Growing demand for analyses of the benefits of environmental improvements (or the costs of environmental damages) has increased interest in using estimates of such benefits in one setting to calculate benefits in another setting. At present, some types of these benefits (or costs)—which can be categorized as effects on health, output, economics assets, and environmental assets—are more amenable to such benefit transfers than others. Original studies that value health effects, for example, do not always lend themselves to benefit transfers. Most of the studies that value mortality risk address the risk of accidental death, which is an inappropriate context for valuing deaths from environmental causes. One way to make benefit transfers more feasible and reliable is to design original research with the purpose of obtaining results to be used in benefit transfers.
In the United States and elsewhere, there is a growing demand for analyses that quantify the benefits to society of improvements in environmental quality or the costs to society of environmental degradation. Since the establishment of environmental and natural resource economics as a discipline in the early 1970s, the primary demand for such benefit (or damage) analyses in the United States has come from two sources: government agencies interested in determining how projects and policies affecting water resources will change the value of and demand for recreation sites, and agencies seeking to comply with Executive Order 12291, which mandates that a cost-benefit analysis be conducted for all major regulations.
Because it would not be feasible to conduct new research to analyze the benefits of every policy and regulation to improve the environment, interest has arisen in developing techniques for benefit transfers—that is, for applying benefit studies made in a specific policy context and location to another context, or location, or both. One of the most successful benefit-transfer exercises to date involves the U.S. Environmental Protection Agency's cost-benefit analysis of regulations for phasing down the lead content of gasoline. In this analysis, the agency used existing benefit studies to estimate the values of reducing premature deaths and of avoiding acute health effects by decreasing individuals' exposure to lead in gasoline. On the basis of these and other estimates, EPA argued that the phasedown made economic sense.
Demand for benefit transfers has increased in recent years. In the wake of several disastrous oil spills, the concept of benefit transfer has been embodied in the U.S. Department of the Interior's Type A natural resource damage assessment model. This model uses original benefit studies of damages to recreational and commercial fishing and hunting from a given type and size of oil spill at given coastal locations to estimate similar damages from similar oil spills at any coastal location.
A fairly new demand for benefit transfers stems from a movement in the United States by state public utility commissions (PUCs) to formally introduce estimates of the environmental costs of alternative means for generating electricity into the decision making of electric utilities. At present, more than half of the states in the country require or are considering requiring electric utilities to account for residual environmental damages from alternative generation technologies when making decisions about which technologies to invest in. Without estimates of these damages, the utilities cannot make investments that minimize the social costs of electricity production.
Another potentially major source of demand for benefit transfers comes from international aid organizations, such as the World Bank and the U.S. Agency for International Development. These organizations are trying to value the environmental effects of projects in developing countries for which they are considering making loans or grants. Because there are few original studies of the benefits of environmental improvements in developing countries, the organizations are attempting to use the results of benefits studies in developed countries to estimate the value of the environmental effects of projects in developing countries.
Given the burgeoning demand for benefit transfers, it is important to consider limitations to the feasibility and usefulness of benefit-transfer exercises involving some types of benefits or damages. Benefit transfers involving environmental damages that result from electricity production, for example, are hindered at the start by the lack of original studies that provide comprehensive estimates of such damages. Even if such studies were conducted, the use of their damage valuations in other contexts would be challenging for at least two reasons. First, because the extent and nature of environmental damages associated with electric power plants hinge on the location of the plants, analysts would have to develop and codify techniques for using the valuations to estimate the value of damages at each plant site. Second, in assessments of these damages, they would have to include the nonuse values of environmental assets not exchanged in the marketplace. The problem here is that studies of nonuse values cannot yet support benefit transfers involving the environmental damages resulting from electricity production. Most studies of nonuse values examine very large changes in the quantity or quality of environmental assets—such as the extinction of a species or the loss of an ecosystem—within unique environments. However, the effects of a single power plant on environmental assets at locations where it is acceptable to site power plants are likely to be trivial.
Estimates of the environmental damages from electricity production that would result from benefit transfers are not yet credible enough to support the pricing of electricity at least social cost, but could be used to rank generation technologies according to social costs.
In the context of electricity production, estimates of damages that would result from benefit transfers would not be sufficient or credible enough to support certain forms of social costing—namely, the dispatch of generating units and the pricing of electricity at least social cost. To rely on benefit transfers to determine the order in which different electricity-generation technologies should be used and the price that consumers should pay for electricity would be to push benefit-transfer techniques and original benefit studies beyond their present capabilities. However, benefit transfers can credibly support a rank ordering of new options for power generation on the basis of social costs—that is, they can indicate to utilities which technologies to invest in so as to minimize such costs. This capability is of significant value.
As in the case of environmental damages resulting from electricity generation, benefit transfers involving the environmental effects of foreign aid projects in developing countries may also have limitations. In using valuations from benefits studies conducted in developed countries to estimate such effects, analysts must take into account the differences between developed countries and developing countries with respect to personal income, institutions, cultures, climate, resources, and so on. Even if these differences can be reflected in estimates of benefits, it is debatable whether benefit transfers are legitimate for valuing certain types of nonmarket commodities in developing countries. This is because the basic tenet of individual sovereignty underlying benefit estimation in the United States and most other developed countries may not be applicable in societies that place emphasis on group welfare. Nevertheless, it may be better to make benefit transfers involving the environmental effects of foreign aid projects than to make no attempt to quantify these effects. Until developing countries conduct their own benefit studies, the careful use of benefit transfers should help analysts determine the impacts of such projects on the environment.
Which benefits can be transferred now?
The benefits of environmental improvements can be categorized as effects on health, output, economic assets, and environmental assets. As regards effects on output and economic assets, benefit transfers can be ignored. Damages to output—for example, damages to crops from air pollution or to commercial fishing from oil spills—are easy enough to estimate by conducting original research. On the other hand, damages to economic assets—damages to buildings from acid rain, for example—cannot reliably be estimated in original studies, let alone in benefit transfers, because inventories of such assets are lacking. In addition, no major studies have examined how owners of economic assets will act to minimize damages to such assets from pollution. Until behavior with regard to protective and mitigative measures is better understood, no benefit transfer to estimate the benefits of reducing damage to economic assets can be reliably made.
The category of benefits most amenable to benefit transfer is effects on health. Once atmospheric or other natural processes are taken into account—for example, in the estimation of the effect of reduced emissions on ambient air quality—it can be assumed that the health effects and the values people place on avoiding these effects are reasonably similar across locations.
Methods for estimating health benefits have been codified for many years. To obtain the mortality benefits from a particular program, for instance, estimates of the value of a statistical life are taken from original studies and multiplied by the number of deaths the program is expected to delay. A similar protocol is followed in using the literature on the values of avoiding acute health effects to estimate the benefits of pollution reductions. Indeed, there are spread sheet models available to obtain estimates of the health benefits of environmental improvements. These models match estimates of changes in air pollution concentrations to dose-response functions for a wide variety of health effects and then match these functions to unit values for avoiding adverse effects.
Although benefit transfers involving health effects are the easiest to undertake, they are of the crudest type. Few of the spread sheet models used in benefit transfers account for differences between the age, personal income, and other socioeconomic characteristics of individuals at the site considered in the original study and those of individuals at the site being considered in a benefit transfer study. In addition, methods for establishing best estimates and margins of error in estimates of such benefits are ad hoc and differ from one benefit transfer study to another.
Not all original studies that value health effects lend themselves to benefit transfers. Most of the literature on valuing mortality risk addresses the risk of accidental death in prime-age adults. With the possible exception of deaths due to accidental releases of toxic substances, this risk context is inappropriate for valuing deaths associated with environmental causes—for example, deaths due to cancer in which pollution is a factor. One mortality-risk study does address the latency issue so important to valuing deaths due to cancer, but it is silent on the effect of prior health status and age on the willingness of people to pay for reductions in cancer risks.
The most problematic area for benefit transfer is damages to environmental assets such as plant and animal species in natural ecosystems, climate, visibility, and sites where natural geographic features provide recreational opportunities. Consider the difficulties the last two pose for benefit transfers.
Applying recreation values and demand functions from one study to another may present the greater challenge. Analysts may find it hard to account for regional factors (such as the range and quality of substitute recreation sites) and site-specific factors (such as the amount of congestion at a given recreation area) that affect individuals' valuation of and demand for recreation sites. In addition, they have no acceptable procedures for determining the size of the population affected by a change in the quality or quantity of recreation sites.
Applying recreation values from one study to another is difficult because such values are highly sensitive to site-specific variables.
Benefit transfers involving recreation values present other difficulties as well—among them, the measurement of recreation values in dollars per day. Because unit-day values represent the average values of a wide range of site characteristics and policy scenarios, their application to specific sites is problematic. In fact, it is more difficult than the application of unit values to health effects because of the presumption that individuals' willingness to pay to avoid health damages is less influenced by site variables than is their willingness to pay for recreation. Another difficulty is the somewhat limited usefulness of the recreation literature in estimating some social costs. The majority of this literature focuses on changes in the availability of recreation resources, ignoring changes in their quality.
Benefit transfers involving visibility values present formidable challenges due to the sensitivity of these values to the characteristics of individual regions, sites, and markets. The characterization of the study site is particularly difficult in visibility benefit transfers. Visual range can be characterized in a relatively straightforward way, but the vista being affected by a change in visibility is hard to characterize beyond "urban," "rural," and "recreational area." Defining the geographic extent of the population affected by such a change is even more difficult than defining that of the population affected by a change in the quality or quantity of recreation. The market for a particular recreation site can be defined (somewhat arbitrarily) as the residential location of users of the site; but the market affected by smog in an urban area, for example, is unclear.
Although visibility values are highly sensitive to regional, site, and market characteristics, the literature on visibility benefits is fairly conducive to benefit transfer. Studies of visibility values in a number of cities permit examination of city-specific factors that affect the values and derivation of functional relationships that terminate in individuals' willingness to pay for improvements in visibility, given a specific baseline visual range and size of change. There are a number of examples of benefit transfers involving visibility. An analysis by the Electric Power Research Institute and Decision Focus Inc. of the benefits of improved visibility in the eastern United States due to reductions in sulfur dioxide (SO2) emissions is a good example of a benefit transfer in which all the steps of the damage function approach are linked together. The analysis showed the linkages between SO2 emissions and ambient SO2 concentrations, concentrations and changes in visibility, changes and individuals' awareness of them, and awareness and willingness to pay for improvements in visibility.
There is significant debate about the protocols used in contingent valuation studies to elicit willingness to pay (WTP) for improvements in visibility. In these studies, individuals are queried about their WTP under a particular set of circumstances. It has been observed that when respondents are shown photographs depicting air pollution, the size of the photographs appears to influence their willingness to pay. In addition, analysts are concerned that respondents are, in part, using visibility as a proxy for health effects.
International International aid organizations are using studies of the environmental damages resulting from development in industrialized countries to value the environmental effects of development projects they are considering funding in nonindustrialized countries. Using valuations from these studies to estimate the potential environmental damages of development projects in a developing-country setting is difficult but necessary until developing countries conduct their own environmental damage studies
Protocols for benefit transfers
As noted above, original benefits studies often do not lend themselves to use in other contexts. Because original studies are not conducted with benefit transfers in mind, they often do not report information that would facilitate transfers.
There are some general protocols for using original studies of the benefits of environmental improvements to estimate benefits in other contexts. They emphasize the use, whenever possible, of demand or value functions, as opposed to average unit values. Use of the function approach puts additional burden on the analyst. For instance, the analyst must gather data on those variables at the site being considered in the benefit transfer study that were found to affect WTP for environmental improvements at the site considered in the orginal study. Without careful reporting of results in the original study, the demand or value function approach may be impossible.
Protocols are needed to guide varying practices for using multiple studies of a given effect to establish a range of uncertainty.
In the practical application of the broad protocols for benefit transfers, there are many choices to be made and few guidelines to follow. How should analysts proceed when original studies value changes in the output of a product but do not link such changes to changes in the quantity or quality of environmental resources? When the underlying science is poor, should the analyst spend a lot of time on the valuation step when he or she knows that the final benefit estimate is only as good as Its weakest component? When original valuation studies have significant problems, either in their own right or for benefit transfers, does the analyst even attempt a benefit transfer?
Protocols are perhaps most needed to guide the widely varying practices for using multiple studies of a given effect (each study with significant flaws) to establish a range of uncertainty. Consider the use of the three existing contingent valuation studies that provide symptom-day values—that is, values for avoiding symptoms of illness for a 24-hour period. The small absolute differences among the values, which plausibly range from $2 to $20, in each of the studies can translate into large percentage differences. Thus benefit estimates depend significantly on the values chosen in each study. In benefit transfers, analysts have established a range of uncertainty for the three studies' symptom-day values in two ways. Some analysts average the midpoint values of the studies and obtain a range of values by averaging each study's estimates of "low" and "high" values, respectively. Others use only midpoint values from the three studies to represent low, middle, and high estimates of unit values. Rather than trying to establish a range of uncertainty, some analysts merely use the values given in the study they deem least flawed.
Analysis and codification of criteria for evaluating original studies would assist analysts in the conduct of benefit transfers. However, the establishment of detailed criteria may not yield many benefits. Beyond ascertaining whether original studies are credible—that is, whether they are based on acceptable theory, whether the theory links up with careful empirical research, and whether essential results are reported—it is not clear what more can be done.
The weighting of criteria for evaluating original studies is crucial. Weighting depends, among other factors, on the use to which the studies will be put, the policy setting in which the studies' valuations will be applied, and the degree to which other data will supplement the valuations. In weighting criteria, a premium should be placed on giving analysts the flexibility to include studies considered most appropriate for the problem at hand. In return for this flexibility, analysts should be responsible for documenting their choice of studies.
Research agenda
To meet the demand for reliable benefit transfers in the context of environmental policy, major research efforts are needed. They could include developing methods to make better use of existing studies in the benefit transfer process, improving the quality of original studies so that the results of benefit transfer studies will be more credible, routinely including certain design elements in original studies to facilitate benefit transfers, and designing original research with the sole purpose of obtaining results to be used in benefit transfers.
Original studies can be more efficiently used in benefit transfer exercises to the extent that their results can be analyzed as a group—either in a meta-analysis in which the results of each study are treated as a single observation, or (if the data used in each of the studies can be obtained) into new analyses of the combined data sets. Such analyses could, in theory, estimate values or functions that eliminate (or at least reduce) the need for ad hoc consideration of a number of studies of a given effect in order to establish best estimates and a margin of error.
In the context of social costing, original studies that examine willingness to pay (WTP) to avoid further environmental degradation would be of greater use in benefit transfers than studies that examine WTP for environmental improvements.
Obviously good original studies will make for more credible benefit transfers. A "good" study is one that explicitly links what it is valuing to what it is measuring. Studies that value recreation benefits are not particularly good at linking the commodity being valued to the physical measures that affect the quality or quantity of the commodity. As noted above, most of these studies value quantitative changes in output, particularly changes in the rate of fish caught by recreational fishers, but rarely link these changes to measurements of changes in environmental assets, such as changes in water quality or in the size of fish populations. Protocols for ensuring that the commodities being valued are linked to the commodities being measured are needed if benefit transfers are to be broadly successful. Studies of the acute health effects of air pollution, for instance, would prove more useful for and more amenable to benefit transfers if they measured changes in the number of days or episodes in which individuals experience symptoms of illness, instead of measuring lung function.
Researchers engaged in original benefits analyses could also make these analyses more amenable to benefit transfers if they considered how the results of their studies will be used. At a minimum, they could facilitate benefit transfers by reporting mean values for independent variables and the equations used to estimate changes in consumer surplus.
Several changes in the design of original benefits analyses would also be helpful. One beneficial change would be a greater focus on site, regional, and market variables that might influence valuations. Another would be an examination of WTP to avoid further environmental degradation rather than an examination of WTP for environmental improvements. There is no reason to believe that WTP to obtain a given degree of environmental improvement is equal to WTP to avoid an equal degree of environmental damage, and the latter is more germane to social costing. The premise of formulating environmental policy on the basis of the costs to society of the environmental damages of polluting activities is that at least some environmental conditions will worsen without regulation of polluting activities.
Most studies of recreation benefits value quantitative changes in output—particularly changes in the rate of fish caught by recreational fishers—without linking these changes to measurements of changes in environmental assets such as changes in water quality. If benefit transfers involving recreation values are to be successful, original studies of recreation benefits must explicitly link what they are valuing to what they are measuring
Benefit transfers would also be facilitated by designing original benefits analyses for the purpose of obtaining results to be used in transfers. As the characteristics of commodities, regions, sites, and markets are likely to affect willingness to pay to avoid environmental degradation, it would be helpful if original studies included information about these factors. Doing so would help analysts to capture the differences among the factors and to investigate which factors weigh most heavily in valuations.
In the area of health benefits, original valuation studies that provide estimates of willingness to pay for reductions in the risks of premature mortality due to diseases with environmental causes would reduce reliance on the largely inappropriate literature dealing with individuals' willingness to pay some amount of wages to avoid accidental death in the workplace. It is particularly important to establish the effects of age and sex on willingness to pay to reduce premature mortality risks associated with environmental exposures. Original studies that estimate willingness to pay for life-years saved directly would obviate the ad hoc approaches currently being used in benefit-transfer exercises to modify the valuations obtained from the literature on accidental death.
Benefit transfers would also be aided by changes in original studies to estimate individuals' willingness to pay to avoid episodes of illness. Most morbidity studies seek values for single symptoms of illness rather than for illness complexes or episodes. Studies that provide values for the latter would help analysts come up with an overall estimate of the value of multiple acute-health effects over periods in which these effects actually occur, although health science provides little guidance as yet on the relationship between air pollution and episodes of illness.
Alan J. Krupnick is a senior fellow in the Quality of the Environment Division at Resources for the Future.
A version of this article appeared in print in the January 1993 issue of Resources magazine.
A version of this article appeared in print in the June 1993 issue of Resources magazine.
