232
ecosystem, depending on the nature of the interactions between the specific ecosystem and the human population. Thus generic lists of ecosystem
services by type of ecosystem are only of very cursory use at this stage. For any given policy setting,
a full accounting of relevant ecosystem goods and
services must be compiled.
The second step is to determine how the units in
which physical impacts are measured can be translated into the units that are relevant for human preferences and decision making. Often, it is conceptually easier for scientists to describe impacts of
human activities on ecosystems in the physical
units that are used to measure relationships within
the ecosystem. However, these units may not be
consistent with the units that individuals and firms
are compelled to consider when deciding whether
to undertake actions that would have an impact on
the environment. 3 Since the goal of ecosystem valuation is to use estimates to develop policies and
instruments to change incentive patterns and to encourage people to make decisions that are consistent with long-term interests of society, the units of
analysis need to be relevant to human options and
choices. 4 Heimlich et al. (1998) provide an excellent example for the case of wetlands in which they
identify physical impacts, then identify the units
that are appropriate for measuring how the physical impacts translate into human values, and then
identify which human activities will be affected and
by how much. They categorize and review nonmarket valuation estimates for wetlands from existing literature at this level.
Third, the scope of changes induced by the policy must be well understood. It is necessary to determine the extent of the potential changes in quantity or quality of the goods and services that might
be affected by a policy. It is generally not useful
to measure the total value of ecosystem services,
unless the proposed policy would result in total loss
of these services. Of greater value for informing
3The issue of reconciling measurement units to bridge
human values and ecosystem impacts can seem especially difficult in very complex ecosystems. This is one
area of research that could receive more attention.
4The intention here is not to minimize the importance of
measuring the significance of a physical impact on the
overall function of a complex ecosystem, when some of
these impacts or changes in overall function may not
have an obvious role in human values. There are other
ways that the term ecosystem values may be used, which
refer purely to physical impacts. However, these notions
of value are different from the definition here. More importantly, in terms of the goal of understanding and altering incentives that influence human interactions with
ecosystems, a purely physical definition is of little use.
Measuring Ecosystem Values
decisions are measures of marginal changes in
value that result from specific policy actions. Thus
a given baseline level of the provision of the services must be established, along with the potential
changes from this baseline. For the majority of situations in which ecosystems may be altered but not
destroyed, valuation must be based on the marginal
changes that would be expected to occur. A large
source of bias can occur if a valuation estimate that
is based on the total value of an ecosystem service
is expressed as an average (by dividing over the
number of units provided, for example) and then
used as a proxy for a marginal value. If a marginal
value is required, then the study design needs to assure that a marginal value is measured. For example, Rollins et al. (1997b) estimate the value of
maintaining water quality for municipal drinking
water services. The policy context required that
they estimate the value of an incremental change
in water quality. More precisely, they estimated the
willingness to pay of users to maintain the level of
quality to be consistent with health guidelines. A
measure of total value would have provided information that would have been of little practical use
in their policy context.
Fourth, the relevant population for each of the
identified ecosystem values must be properly identified to avoid biased valuation estimates. The relevant population can vary, depending on which
ecosystem services are considered. In the example
of the public forestland taken out of commercial
use, any potential changes in recreational use patterns may affect mostly local popUlations. The
change in recreational benefits generated by the
policy would depend on the size of the population
in the region where the forest is located, the number of available substitutes for recreational land,
and the change in the quality of the recreational experience. However, any changes in habitat that affect threatened or endangered species might be valued by a much larger population, most of which
may not live in the region.
Fifth, the choice of valuation method must be appropriate to the circumstances of the proposed policy. For example, most travel cost models are not
appropriate for determining marginal changes in
the quality of a site for recreational purposes, but
they work well for determining the economic value
of creating a new site or the loss of totally eliminating a site. On the other hand, direct methods,
which include contingent valuation and choice experiments, work well for marginal changes. Only
direct methods are available for measuring existence, or passive use, values; it is simply not possible to use any kind of indirect method, such as a
travel cost model for this purpose. While it is com-
ecosystem, depending on the nature of the interactions between the specific ecosystem and the human population. Thus generic lists of ecosystem
services by type of ecosystem are only of very cursory use at this stage. For any given policy setting,
a full accounting of relevant ecosystem goods and
services must be compiled.
The second step is to determine how the units in
which physical impacts are measured can be translated into the units that are relevant for human preferences and decision making. Often, it is conceptually easier for scientists to describe impacts of
human activities on ecosystems in the physical
units that are used to measure relationships within
the ecosystem. However, these units may not be
consistent with the units that individuals and firms
are compelled to consider when deciding whether
to undertake actions that would have an impact on
the environment. 3 Since the goal of ecosystem valuation is to use estimates to develop policies and
instruments to change incentive patterns and to encourage people to make decisions that are consistent with long-term interests of society, the units of
analysis need to be relevant to human options and
choices. 4 Heimlich et al. (1998) provide an excellent example for the case of wetlands in which they
identify physical impacts, then identify the units
that are appropriate for measuring how the physical impacts translate into human values, and then
identify which human activities will be affected and
by how much. They categorize and review nonmarket valuation estimates for wetlands from existing literature at this level.
Third, the scope of changes induced by the policy must be well understood. It is necessary to determine the extent of the potential changes in quantity or quality of the goods and services that might
be affected by a policy. It is generally not useful
to measure the total value of ecosystem services,
unless the proposed policy would result in total loss
of these services. Of greater value for informing
3The issue of reconciling measurement units to bridge
human values and ecosystem impacts can seem especially difficult in very complex ecosystems. This is one
area of research that could receive more attention.
4The intention here is not to minimize the importance of
measuring the significance of a physical impact on the
overall function of a complex ecosystem, when some of
these impacts or changes in overall function may not
have an obvious role in human values. There are other
ways that the term ecosystem values may be used, which
refer purely to physical impacts. However, these notions
of value are different from the definition here. More importantly, in terms of the goal of understanding and altering incentives that influence human interactions with
ecosystems, a purely physical definition is of little use.
Measuring Ecosystem Values
decisions are measures of marginal changes in
value that result from specific policy actions. Thus
a given baseline level of the provision of the services must be established, along with the potential
changes from this baseline. For the majority of situations in which ecosystems may be altered but not
destroyed, valuation must be based on the marginal
changes that would be expected to occur. A large
source of bias can occur if a valuation estimate that
is based on the total value of an ecosystem service
is expressed as an average (by dividing over the
number of units provided, for example) and then
used as a proxy for a marginal value. If a marginal
value is required, then the study design needs to assure that a marginal value is measured. For example, Rollins et al. (1997b) estimate the value of
maintaining water quality for municipal drinking
water services. The policy context required that
they estimate the value of an incremental change
in water quality. More precisely, they estimated the
willingness to pay of users to maintain the level of
quality to be consistent with health guidelines. A
measure of total value would have provided information that would have been of little practical use
in their policy context.
Fourth, the relevant population for each of the
identified ecosystem values must be properly identified to avoid biased valuation estimates. The relevant population can vary, depending on which
ecosystem services are considered. In the example
of the public forestland taken out of commercial
use, any potential changes in recreational use patterns may affect mostly local popUlations. The
change in recreational benefits generated by the
policy would depend on the size of the population
in the region where the forest is located, the number of available substitutes for recreational land,
and the change in the quality of the recreational experience. However, any changes in habitat that affect threatened or endangered species might be valued by a much larger population, most of which
may not live in the region.
Fifth, the choice of valuation method must be appropriate to the circumstances of the proposed policy. For example, most travel cost models are not
appropriate for determining marginal changes in
the quality of a site for recreational purposes, but
they work well for determining the economic value
of creating a new site or the loss of totally eliminating a site. On the other hand, direct methods,
which include contingent valuation and choice experiments, work well for marginal changes. Only
direct methods are available for measuring existence, or passive use, values; it is simply not possible to use any kind of indirect method, such as a
travel cost model for this purpose. While it is com-