• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

Natural Resource Damage Assessment Methods: Lessons in Simplicity From State Trustees

Posted on: Tuesday, 19 October 2004, 12:00 CDT

When natural resource damages are caused by releases of hazardous materials into the environment, government trustees must conduct Natural Resource Damage Assessments (NRDAs) to support claims to recover the value of lost or damaged resources. This article sets forth theoretical arguments that support efforts to develop unbiased simplified NRDA methods for use by government trustees and proposes a set of criteria that can be used to evaluate the quality of any such simplified method. The authors then describe the simplified methods being used by five states across the country, affording academic economists a rare view of the kinds of methods state agencies use in-house. The article evaluates those methods against the criteria set forth and discusses the potential of other nonstate- specific simplified NRDA methods (benefit transfer and Type A models) to do the job better. The new framework established can guide future research to design simplified methods that are less biased than the simplified methods currently in use by some states without compromising ease of implementation. (JEL Q5, K32)

ABBREVIATIONS

CERCLA: Comprehensive Environmental Response, Compensation, and Liability Act

DECY: Department of Ecology

DEP: Department of Environmental Protection

MPCA: Minnesota Pollution Control Agency

NOAA: National Oceanographic and Atmospheric Administration

NRD: Natural Resource Damages

NRDA: Natural Resource Damage Assessment

OPA: Oil Pollution Act

PRP: Potentially Responsible Party

I. INTRODUCTION

Natural resource damages can be caused by releases of hazardous materials, such as oil and toxic chemicals into the environment. The damage may be extensive, as in the case of polychlorinated biphenyl contamination of the Lower Fox River in Wisconsin, which led to national resource damage (NRD) settlements exceeding $42 million (Ando and Wildermuth, 2004) and the American Trader spill of 40,000 gallons of oil in southern California, which led to a jury verdict of $12.7 million for damages to beach and boating uses alone (Helton and Penn, 1999). However, most releases are very small-in 2000 there were 8058 oil spills of less than 100 gallons and only 77 spills of more than 1000 gallons (U.S. Coast Guard, 2001). In all cases, government trustees (the state and federal agencies designated to act on behalf of the public to safeguard public resources) have the right and responsibility to file claims to recover the value of lost or damaged resources according to the NRD provisions of the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA), the Oil Pollution Act (OPA), or related state law.

The trustees must conduct Natural Resource Damage Assessments (NRDAs) to support NRD claims. This process is challenging because most values associated with natural resources cannot be obtained from the market trades. Environmental economists have done much to address this problem by developing economic approaches to placing monetary values on the consequences of pollution discharges. Some approaches (e.g., travel cost and hedonic wage models) rely on revealed preferences by consumers in the market place. Other approaches analyze stated preferences with such tools as contingent valuation surveys. These methods have supported NRD claims in high- profile cases such as the Exxon Valdez accident in Alaska and the Upper Clark Fork River Basin Superfund Site in Montana.1

The state of the art in this type of NRDA methodology is impressively sophisticated. The voluminous literature on nonmarket valuation techniques embodies the efforts researchers have made to refine and improve the reliability and accuracy of these techniques in yielding values consistent with economic theory (Brookshire and Scrogin, 2000).2 These approaches are also expensive and time- consuming to execute. For example, a contingent valuation study of the damage caused by the study the Exxon Valdez spill is estimated to have cost $3 million. Such studies can therefore only be conducted in cases in which potential damages are very large (Harrison and Lesley, 1996).

Most government trustee agencies, however, are small, budget- constrained, and lightly staffed, often with no economists on board. There exist few simplified NRDA methods-tools that can be used by a person with minimal training to perform simple, low-cost, expedited NRDAs. This paucity of simplified methods has led most state trustees to opt not to pursue the majority of potential NRD cases. A few state trustees have developed such methods to estimate the monetary value of NRDs, using data on physical damages to resources and parameters that impute a value to those damages.

This article sets forth theoretical arguments that support efforts to develop unbiased simplified NRDA methods for use by government trustees and proposes a set of criteria which can be used to evaluate the quality of any such simplified method. The authors then describe the simplified methods being used by some states, affording academic economists a rare view of the kinds of methods state agencies use when they do not hire expert consultants to carry out a NRDA. The authors evaluate those methods against the criteria set forth and discuss whether other nonstate-specific simplified NRDA methods (benefit transfer and Type A models) have the potential to do the job better.

II. CONCEPTUAL ISSUES IN THE DESIGN OF SIMPLIFIED METHODS FOR NRDA

Work dating back to Becker (1968) has emphasized that liability law can provide firms with proper incentives to engage in socially beneficial precautionary activity (Segerson, 2000). In the case of NRDs, one might in theory be able to induce firms to take efficient levels of precaution to avoid and mitigate accidents that cause NRDs if one forces firms to compensate the public for the lost values of the injured resources. Even though simplified methods for NRDA may yield imprecise damage estimates, they can play an important role in reducing the total social costs associated with NRDs from environmental contamination. Theoretical work by Kaplow and Shavell (1996) argues that under some circumstances, the victim who brings suit for damages against an injurer needs only to use an unbiased estimator to induce efficient levels of precaution. As long as the potentially injuring party faces an ex ante expected payment that is equal to their ex ante expectation of damages, that party will have efficient incentives to avoid and mitigate damages. In this model, it is actually welfare reducing for litigants to invest any money in the accuracy of ex post damage estimates.

This result does not apply so broadly, however, if firms can evade damages by challenging inaccurate damage assessments (Ando and Polasub, 2003). In that case, a welfare-maximizing trustee should use inexpensive and inaccurate (but unbiased) methods for small cases but adopt costly case-specific methods for big cases. This strategy balances transaction costs and power of incentives. Ando and Polasub (2003) also find that welfare can be improved if unbiased but simplified methods are granted rebuttable presumption or are otherwise blessed by the courts. This lowers the extent to which firms can evade payment due to inaccuracy, and thus reduces the extent to which trustees must engage in otherwise wastefully expensive NRDAs to induce efficient levels of protection.

Theory and intuition suggest that it may be welfare-enhancing to add low-cost assessment methods to trustees' NRDA toolkits. What characteristics, then, should a good simplified method have? The authors propose the following criteria to be used as guidelines for trustees and economists working on efforts to develop and improve low-cost NRDA methods.

Simple to Use

The method should be inexpensive, easy to understand, and quick to implement. It should produce damage estimates using limited spill- specific information (such as the type and volume of oil released, location of spill, or amount of groundwater contaminated). Ideally, such a method should be amenable to use by inexpert personnel and should not require specialized computer support so that NRD estimates can be developed by field staff away from the home office. State agencies often do not have economists and scientists on their staff that can apply sophisticated econometrics or habitat equivalency modeling to determine damage estimates.

Legal Recognition

A simplified method is more likely to be effective if authorized by a state or federal legal establishment. Firms would then be unable to evade payment by arguing about the inaccuracy of the method, transaction costs would be lower, and social welfare would be increased.

Transparent

The method should be conceptually transparent and the basis for the choice of parameters used in the calculations of monetary damages should be specified. This would allow trustees to adjust damage estimates to changes in factors, such as inflation rates, the costs of cleanup, and demographic and socioeconomic characteristics, that affect willingness to pay.

Damage Estimates Vary with Scope

Damage estimates should vary in magnitude with the magnitude of the injury to resources. Only then will firms' expected payments be equal to the expected damage to society of a spill, giving firms the right incentives to lower the probability of accidents occurring and the extent of damage caused when an accident does occu\r.

Net Present Value Calculated Appropriately

In many cases, NRDs occur over an extended period of time. These need to be aggregated temporally using appropriate discounting rules. Failure to compound past damages would lead to undercompensation for losses, whereas failure to discount expected future damages would imply that the public is overcompensated for losses.

Unbiased Estimate of Public Use and Nonuse Values

Estimates of damages to natural resources should reflect the loss of both use and nonuse values. Total damages should be a meaningful sum of unbiased estimates of society's willingness to pay to avoid both elements of the loss. Failure to account for all elements of the damages will bias estimates downward. Because environmental goods have public good aspects, it is also important that damage estimates reflect public values for the damaged resources (Hanemann, 1994). It is temptingly easy to let well-trained experts pass judgment on which resources are relatively more valuable to society. However, to induce efficient precaution on the part of potential polluters, NRD enforcement must hold polluters liable for damages equal to the loss to society of resource injuries, and economic theory tells us that this loss is measured by aggregate social willingness to pay to avoid the injury.

Reflect Socioeconomic Characteristics of Affected Population

The literature on nonmarket valuation shows that demographic and socioeconomic characteristics of households can influence their willingness to pay for environmental amenities and therefore their estimation of the benefits associated with such amenities. An unbiased assessment method should allow the NRD estimate for the same physical damage to vary across areas with the size of the population whose benefits should be included, with the nature of public preferences, and the willingness to pay for resources in those areas. Additionally, in developing aggregate measures of benefits, it is important to consider the geographical extent of the population whose benefits should be included (Dunford etal., 1997; Randall, 1997).

III. SIMPLIFIED METHODS IN PRACTICE: FIVE STATE PROFILES

A. Background

We asked 64 state agencies in 50 states and the District of Columbia about their NRD assessment and recovery activity (see Ando et al., 2004, for the full report). The states' responses are summarized in Table 1. For each state we have also provided information on the number of oil spills and Superfund sites (U.S. Coast Guard, 2001; Environmental Defense, 2002), which show the occurrences of cases that require NRDA.3

Agencies in 29 of the states responded to the survey with information about their programs; of those, 19 states had NRD program activity on which the agencies reported, and 10 responded to tell us that their agency had no NRD program at the current time. Many state agencies did not respond formally to the survey, often because they could not afford to devote human resources to answering questions. Of the 22 states from which the authors did not receive a formal response, they know that 6 have active programs and suspect that 4 others may have active programs. The authors remain uncertain of the status of NRD activity in the remaining 12 states.

The survey specifically asked each agency if it had a simplified NRDA method in use. At the time of the survey, only 4 agencies in the 28 responding states replied in the affirmative: Washington, Florida, New Jersey, and Minnesota. At that time, California's trustees did not use a simplified method, opting instead to do case- specific habitat-equivalency analyses. However, that agency has adopted simplified methods since that time. Interestingly, four of these five states rank highly in measures of NRD intensity. California, Florida, and Washington rank third, fourth, and sixth, respectively in the number of oil spills in 2000. New Jersey, California, and Florida rank fourth, fifth, and seventh in volume of oil spilled in 2000. New Jersey, California, Florida, and Washington rank first, second, sixth, and seventh in the number of Superfund sites located there in 2002. Only Minnesota went to the trouble of developing a simplified method in the absence of a large NRD case load; as will be seen, the method used there addresses qualitatively different cases than those assessed with the simplified methods in the other four states.

Staff members in each of these five states have generously shared information regarding their methods. The profiles are digested summaries of the information they provided. Note that it is possible that some of the trustees in the 23 nonrespondent states are also using codified simplified methods or that NRDA practice has changed in these 5 profiled states since the time of the survey. The authors do not claim to present an exhaustive catalog of state simplified methods but present these as examples of the kinds of methods that states are developing and adopting.

B. California4

The Office of Spill Prevention and Response in the California Department of Fish and Game has primary responsibility for NRD recovery in the state of California.5 Historically, staff members in that office relied on case-specific habitat-equivalency analysis to conduct NRDAs for cases in their state. Recently, agency staff members have developed a simplified method to use in some cases,6 even though state law does not include any particular mandate for such a method to be used. The California method can be used for oil spills that are small enough not to have NRD values in excess of $50,000 total. It provides a measure of the value of lost habitat uses but is not used to estimate lost human use values, such as recreation.

TABLE 1

Survey Responses by State

TABLE 1

Survey Responses by State

To use this method, field staff need only have information on what kind of habitat was affected by the spill, how much of the habitat was affected (either in acres or stream miles), degree of injury, and duration of injury. The method is designed to allow staff members to use this limited information to estimate quickly what the results of a habitat-equivalence analysis-based NRDA would be for the case at hand.

The tool is a set of eight tables, one for each of eight types of California habitat. Each table has degree of injury on one axis and duration of injury on the other. The cells in the table provide a dollar value per unit of habitat, which is then multiplied by the total amount of habitat affected to give a total NRD estimate. The dollar values are derived from calculations of the average costs of habitat restoration in the state of California. The function that translates degree and duration of injury into habitat restoration needs is based on a linearized closed-form solution of habitat- equivalency analysis itself, which was developed by the California staff.

C. Florida

The great majority of NRDAs in Florida are conducted using a simplified method set out in Florida law (State of Florida, 2001). Most of the details of the compensation schedule and its proper use are stipulated in the law itself, except that the law charges the Department of Environmental Protection (DEP) with the task of ranking nonpetroleum pollutants on a one-to-three scale of harmfulness.

The simplified method is used to estimate damages associated with small or moderate discharges into coastal and offshore waters. It is intended to cover oil and other hazardous materials but has been used almost exclusively to assess damages from oil spills. If a discharge involves fewer than 30,000 gallons, the DEP uses the compensation schedule to assess the damages for which the potentially responsible party (PRP) is liable. In case of a discharge of greater than 30,000 gallons, the PRP may choose either to pay the amount calculated with the compensation schedule or to pay the damages estimated by a case-specific damage assessment.

The Florida method uses data on several spill-specific variables: the volume and type of oil spilled, the proximity of the discharge to the Florida coastal shore and/or special management areas, the type and quantity of habitat affected, and the number of endangered or threatened species that are killed as a result of the discharge. Noncase-specific data information is used in this assessment method as well. Each type of oil is placed into one of three categories of harmfulness, and 12 habitat types have been given unit monetary values for use in this tool.

The first segment of the equation assigns some damages just for the existence of a release, even if there is no contact with habitat or harm done to endangered animals. That amount of base damages increases with the volume of the release and proximity to shore. The second segment of the equation adds to total damages in cases where habitat has been affected. That amount of damages increases with the size of habitat and with the harmfulness of the pollutant.

Base damages are higher if the release is in (or spreads to affect) a special management area, such as a state or national park or aquatic reserve. Damages are greater per unit of habitat for highly valued habitat types (such as coral reef) than for others, and damages are greater for the same amount of a given habitat type if it is located in a special management area. Additional damages are added for the deaths of animals listed as either endangered or threatened.

D. Washington

Washington State Law (State of Washington, 2001a, 2001b) established strict and several liability for damages associated with releases of oil into the waters of Washington State. That law authorized the Department of Ecology (DECY) to develop regulations to facilitate the process of estimating the damages associated with oil spills as described by the state law. The compensation schedule that office developed was codified in the body of state regulations (State of Washington, 1992). A different agency handles NRDs associated with \other types of environmental contamination; to these authors' knowledge, they do not use a simplified method in their work.

The DECY uses the compensation schedule to estimate adequate compensation for damages from oil spilled in four types of state waters: marine and estuarine waters (general), the Columbia River estuary, freshwater streams, rivers and lakes, and freshwater wetlands. Following a spill, a Resource Damage Assessment Committee meets to conduct a preassessment screening. If the committee determines (among other things) that a full damage assessment cannot be conducted at reasonable cost, then the simplified compensation schedule is used. In practice, this schedule is used to develop compensation estimates for all but a small number of the largest spills in the state.

The DECY uses data on several features of a spill as inputs to the assessment method: the volume and type of oil spilled, the location of the spill, the date on which the spill occurred, and the nature of any cleanup or protective actions that the PRP might have carried out to mitigate the damage caused by the spill. For spills of 1000 or more gallons, the ageney uses data on the composition of the habitat types that were actually exposed to the spilled oil (Washington DECY, 2002).

Other, noncase-specific information is used in the assessment method. Each type of oil is ranked on toxicity, persistence, and potential for mechanical injury. The state marine/estuarine waters were divided up into 131 zones and a scientific advisory board assigned each zone environmental sensitivity/vulnerability scores for habitat, birds, shellfish, marine fish, salmon, marine mammals, and recreation. Marine sensitivity scores vary by season; the method uses the score for the most sensitive season affected. Freshwater surface water and wetlands are assigned sensitivity categories as well.

When oil spills affect more than one type of environment, damages are calculated using the methods described for each of the receiving environment types exposed to the spilled oil. Total damages are then estimated as the greatest of the damages calculated for the receiving environment types exposed to the spill. This allows damage assessment to proceed without trying to ascertain how much oil spread onto each of the different types of environments.

E. New Jersey

There are multiple state statutes establishing the authority of New Jersey's DEP to require the investigation and restoration of natural resource injuries associated with hazardous-waste sites. The DEP has promulgated Technical Regulations for Site Remediation, which provides a mechanism for assessing natural resource damages and restoration required as part of the site remediation process for contaminated sites.9 A simplified method for determining injuries to groundwater, in particular, is included in that document, which is a formal part of the New Jersey Administrative Code (State of New Jersey, 1999). In NRD cases where resources other than groundwater are injured (e.g., beaches, wildlife), trustees in New Jersey use other NRDA methods to develop damage estimates.

In the case of groundwater contamination caused by a release of hazardous materials, the trustee office has a strong preference for having the responsible party implement an aquifer recharging project through property acquisition rather than simply paying a monetary sum. However, the responsible party has the option to provide monetary damage damages in lieu of implementing a restoration project. Groundwater injuries need not be restored (and thus are not assessed) if the contamination has not traveled beyond the property on which the discharge occurred, the remedial action involved is nondepletive (such as pump-and-treat systems) or no other natural resources (such as wetlands) have been affected by the contamination. De minimis exemptions may also apply; groundwater damages are not assessed if the contamination is small relative to the water needs of the area or if there is no prospective lost use.

To assess monetary damages due to groundwater contamination, New Jersey uses a simplified assessment method that uses case-specific data on the area and location of the contamination plume, the type of remedial action proposed, and the period of time over which the concentrations of contaminants in the groundwater have or will exceed the state groundwater quality standards. All of this information must be obtained as part of the remedial investigation; no new data collection is required to perform the damage assessment. This method takes two other major inputs that are not case- specific: groundwater recharge rates specific to the area of the site given by the Water Supply Master Plan (to determine the volume of contamination) and the dollar cost to consumers of potable water in the area of the site.

F. Minnesota

In 1994, the State of Minnesota launched a Closed Landfill Program to clean up and handle the long-term care for over 100 closed municipal solid waste landfills in the state. The 1996 Amendments to the Minnesota Landfill Cleanup Act of 1994 (Minn. Stat. Ch. 115B.441-115B.445) set forth a mechanism for obtaining contributions from insurance carriers to the funding for the program. In many cases, the people who bore legal responsibility for environmental response costs at these facilities (e.g., the owners or operators) held liability insurance for environmental response costs at the landfill. The Minnesota Pollution Control Agency (MPCA) works with the attorney-general's office to obtain appropriate payments from those insurance carriers that have outstanding exposure for cleanup liability at closed landfill sites. To have a foundation for the settlement negotiation process, the MPCA has developed estimates of NRDs, because the carriers can request settlement for NRDs at the time they settle for response costs and thus be granted immunity from future legal action regarding NRD compensation. The NRD estimation method has been developed in agency documents by the MPCA (Ruotsinoja, 1997; MPCA, 2002).

The MPCA bases its NRD estimates on groundwater contamination at the 66 landfills at which such contamination exceeds the state's health risk limits. The NRD estimate at each of the 66 landfills uses data on the volume of contaminated groundwater at the site. This method also uses nonsite-specific information on the average cost of lost groundwater services at a landfill. In a departure from the typical case-specific NRD assessment, this groundwater damage assessment method yields a monetary damage estimate used as a starting point for NRD negotiations with insurance carriers to settle their global liability for damages associated with all of the sites at which they have outstanding liability. Groundwater was chosen as the resource to be valued because it is commonly contaminated at these sites, and it was relatively straightforward to develop a rule of thumb for estimating the value of that damage. In NRD cases other than closed landfills, the MPCA does not use this simplified method but draws on a diverse toolkit of other NRDA methods.

The average cost of lost groundwater services is estimated as the amount by which it costs more to provide potable municipal water when the source groundwater is contaminated. The average cost to design, construct, operate, and maintain a groundwater treatment system over 30 years is calculated (based on information from six landfills in the program) to be 0.57 cents/gallon. The cost of supplying drinking water based on uncontaminated groundwater supplies is estimated (based on five municipalities) to be 0.1 cents per gallon. Thus, the average cost of groundwater services lost due to contamination is assumed to be 0.47 cents/gallon (the difference between the two numbers).

IV. EVALUATING THE STATES' SIMPLIFIED NRDA METHODS

The simplified methods described have some desirable features but lack some of the characteristics identified as desirable in section II. Here the authors evaluate the real-world simplified methods against the checklist of features; this discussion is summarized in Table 2.

Simple to Use

All of the methods described above have this feature. Data requirements are well specified and methods for using them to obtain monetary estimates of damages do not require specialized technical training.

Transparent

The New Jersey, Minnesota, and California assessment methods are reasonably transparent, because the water rate per gallon, the average cost of groundwater cleanup, and the average costs of habitat restoration can be adjusted with changes in the demand for water, inflation, and the costs of cleanup. However, the documentation for the methods used by Washington and Florida do not provide rationales for the choices of parameters used in those methods; this makes these methods more inflexible to such changes over time.

TABLE 2

Evaluation of Simplified Damage Assessment Methods

Damage Estimates Vary with Scope

The state methods all conform at least somewhat to this critrium. The damage assessments from oil spills in Washington and Florida are spill-specific, because the parameters used vary by size of spill and type of resource damaged, such as the physical nature of the oil, the quantity of oil, and the susceptibility of the resources to injury; these methods would more reliably pass the scope test, however, if they used measures of injury rather than volume of oil spilled, because injury and volume can be poorly correlated. Damage estimates from the Washington and Florida methods are also influenced by some factors related to the expected value of the injured resources. Specifically, the damage estimate will be higher, all else equal, if species of special importance (such as those thought to be endangered) are likely to have been affected; if damaged freshwater systems are important for benefits, such as municipal water supply, recreation, or wildlife habitat; i\f damaged recreational areas are visited by relatively large numbers of people; or if damaged freshwater wetlands are important (for example, those that harbor endangered species) or unusual. Similarly, damages to groundwater in New Jersey and Minnesota increase with an increase in the volume of groundwater contaminated and the expected value of the injured groundwater, represented either by the assessed price paid for municipal water in the area of the site or the average cost of providing potable water from alternative sources. Finally, damages estimated by the California method will increase with the duration and extent of the injury to natural resources.

Net Present Value Calculated Appropriately

Most of the states' methods do a poor job of accounting for injury duration. Neither the Washington nor the Florida method considers duration of injury explicitly while estimating damages, although the Washington method does include a persistence score to reflect persistence of injury. New Jersey and Minnesota consider groundwater damages to last for a maximum of 30 years. Given the slow-moving nature of many contaminants in aquifers and the irreversible impact of some contaminations, this may underestimate the duration of the injury. Unlike Minnesota, New Jersey does include the duration of injury explicitly in its calculation but considers the lost use value to be the same over the duration of the injury and estimates it simply by multiplying the per annum loss by the number of years of lost use, thereby ignoring discounting and possibly overestimating damages. The California method seems to score somewhat better on this point, because it accounts explicitly for the duration of the injury, and the underlying reduced-form analysis incorporates variation in duration in a manner that discounts streams of benefits over time appropriately.

Unbiased Estimate of Public Use and Nonuse Values

The damage assessment schedules in Washington and Florida impose absolute and relative values to species, habitats, and resources that may or may not reflect society's willingness to pay for them. For example, the correct damages associated with the deaths of endangered and threatened species may or may not be $10,000 and $5000, respectively as assigned by Florida (see Loomis and White, 1996, for some estimates of values of endangered species.) Similarly, although the habitat factors have the appearance of mean values per unit of resource damaged, it is not clear how these were determined. Additionally, many features of these assessment schedules seem to impose arbitrary and possibly undesirable structures on the damages that are assessed. For example, in the marine and Columbia River compensation schedules in Washington, vulnerability and sensitivity scores for a variety of potentially damaged resources are added together in a manner that forces the different uses to have identical weights. Thus, a spill that causes maximal damage to shellfish but leaves salmon untouched will be deemed as damaging as a spill of similar volume that causes maximal damage to salmon but leaves shellfish untouched. Similarly, the method used in Florida assumes that the marginal value of a species killed is constant; it is quite possible that the incremental value of a threatened creature is increasing or decreasing with the number of such creatures that are lost.

The methods used by New Jersey and Minnesota for valuing groundwater contamination have biases in their estimates of willingness to pay. Some of those biases may cause damages to be overestimated of damages. For example, both methods implicitly assume that all stocks of groundwater are (or will be) used as source for municipal water supplies (Desvousges et al., 1999), and the New Jersey method does not account for the fact that contaminated ground water is not devoid of value (it could be used after decontamination). On the other hand, other features may bias the estimates downward: The water rates used by New Jersey to proxy for the value of lost services are based on average (rather than marginal) costs and are chronically lower than market prices would be, and neither state's method provides a way to include nonconsumptive values (National Research Council, 1997).

The California method is based on the principle of identifying the cost of replacing lost natural-resource services rather than identifying social willingness to pay for those services. Although this method has been endorsed by one of the major federal trustees, it may well yield damage estimates that do not accurately represent true damages (Mazzotta et al., 1994; Unsworthand Bishop, 1994; Flores and Thacher, 2002). On the other hand, this method will not neglect to incorporate any element of damaged natural resource values because it finds the cost of providing equivalent natural- resource services to those that were lost.

Reflect Socioeconomic Characteristics of Affected Population

The New Jersey method may allow for some variation in preferences by including water rates as a measure of damages, and the Washington method does allow intensity of recreational use of different places to affect the implicit estimates of lost recreational use caused by spills. Otherwise, the simplified NRDA methods described do little to incorporate variation in consumer preferences and in willingness to pay among areas in which environmental damage occurs. The size of the population affected by natural resource damages is also not incorporated into any of the simplified methods described.

In sum, the authors cannot say that the simplified methods yield values that are even unbiased estimates of the true damages at hand. However, these state methods do yield damage estimates that covary in a useful manner with the scope of the accidents they are designed to evaluate, and they are desirable in that they are parsimonious in the time, expense, and data needed to use them. This latter virtue is not trivial. Assessment costs reduce net social welfare ceteris paribus. Furthermore, states with simplified methods added to their toolkits may be able to pursue more NRD cases and thus induce firms to engage in higher levels of precautionary activity. A comparison of Louisiana and Washington states supports this hypothesis. Louisiana has many more oil spills per year than does Washington-in 2000 there were 1704 spills in Louisiana and only 489 in Washington (U.S. Coast Guard, 2001). However, in their response to the survey, staff of the Louisiana Department of Environmental Quality did not report using a simplified method,10 and indeed reported that their agency has pursued only 13 cases from 1995 to 2001, whereas the Washington DECY pursued NRD recovery in 168 cases during the same time period (Ando et al., 2002).

V. ALTERNATIVE SIMPLIFIED APPROACHES FOR NRDA

The authors now briefly discuss two alternative approaches that are used by the National Oceanographic and Atmospheric Administration (NOAA) for NRDA. Benefit transfer methods have been allowed for NRDA by NOAA's Final Rule issued in 1996. The rule also provided guidelines for determining transfer appropriateness (NOAA, 1996a). NOAA also developed Type A NRDA methods in compliance with CERCLA regulations that mandate the development of a simplified method. Type A models are only to be used for minor and singular spills into coastal, marine, or Great Lakes waters of a substance that has been included in the models' databases. Estimates using Type A models will be granted rebuttable presumption if they are applied to appropriately small cases, which then are the subjects of litigation.

A. Benefit Transfer Method

The benefit transfer method uses information on environmental values at an existing study site to assess the value of damages to an environmental resource at a different policy site. Transfer of values to the policy site may take one of two forms: a direct transfer of the mean value for the environmental resource at the study site or a transfer of the estimated benefit function at the study site to the policy site. Data on socioeconomic characteristics of individuals and characteristics of resources at the policy site are used to compute the value of environmental damages to a resource at that site. Benefit transfer is based on the assumption that individuals studied at the original site have the same underlying preferences as individuals at the new site.

Benefit transfer methodology has the potential to provide a lower- cost, less time-intensive, and more practical approach for assessing economic values for environmental goods than other valuation approaches, such as contingent valuation and travel cost methods, which require primary data collection. However, the application of benefit transfer does require informed judgment on the several steps involved in transferring benefits to obtain damage estimates. These include identifying comparable and high-quality primary studies from which to transfer estimates, evaluating the applicability of these studies, obtaining a measure of value of damages from the willingness to pay estimates obtained by the original studies, and constructing aggregate estimates from measures available at a household level with little basis for judging the appropriate geographical extent of the affected population. A number of studies have supported the need to adjust willingness to pay values and value functions being transferred for differences such as those in the commodity (quality and quantity) being valued, socioeconomic characteristics, and welfare baseline conditions and other technical issues (Bergstron and De Civita, 1999). The methods for making these adjustments and implementing the benefit transfer approach is by no means straightforward, and different analysts can obtain considerably different values for the same resource, as shown in Smith (1992).

At a more practical level, trustees seeking to use benefit transfer to assessdamages to resources, for example for groundwater contamination, are likely to encounter several difficulties. First, the definition of the commodity that is typically valued by existing primary studies and that required to be valued for NRDA is often different. For example, in the case of studies valuing groundwater, the commodity valued by primary studies is typically stated as "protection of groundwater from nitrate or pesticides contamination" or "increased water supply protection," and its value is measured in terms of dollars per year that a household is willing to pay. In the case of NRDA of groundwater contaminated by a spill, the commodity is defined more quantitatively, in terms of gallons of water contaminated and area spanned by the contaminated plume of water. One must use models, assumptions, and expert judgment to convert the values of willingness to pay that emerge from economic valuation studies to a value per unit of the damaged resource (see Banzhaf et al., 1996, for a description of the methods used to transfer values of the externality costs of electricity generation).

Second, it has generally been found that using benefit function transfers rather than direct transfers of values improves accuracy of the approach. Functional transfers, however, require data for the sites to which the benefits are being transferred. Data requirements depend on the variables used in the original specification of the benefits function. In the case of groundwater quality studies, these variable include perceptions of groundwater safety, likelihood of future contamination, concern for water safety, and averting actions by households to avoid health risks in addition to socioeconomic characteristics of the population (Bergstrom et al., 2001). Transfers of benefits functions for NRDA would require trustees to gather data on these variables or make subjective assumptions about their values among the population at the contaminated site.

Finally, a major factor influencing the value of aggregate benefits from a resource is the geographical extent of the market affected by damages to that resource. Even small uncertainties in willingness to pay per household can lead to several fold differences in aggregate damages depending on the number of households assumed to be affected by resource injuries.

Studies examining the accuracy of alternative benefit transfer methods, for obtaining willingness to pay for improvements in groundwater quality, find that even under ideal conditions (that is, using similar survey instruments across sites, similar socioeconomic characteristics, and the same definition of the nonmarket commodity) benefits were unlikely to be effectively transferred (VandenBerg et al., 2001; Delavan and Epp, 2001). Smith (1992) calls for a need for better protocols and more research that would provide a well- stocked database of prices for nonmarket resources to make benefit transfer method more of an economic science than economic alchemy.

B. NOAA Type A Models

The Type A models-the NRDA Model for Coastal and Marine Environments and for Great Lakes Environments-are large computerized simulation models designed for use across the entire United States (Grigalunas et al., 1988; NOAA, 1996b). To generate an NRDA with one of these models, the user must have data on the following features of the spill: type of substance, amount spilled, time of spill, duration of release, location of release point, wind and tidal information, extent of cleanup, a reference year and price index to calculated dollar damages, and information on closing of commercial fisheries or recreational activities. This information is used as input for a complex set of linked physical, biological, and economic submodels. The model chooses the most technically feasible restoration response, calculates the cost of that response and the compensable values of lost uses that are not restored by the chosen restoration option. Total damages in U.S. dollars are calculated as the sum of the restoration cost and lost use values. Discounting is done using a discount rate of 3% (Hampton, personal communication).

These models have some of the best and worst features of the other methods discussed herein. They are simple to use and require little data to generate a dollar value of estimated NRDs. Use of these models gives a trustee a presumption of correctness in a court of law and thus greater bargaining power with a PRP. The models are transparent due to their extensive and public documentation, discounting is handled in a meaningful way, and the results certainly vary with the scope of a spill. However, there are some major weaknesses in these models. The damage estimates of traditional Type A models do not include nonuse values.11 The damage calculated includes a mixture of restoration costs and lost values, and although the lost use values are found using reasonably sophisticated benefit transfer, the combination of the two types of damages is difficult to identify as the theoretically correct measure of consumer surplus lost due to an accident.12 The damages that emerge from these models do not explicitly account for variation in the characteristics of the affected human populations. Perhaps most important, the models use parameters that are gross averages over very large geographic ranges. The marine and estuarine waters of the entire United States are divided by these models into only 10 different regions. Thus, there is almost no ability for variation in local conditions to influence the damage estimates.

In sum, these two national simplified methods are not necessarily superior to the states' methods. Though benefit transfer has the potential to be more consistent with desired economic principles than the states' methods, its implementation is not straightforward. Type A models, though simple to apply, are based on parameters that are gross averages over such large geographic ranges that they cannot provide damage estimates that are site-specific.

VI. CONCLUSIONS

Economists have contributed much to the process of policy analysis, development, and evaluation by developing state-of-the- art valuation methods that measure a broad range of nonmarket values with impressive accuracy. However, case-specific methods, such as contingent valuation or travel-cost analysis, are expensive, time- consuming, and data-intensive. As such, those methods are simply not a good fit for the needs of the government trustees that seek to hold firms liable for the natural resource damages that result from small releases of hazardous materials into the environment. State agencies themselves have developed simplified methods for NRDA that can be implemented quickly, at low cost, and without highly sophisticated technical training. Although the availability of these methods enable states to pursue NRD compensation and create incentives for precautionary behavior by firms to avoid liability for damages, these methods are flawed in ways that are likely to induce chronic bias in the estimates they yield. Thus, there could be a large payoff to society if simplified methods that are less egregiously at odds with economic theory of valuation could be used instead.

Methods to assess damages for use in NRDA do not need to meet the standards of accuracy established by primary-study methods. To induce firms to engage in optimal ex ante damage avoidance activities, the assessment method need only be unbiased (such that a firm's expected payment is equal to the damage it causes) and accurate enough to forestall successful challenges in a court of law. Indeed, if a simplified method is codified by law (as in Washington and Florida) they can be inoculated against legal challenge in cases of modest scope. Further research is needed that focuses on designing simplified methods that are less biased than the simplified methods currently in use by some states without compromising the ease of implementation of the simplified state methods. The criteria set forth in this article can serve to guide that future research.

1. See State of Montana Natural Resource Damage Program (2003) for details on this case. The Atlantic Ritchfield Corporation has partially settled with the trustees for $129 million in NRDs; further litigation is pending.

2. There have been 5500 studies and papers from over 100 countries using contingent valuation methods alone (Carson, forthcoming).

3. These are sites placed on the National Priorities List under CERCLA.

4. The information in this section was provided by Steve Hampton in the California Department of Fish and Game (personal communication).

5. Other agencies in California may collect damages as trustees, but the office is usually most prominent and works cooperatively with the other agencies when they are involved.

6. The agency also has access to a Type A model that was designed by a consulting company for the state of California in 1998, but staff members have not used this method very often and do not expect it to become a more prominent part of their toolbox in the future.

7. This representation of the schedule comes from Nick Stratis in the Florida DEP.

8. The regulations specifying the details of the compensation are roughly 100 pages long, with many pages of detailed scores for different items in the various geographical areas.

9. More information on the method used by New Jersey for assessing natural resource injuries to resources, such as fisheries, sediments, wetlands, forests, wildlife habitat, wildlife, and lost public use/recreation can be found online at www.state.nj.us/dep/ nrr/nri/nri_eco.htm.

10. Note that there are other agencies in Louisiana that might have pursued NRD cases, but the number of additional cases is not likely to be large.

11. The Type A model developed for the state of California added nonuse values in an ad hoc manner (Hampton, personal communication), so this omission need not be a necessary fe\ature of such models.

12. This approach to damage assessment is supported by some of the regulatory frameworks; see Dunford (1992) for a useful discussion of this matter.

REFERENCES

Ando, A. W., and W. Polasub. "Envelope Backs or the Gold Standard: Choosing the Accuracy of Natural Resource Damage Assessment Methods." UIUC Working Paper, 2003.

Ando, A., and A. Wildermuth. "Natural Resource Damages Legal Overview and Sample cases." In Natural Resource Damage Assessment: Methods and cases, edited by A. W. Ando, M. Khanna, A. Wildermuth, and S. Vig. Report to the Illinois Department of Natural Resources, Waste Management and Research Center, 2004.

Ando, A. W., M. Khanna, A. Wildermuth, and S. Vig. Natural Resource Damage Assessment: Methods and cases. Report to the Illinois Department of Natural Resources, Waste Management and Research Center, 2004.

Banzhaf, H. S., W. H. Desvousges, and F. R. Johnson. "Assessing the Externalities of Electricity Generation in the Midwest." Resource and Energy Economics, 18, 1996, 395-421.

Becker, G. S. "Crime and Punishment: An Economic Approach." Journal of Political Economy, 76, 1968, 169-217.

Bergstrom, J. C., and P. De Civita. "Status of Benefits Transfer in the United States and Canada: A Review." Canadian Journal of Agricultural Economics, 47(1), 1999, 79-87.

Bergstrom, J. C., K. J. Boyle, and G. L. Poe. The Economic Value of Water Quality. Cheltenham, UK: Edward Elgar, 2001.

Brookshire, D. S., and D. O. Scrogin. "Reflections upon 25 Years of the Journal of Environmental Economics and Management." Journal of Environmental Economics and Management, 39(3), 2000, 249-63.

Carson, R. Contingent Valuation: A Comprehensive Bibliography and History. Northampton, MA: Edward Elgar, forthcoming.

Delavan, W., and D. J. Epp. "The Case of Nitrate Contamination in Pennsylvania, Georgia and Maine." In The Economic Value of Water Quality, edited by J. C. Bergstrom, K. J. Boyle, and G. L. Poe. Cheltenham, UK: Edward Elgar, 2001, pp. 100-120.

Desvousges, W. H., R. W. Dunford, K. E. Mathews, C. L. Taylor, and J. L. Teague. A Preliminary Economic Evaluation of New Jersey's Proposed Groundwater Damage Assessment Process. Prepared for New Jersey Site Remediation Industry Network by Triangle Economic Research, Durham, NC, 1999.

Dunford, R. W. "Natural Resource Damages from Oil Spills." In Innovation in Environmental Policy: Economic and Legal Aspects of Recent Developments in Environmental Enforcement and Liability, edited by T. H. Tietenberg. Brookfield, VT: Ashgate, 1992.

Dunford, R. W., R. F. Johnson, and E. S. West. "Whose Losses Count in Natural Resource Damages?" Contemporary Economic Policy 15(4), 1997, 77-87.

Environmental Defense. Pollution Locator; Land Contamination; Rank States. Available online at www.scorecard.org/env-releases/ land/rank-states.tcl, 2002.

Flores, N., and J. Thacher. "Money, Who Needs It? Natural Resource Damage Assessment." Contemporary Economic Policy, 20(2), 2002, 171-78.

Grigalunas, T. A., J. J. Opaluch, D. French, and M. Reed. "Measuring Damages to Marine Natural Resources from Pollution Incidents under CERCLA: Application of an Integrated Ocean Systems/ Economic Model." Marine Resource Economics, 5, 1988, 1-21.

Hanemann, M. W. "Valuing the Environment through Contingent Valuation.", Journal of Economic Perspectives, 8(4), 1994, 1-43.

Harrison, G. W., and J. C. Lesley. "Must Contingent Valuation Surveys Cost So Much?" Journal of Environmental Economics and Management, 31(1), 1996, 79-95.

Helton, D., and T. Penn. Putting Response and Natural Resource Damage Costs in Perspective. Paper ID #114, 1999 International Oil Spill Conference. Available online at www.darp.noaa.gov/pdf/ costsofs.pdf,1999.

Kaplow, L., and S. Shavell. "Accuracy in the Assessment of Damages." Journal of Law and Economics, 39, 1996, 191-210.

Loomis, J. B., and D. S. White. "Economic Benefits of Rare and Endangered Species: Summary and Meta-Analysis." Ecological Economics, 18, 1996, 197-206.

Mazzotta, J. J., J. J. Opaluch, and T. A. Grigalunas. "Natural Resource Damage Assessment: The Role of Resource Restoration." Natural Resources Journal, 34, 1994, 153-78.

Minnesota Pollution Control Agency (MPCA). Closed Landfill Program-Insurance Recovery Effort. Available online at www.pca.mn.us/ cleanup/landfillclosed.html, 2002.

National Research Council. Valuing Ground Water: Economic Concepts and Approaches. Washington, DC: National Academy Press, 1997.

New Jersey Department of Environmental Protection (NJDEP). Frequently Asked Questions about Natural Resource Damages in New Jersey. Office of Natural Resource Damages, NJDEP, 1999.

National Oceanic and Atmospheric Administration. "Natural Resource Damage Assessments: Final Rule." Federal Register, 5 January 1996a, 440-510.

_____. "Natural Resource Damage Assessment Final Rule: Type A Procedures." Federal Register, 7 May 1996b, 440-510.

Randall, A. "Whose Losses Count? Examining Some Claims about Aggregation Rules For Natural Resources Damages." Contemporary Economic Policy, 15(4), 1997, 88-97.

Ruotsinoja, S. Natural Resource Damages Calculations for Closed Landfill Program Insurance Settlements. State of Minnesota Office Memorandum, 1997.

Segerson, K. "Liability for Environmental Damages." In Principles of Environmental and Resource Economics: A Guide for Students and Decision Makers. Cambridge, MA: Edward Elgar, 2000.

Smith, V. K. "On Separating Defensible Benefit Transfers from 'Smoke and Mirrors.'" Water Resources Research, 28(3), 1992, 685- 94.

State of Florida. Pollutant Discharge, Prevention, and Removal. Title XXVII Florida Statues, chapter 376, 2001.

State of Montana Natural Resource Damage Program (NRDP). Fact Sheet on Upper Clark Fork River Basin Natural Resource Damage Litigation and Restoration. Available online at www.clarkforkrestoration.com, 2003.

State of New Jersey. Technical Requirements for Site Remediation. New Jersey Administrative Code 7:26E, 1999.

State of Washington. Preassessment Screening and Oil Spill Compensation Schedule Regulations. Title 173 Washington Administrative Code, chapter 173-183, 1992.

_____. Water Pollution Control. Revised Code of Washington, chapter 90.48, 2001a.

_____. Oil and Hazardous Substance Spill Prevention and Response. Revised Code of Washington, chapter 90.56, 2001b.

U.S. Coast Guard. Oil Spills in U.S. Waters. Available online at www.uscg.mil/hq/g-m/nmc/response/stats/chpt2000.pdf, 2001.

Unsworth, R. E., and R. C. Bishop. "Assessing Natural Resource Damages Using Environmental Annuities." Ecological Economics, 11, 1994, 35-41.

VandenBerg, T. P., G. L. Poe, and J. R. Powell. "Assessing the Accuracy of Benefits Transfers: Evidence from a Multi-site Contingent Valuation Study of Ground Water Quality." In The Economic Value of Water Quality, edited by J. C. Bergstrom, K. J. Boyle, and G. L. Poe. Cheltenham: Edward Elgar, 2001, pp. 100-120.

Washington Department of Ecology. Focus Sheet: Assessing Oil Spill Damage. Available online at www.ecy.wa.gov/biblio/ 0208004.html, 2002.

AMY W. ANDO and MADHU KHANNA*

* This is a revision of a paper presented at the Western Economics Association, International 78th annual conference, Denver, 12 July 2003, in a Contemporary Economic Policy invited session. The authors gratefully acknowledge funding from the Waste Management and Research Center of the Illinois Department of Natural Resources. This material is also based in part on work supported by the Cooperative State Research, Education, and Extension Service, U.S. Department of Agriculture, under project no. ILLU 05-0305. The authors are grateful to John Braden, Nick Flores, Steve Hampton, Peter Schwarz, two anonymous referees, and staff members at the Illinois Department of Natural Resources for helpful comments and advice. They thank Wallapak Polasub for research assistance and the staff members of the agencies that worked to provide information about their programs. Any errors remain the responsibility of the authors.

Ando: Assistant Professor, Dept. of Agricultural and Consumer Economics, 326 Mumford Hall, 1301 W. Gregory Dr., University of Illinois at Urbana-Champaign, Urbana, IL 61801. Phone 1-217-333- 5130, Fax 1-217-333-5538, E-mail amyando@uiuc.edu

Khanna: Associate Professor, Dept. of Agricultural and Consumer Economics, 326 Mumford Hall, 1301 W. Gregory Dr., University of Illinois at Urbana-Champaign, Urbana, IL 61801. Phone 1-217-333- 5176, Fax 1-217-333-5538, E-mail khannal@uiuc.edu

Copyright Western Economic Association Oct 2004

Source: Contemporary Economic Policy

More News in this Category