

(9) We have not incorporated the 'infrastructure' value of ecosystems, as noted
above, leading to an underestimation of the total value.

(10) Inter-country comparisons of valuation are affected by income differences.
We attempted to address this in some cases using the relative purchasing power
GNP per capita of the country relative to the USA, but this is a very crude way
to make the correction.

(11) In general, we have used annual flow values and have avoided many of the
difficult issues involved with discounting future flow values to arrive at a net
present value of the capital stock. But a few estimates in the literature were
stated as stock values, and it was necessary to assume a discount rate (we used
5%) in order to convert them into annual flows.

(12) Our estimate is based on a static 'snapshot' of what is, in fact, a complex,
dynamic system. We have assumed a static and 'partial equilibrium' model in the
sense that the value of each service is derived independently and added. This
ignores the complex interdependencies between the services. The estimate could
also change drastically as the system moved through critical non- linearities or
thresholds. Although it is possible to build 'general equilibrium' models in which
the value of all ecosystem services are derived simultaneously with all other
values, and to build dynamic models that can incorporate non-linearities and
thresholds, these models have rarely been attempted at the scale we are
discussing. They represent the next logical step in deriving better estimates of the
value of ecosystem services.

We have tried to expose these various sources of uncertainty wherever possible
in Supplementary Information and its supporting notes, and state the range of
relevant values. In spite of the limitations noted above, we believe it is very useful
to synthesize existing valuation estimates, if only to determine a crude, initial
magnitude. In general, because of the nature of the limitations noted, we expect
our current estimate to represent a minimum value for ecosystem services.

Total global value of ecosystem services

Table 2 is a summary of the results of our synthesis. It lists each of the major
biomes along with their current estimated global surface area, the average (on a
per hectare basis) of the estimated values of the 17 ecosystem services we have
identified from Supplementary Information, and the total value of ecosystem
services by biome, by service type and for the entire biosphere.

We estimated that at the current margin, ecosystems provide at least US$33
trillion dollars worth of services annually. The majority of the value of services
we could identify is currently outside the market system, in services such as gas
regulation (US$1.3 trillion yr-1), disturbance regulation (US$1.8 trillion yr-1),
waste treatment (US$2.3 trillion yr-1) and nutrient cycling (US$17 trillion yr-1).
About 63% of the estimated value is contributed by marine systems (US$20.9
trillion yr-1). Most of this comes from coastal systems (US$10.6 trillion yr-1).
About 38% of the estimated value comes from terrestrial systems, mainly from
forests (US$4.7 trillion yr-1) and wetlands (US$4.9 trillion yr-1).

We estimated a range of values whenever possible for each entry in
Supplementary Information. Table 2 reports only the average values. Had we
used the low end of the range in Supplementary Information, the global total
would have been around US$19 trillion. If we eliminate nutrient cycling, which is
the largest single service, estimated at US$17 trillion, the total annual value
would be around US$16 trillion. Had we used the high end for all estimates,
along with estimating the value of desert, tundra and ice/rock as the average
value of rangelands, the estimate would be around US$54 trillion. So the total
range of annual values we estimated were from US$16--$54 trillion. This is not a
huge range, but other sources of uncertainty listed above are much more critical.
It is important to emphasize, however, that despite the many uncertainties
included in this estimate, it is almost certainly an underestimate for several
reasons, as listed above.

There have been very few previous attempts to estimate the total global value of
ecosystem services with which to compare these results. We identified two,
based on completely different methods and assumptions, both from each other
and from the methods used in this study. They thus provide an interesting check.

One was an early attempt at a static general equilibrium input--output model of
the globe, including both ecological and economic processes and
commodities26,27. This model divided the globe in to 9 commodities or product
groups and 9 processes, two of which were 'economic' (urban and agriculture)
and 7 of which were 'ecological', including both terrestrial and marine systems.
Data were from about 1970. Although this was a very aggregated breakdown
and the data was of only moderate quality, the model produced a set of 'shadow
prices' and 'shadow values' for all the flows between processes, as well as the
net outputs from the system, which could be used to derive an estimate of the
total value of ecosystem services. The input--output format is far superior to the
partial equilibrium format we used in this study for differentiating gross from net
flows and avoiding double counting. The results yielded a total value of the net
output of the 7 global ecosystem processes equal to the equivalent of US$9.4
trillion in 1972. Converted to 1994 US$ this is about $34 trillion, surprisingly
close to our current average estimate. This estimate broke down into US$11.9
trillion (or 35%) from terrestrial ecosystem processes and US$22.1 trillion (or
65%) from marine processes, also very close to our current estimate. World
GNP in 1970 was about $14.3 trillion (in 1994 US$), indicating a ratio of total
ecosystem services to GNP of about 2.4 to 1. The current estimate has a
corresponding ratio of 1.8 to 1.

A more recent study28 estimated a 'maximum sustainable surplus' value of
ecosystem services by considering ecosystem services as one input to an
aggregate global production function along with labour and manufactured capital.
Their estimates ranged from US$3.4 to US$17.6 trillion yr-1, depending on
various assumptions. This approach assumed that the total value of ecosystem
services is limited to that which has an impact on marketed value, either directly
or indirectly, and thus cannot exceed the total world GNP of about US$18
trillion. But, as we have pointed out, only a fraction of ecosystem services affects
private goods traded in existing markets, which would be included in measures
such as GNP. This is a subset of the services we estimated, so we would expect
this estimate to undervalue total ecosystem services.

The results of both of these studies indicate, however, that our current estimate is
at least in approximately the same range. As we have noted, there are many
limitations to both the current and these two previous studies. They are all only
static snapshots of a biosphere that is a complex, dynamic system. The obvious
next steps include building regional and global models of the linked ecological
economic system aimed at a better understanding of both the complex dynamics
of physical/biological processes and the value of these processes to human well-
being29,30. But we do not have to wait for the results of these models to draw
the following conclusions.


What this study makes abundantly clear is that ecosystem services provide an
important portion of the total contribution to human welfare on this planet. We
must begin to give the natural capital stock that produces these services
adequate weight in the decision-making process, otherwise current and
continued future human welfare may drastically suffer. We estimate in this study
that the annual value of these services is US$16--54 trillion, with an estimated
average of US$33 trillion. The real value is almost certainly much larger, even at
the current margin. US$33 trillion is 1.8 times the current global GNP. One way
to look at this comparison is that if one were to try to replace the services of
ecosystems at the current margin, one would need to increase global GNP by at
least US$33 trillion, partly to cover services already captured in existing GNP
and partly to cover services that are not currently captured in GNP. This
impossible task would lead to no increase in welfare because we would only be
replacing existing services, and it ignores the fact that many ecosystem services
are literally irreplaceable.

If ecosystem services were actually paid for, in terms of their value contribution
to the global economy, the global price system would be very different from
what it is today. The price of commodities using ecosystem services directly or
indirectly would be much greater. The structure of factor payments, including
wages, interest rates and profits would change dramatically. World GNP would
be very different in both magnitude and composition if it adequately incorporated
the value of ecosystem services. One practical use of the estimates we have
developed is to help modify systems of national accounting to better reflect the
value of ecosystem services and natural capital. Initial attempts to do this paint a
very different picture of our current level of economic welfare than conventional
GNP, some indicating a levelling of welfare since about 1970 while GNP has
continued to increase31,32,33. A second important use of these estimates is for
project appraisal, where ecosystem services lost must be weighed against the
benefits of a specific project8. Because ecosystem services are largely outside
the market and uncertain, they are too often ignored or undervalued, leading to
the error of constructing projects whose social costs far outweight their benefits.

As natural capital and ecosystem services become more stressed and more
'scarce' in the future, we can only expect their value to increase. If significant,
irreversible thresholds are passed for irreplaceable ecosystem services, their
value may quickly jump to infinity. Given the huge uncertainties involved, we may
never have a very precise estimate of the value of ecosystem services.
Nevertheless, even the crude initial estimate we have been able to assemble is a
useful starting point (we stress again that it is only a starting point). It
demonstrates the need for much additional research and it also indicates the
specific areas that are most in need of additional study. It also highlights the
relative importance of ecosystem services and the potential impact on our
welfare of continuing to squander them.

Present address: Department of Systems Ecology, University of Stockholm,
S-106 91 Stockholm, Sweden.


S. Carpenter was instrumental in encouraging the project. M. Grasso did the
initial identification and collection of literature sources. We thank S. Carpenter,
G. Daily, H. Daly, A. M. Freeman, N. Myers, C. Perrings, D. Pimentel, S.
Pimm and S. Postel for helpful comments on earlier drafts. This project was
sponsored by the National Center for Ecological Analysis and Synthesis
(NCEAS), an NSF-funded Center at the University of California at Santa
Barbara. The authors met during the week of June 17--21, 1996 to do the
major parts of the synthesis activities. The idea for the study emerged at a
meeting of the Pew Scholars in New Hampshire in October 1995.

References Groot, R. S. Environmentalist 7, 105--109 (1987).
2.Turner, R. K. Economics, Growth and Sustainable
Environments , Macmillan, London 1988.
3.Turner, R. K. Ambio 20, 59-- 63 (1991). Groot, R. S. Functions of Nature: Evaluation of Nature in
Environmental Planning, Management, and Decision
Making , Wolters- Noordhoff, Groningen 1992.
5.Daily, G. Nature's Services: Societal Dependence on Natural
Ecosystems , Island, Washington DC 1997.
6.Turner, R. K. & Pearce, D. Economics and Ecology: New Frontiers
and Sustainable Development , 177--194 Chapman and Hall,
London 1993.
7.Costanza, R. & Daly, H. E. Conserv. Biol. 6, 37--46 (1992).
8.Bingham, G. Ecol. Econ. 14, 73-- 90 (1995).
9.Mitchell, R. C. & Carson, R. T. Using Surveys to Value Public Goods:
the Contingent Valuation Method , Resources for the Future,
Washington DC 1989.
10.Costanza, R., Farber, S. C. & Maxwell, J. Ecol.
Econ. 1, 335--361 (1989).
11.Dixon, J. A. & Sherman, P. B. Economics of Protected Areas , Island,
Washington DC 1990.
12.Barde, J.-P. & Pearce, D. W. Valuing the Environment: Six Case
Studies , Earthscan, London 1991.
13.Aylward, B. A. & Barbier, E. B. Biodiv. Cons. 1, 34 (1992).
14.Pearce, D. Economic Values and the Natural World , Earthscan,
London 1993.
15.Goulder, L. H. & Kennedy, D. Nature's Services: Societal Dependence
on Natural Ecosystems , 23--48 Island, Washington DC 1997.
16.Costanza, R. & Folke, C. Nature's Services: Societal Dependence on
Natural Ecosystems , 49--70 Island, Washington DC 1997.
17.Matthews, E. J. Clim. Appl. Meteorol. 22, 474--487 (1983).
18.Deevey, E. S. Sci. Am. 223, 148-- 158 (1970).
19.Ehrlich, R., Ehrlich, A. H. & Holdren, J. P. Ecoscience: Population,
Resources, Environment , W.H. Freeman, San Francisco 1977.
20.Ryther, J. H. Science 166, 72-- 76 (1969).
21.United Nations Environmental Programme First Assessment Report,
Intergovernmental Panel on Climate Change , United Nations, New
York 1990.
22.Whittaker, R. H. & Likens, G. E. Primary Production of the
Biosphere , 305--328 Springer, New York 1975.
23.Bailey, R. G. Ecosystem Geography , Springer, New York 1996.
24.Houde, E. D. & Rutherford, E. S. Estuaries 16, 161--176 (1993).
25.Pauly, D. & Christensen, V. Nature 374, 255--257 (1995).
26.Costanza, R. & Neil, C. Energy and Ecological
Modeling , 745--755 Elsevier, New York 1981.
27.Costanza, R. & Hannon, B. M. Network Analysis of Marine
Ecosystems: Methods and Applications , 90--115 Springer,
Heidelberg 1989.
28.Alexander, A., List, J., Margolis, M. & d'Arge, R. Alternative methods of
valuing global ecosystem services Ecol. Econ. (submitted)
29.Costanza, R., Wainger, L., Folke, C. & Mäler, K.-G.
BioScience 43, 545--555 (1993).
30.Bockstael, N. Ecol. Econ. 14, 143-- 159 (1995).
31.Daly, H. E. & Cobb, J. For the Common Good: Redirecting the
Economy Towards Community, the Environment, and a Sustainable
Future , Beacon, Boston 1989.
32.Cobb, C. & Cobb, J. The Green National Product: a Proposed Index
of Sustainable Economic Welfare , Univ. Press of America, New
York 1994.
33.Max-Neef, M. Ecol. Econ. 15, 115-- 118 (1995).

Reprinted from
Nature © Macmillan Publishers Ltd. 1997
Registered No. 785998 England.