The high cost of cleanups of hazardous waste sites under the Superfund program has prompted critics of the program to question whether the benefits of cleanups are worth what the U.S. Environmental Protection Agency (EPA) says should be spent to achieve them. Recently, researchers at Resources for the Future and the University of Maryland examined the agency's decisions regarding the cleanup of 110 Superfund sites in order to determine whether the factors that influenced the agency's choice of target risk level (the lifetime cancer risk posed by a site after cleanup) and choice of cleanup option (how permanent the cleanup should be) for each site were related to benefit and cost considerations. Their study reveals that EPA balanced benefits and costs to the extent that it was legally entitled to do so. Other factors being equal, the cheaper the cost of a cleanup option, the more likely it was that the agency chose it. However, EPA clearly preferred more permanent remedies to less permanent remedies. The question remains whether the benefits of more permanent remedies warrant the costs of obtaining them.
Passage of the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA), commonly referred to as Superfund, brought about what has become the best-known program for cleaning up hazardous waste sites in the United States. One of the most controversial of all U.S. environmental programs, Superfund has been widely criticized for the slow pace at which remedial actions at such sites are undertaken. It has also come under fire for leading to lengthy and costly litigation as a result of its liability standards, which many parties potentially responsible for cleanup costs consider unfair.
Not least among the criticisms of Superfund, which is administered by the U.S. Environmental Protection Agency (EPA), is the high cost of cleanups under the program. To date, it has averaged $27 million per site. Estimates of the costs to clean up all sites currently on EPA's National Priorities List (NPL) and those likely to be put on the NPL range from $60 billion to $90 billion. In light of these estimates, many question whether the benefits of the Superfund program justify the costs of the program. At issue is how EPA determines the amount of money that should be spent on the cleanup of each Superfund site. Does the agency make this determination on the basis of risks to human health and the environment, or on some other basis?
To answer this question, we recently examined EPA's decisions regarding the cleanup of contaminated soils at 110 Superfund sites. Specifically, we examined the factors that influenced the agency's decisions about how much soil to clean up and how the cleanup was to be effected at each of the sites. In doing so, we investigated whether these factors were related to the benefits and costs of cleanups and whether political variables, as measured by the socioeconomic characteristics and racial composition of the population living near a site, influenced the choices of cleanup strategy at the 110 sites.
If those choices reflected the benefits of cleanup, it would be expected that cleanups at sites where these benefits were great would be both more comprehensive and more permanent. Other factors being equal, the benefits of cleanup should be greater at sites located in densely populated areas than at sites located in sparsely populated areas, for the simple reason that more people are potentially exposed to contaminated soils in the former. To the extent that society considers it important to clean up sites that pose the highest risks, the benefits of cleanups should also be greater at sites where such risks are higher than at sites where they are lower.
If EPA's choices of cleanup strategy reflected the costs of cleanup, it would be expected that, other factors being equal, cleanups at sites where these costs were great would be both less comprehensive and less permanent.
The cleanup process under Superfund
When EPA learns of abandoned hazardous waste sites, it enters them into a computerized database called CERCLIS (Comprehensive Environmental Response, Compensation, and Liability Information Service), which currently contains entries on more than 33,000 sites. The agency then undertakes a preliminary assessment of each site in which it attempts to determine how the site was used and which hazardous substances were disposed of there. In many cases, this assessment indicates that no remedial action is required. In about one-third of the cases, it indicates the need for a site investigation, in which a more detailed assessment of the site is made on the basis of a visual examination and a laboratory analysis of materials found at the site. After a site investigation, many sites are judged harmless enough to warrant no further action.
Sites not so judged go through a formal hazard ranking process in which they are evaluated on the basis of their potential to inflict damage through three pathways—groundwater, surface water, and air. Using its hazard ranking system, EPA assigns each site a numerical score on a scale from 0 to 100. If a site's score is greater than 28.5, the site is put on the National Priorities List. At the end of fiscal year 1991, there were more than 1,200 sites on the list.
Once EPA has selected cleanup goals and remedial actions that protect public health and the environment and that comply with state and federal environ-mental standards, it is allowed to balance the cost of cleanup against other evaluation cri-teria, such as the permanence of remedial actions.
Each site on the NPL is subject to a formal Remedial Investigation/Feasibility Study (RI/FS). RI/FSs may be conducted for individual portions of each large site. In the Remedial Investigation, the wastes at the site are identified and the risks they pose to human health and the environment are assessed. In the Feasibility Study, remedial alternatives are developed and screened.
In most cases, the Feasibility Study must address two pollution problems: surface contamination—contamination of soil, sludge, or surface water—and groundwater contamination. The cleanup options for contaminated soil include capping, in which materials such as clay and asphalt are placed over the soil; in situ treatment, in which the soil is not excavated but treated in place with chemicals or microorganisms that detoxify contaminants; or excavation. Soils that are excavated can be disposed of in a landfill or can be treated by some method such as incineration. The most common cleanup option for contaminated groundwater is to pump the water to the surface, treat it to remove contaminants, and either reinject it into the aquifer from which it was pumped or discharge it into a river or stream. In some cases the use of an alternative water supply is considered.
Following the Feasibility Study, EPA issues a proposed remedial action plan, which describes and justifies the remedial alternative selected. The public, private parties, and state and local governments are then given thirty days to comment on the proposed plan, after which the plan is issued as a Record of Decision. At this point, the actual cleanup of the site can begin. Once the cleanup is completed, the site can be deleted from the NPL.
It is important to note that EPA must not only protect public health and the environment but comply with state and federal environmental standards—the "Applicable or Relevant and Appropriate Requirements"—in selecting cleanup goals and remedial actions. After these standards have been met, the agency is allowed to balance the cost of cleanup against four other evaluation criteria: (1) permanence of remedial actions; (2) short-term effectiveness of the actions; (3) reduction of toxicity, mobility, or volume of waste; and (4) feasibility of the cleanup strategy.
Scope of study
In studying the role that cost and benefit considerations play in the choice of cleanup strategies at Superfund sites, we limited our investigation to EPA's cleanup decisions at sites for which one or more Records of Decision (RODs) had been signed. By the end of the 1991 fiscal year, 945 RODs had been signed. We further limited our investigation to EPA's cleanup decisions at sites where soil contained chemicals used in the treatment of lumber (wood-preserving sites) and at sites where soil contained polychlorinated biphenyls (PCBs) at concentrations in excess of ten parts per million (PCB sites). Thus our study was restricted to cleanup decisions at 110 sites—32 wood-preserving sites and 78 PCB sites. Because EPA chose different cleanup strategies for different parts of some of these sites, we examined 127 cleanup decisions in all—40 at wood-preserving sites and 87 at PCB sites.
We chose to investigate cleanup decisions at two types of sites in order to determine if the influence of cost and benefit considerations was the same at one as at the other. Our decision to focus on wood-preserving sites and PCB sites was prompted by the similarity of their contaminants. Like PCBs, creosote and pentachlorophenol—one or both of which are found in the soil at wood-preserving sites—are organic chemicals. Therefore, the menu of treatment options is similar for wood-preserving sites and PCB sites. Because all three contaminants are suspected human carcinogens, we were able to collect quantitative estimates of health risks in order to determine whether these risks were reflected in the choice of cleanup strategy.
In our study, we examined EPA's two major decisions with respect to the cleanup at each site: how much of the soil should be cleaned up and how the cleanup should be effected. The first decision—how extensive the cleanup should be—involves determining how much of the site to cap, or, if excavation is chosen, how much of the soil to excavate. This decision is usually stated in terms of the maximum concentrations (in parts per million) of toxic substances that will be left at the site once the cleanup is completed. These concentrations place an upper bound on the lifetime cancer risk arising from exposure to contaminated soil at the site. Choosing maximum concentrations of contaminants is therefore tantamount to choosing a target health risk. In choosing a target risk level, EPA is to protect the health of persons living near the site, regardless of cost. It is free to decide how low the target health risk should be at a site, although guidelines suggest that lifetime cancer risk should lie between 1 in 10,000 and 1 in 1,000,000.
As noted above, the second decision involves choosing one or more of three options: capping the soil, treating it in situ, or excavating it. If the last is chosen, EPA must then decide whether the excavated soil will be put in a landfill or treated in some way (such as incineration). This decision is essentially a determination as to how permanent the cleanup will be. If the contaminated soil is capped rather than excavated, for example, the cleanup will last only as long as the life of the cap. If the excavated soil is put in a landfill rather than being incinerated, the cleanup will last only as long as the landfill liner remains intact.
Choice of target risk level
EPA quantified the post-cleanup level of lifetime cancer risk it wanted to achieve in 61 of the 127 decisions we examined. Parties potentially responsible for cleanup costs (so-called potentially responsible parties, or PRPs) paid for RI/FSs on which 29 of these decisions were based. State funds or federal tax revenues earmarked for the purpose were used to pay for RI/FSs on which the remaining 32 decisions were based. Since it is sometimes thought that the party who pays for an RI/FS can influence the menu of cleanup options considered at a site, we examined whether the factors that affected the target risk level chosen for enforcement-lead sites (where RI/FSs were paid for by potentially responsible parties) differed from those that affected the level chosen for fund-lead sites (where RI/FSs were paid for by government funds).
As stated earlier, at least two factors might influence the choice of target risk level at a site if benefits mattered in deciding which cleanup options to implement. One factor is baseline risk—that is, the risk that a site poses to an individual living near the site prior to cleanup. The other is the size of the population exposed to contamination from the site. Other factors being equal, EPA might be expected to choose lower target risk levels for sites with a high baseline risk than for sites with a low baseline risk. Likewise, the agency might be expected to choose lower target risk levels for sites in densely populated areas than for sites in sparsely populated areas, because more people would benefit from a reduction in risk in densely populated areas.
Baseline risk (the risk to health posed by a site prior to cleanup) influenced EPA's choice of target risk level only at enforcement-lead sites, where remedial investigations and feasibility studies were paid for by parties potentially liable for cleanup costs.
In our study, we approximated the size of the population living near a site by noting whether the site was located in an urban, a suburban, or a rural area. We found that the influence of population on the choice of target risk level was no different at enforcement-lead sites than it was at fund-lead sites. Somewhat surprisingly, we did find that the target risk levels selected for sites in urban areas were higher than those selected for sites in suburban or rural areas. One possible explanation for our finding is that residents living near the sites perceived some health threat arising from the excavation of soil—a remedial action that is usually entailed by the choice of a low target risk level. If so, EPA may have chosen a higher rather than a lower target risk level at sites in urban areas so as to avoid a cleanup strategy that involved soil excavation.
While the influence of population on the choice of target risk level was the same at enforcement-lead sites and fund-lead sites, we found that the influence of baseline risk was not. EPA chose high target risk levels at enforcement-lead sites that posed a low baseline risk. This risk had no effect on the choice of target risk level at fund-lead sites. At these sites, however, the target risk level was related to a site's hazard ranking system score: the lower the score, the higher the target risk.
As noted above, EPA is supposed to ignore costs in setting target risk levels. It appears to have done just that. At both fund-lead sites and PRP-lead sites, target risk levels were unrelated to either unit excavation costs or to the size of the site. Target risk levels were also unrelated to political variables, as measured by the racial composition of and the median income in the county where a site is located.
Choice of cleanup options
At the sites, EPA had three choices regarding cleanup options: whether or not to excavate soils; whether or not to treat soils; and whether to dispose of excavated soil off site or on site. These choices can be combined to yield six distinct remedial alternatives: on-site treatment of soil that has been excavated (on-site treatment), off-site treatment of soil that has been excavated (off-site treatment), disposal of excavated but untreated soil in a landfill off the site (off-site landfill), disposal of excavated but untreated soil in a landfill on the site (on-site landfill), on-site treatment of soil that has not been excavated (in situ treatment), and containment of soil that has been neither excavated nor treated (containment) (see figure, p.16).
We estimated the parameters of a model that explains the likelihood that EPA would select each of the six cleanup options. The model describes how much more likely the agency is to select one option rather than another as a function of the cost of the options and the characteristics of the site—its location, the baseline risk it poses, and the volume of contaminated waste it contains. One reason for estimating the parameters of the model is to determine if variables such as cost and baseline risk can explain the cleanup option chosen. The model also enables us to compute, in dollar terms, how much more valuable EPA considers one option to be than another. For example, we can estimate how much more EPA would be willing to spend (or have PRPs spend) to incinerate a given volume of waste rather than to cap it.
When we estimated the parameters of the model for wood-preserving sites and then for PCB sites, we found that costs were significant in explaining the choice of cleanup option at both types of sites. Other factors being equal, we found in both cases that EPA was likely to choose less expensive cleanup options. Unlike costs, baseline risk and the location of the site appear to have had little effect on the agency's choice of cleanup option at either type of site. However, we found that higher hazard ranking scores were positively correlated with a preference for more permanent treatment options—incineration, for example—at PCB sites.
Other factors being equal, EPA was likely to choose less expensive cleanup options at sites contaminated with polychlorinated biphenyls (PCB sites) and at sites where chemicals had been used in the treatment of lumber (wood-preserving sites).
Remedial alternatives for soil contamination
By law, EPA is supposed to show a preference for treatment options, as opposed to containment options, and for on-site remedies, as opposed to off-site remedies. Our study revealed that it chose on-site treatment of excavated soil more often than the five other cleanup options. Moreover, it revealed that the agency was willing to pay (or have PRPs pay) at least twice as much for on-site incineration of contaminated soil than for capping of the soil at both PCB sites and wood-preserving sites.
If a PCB site were to contain 10,000 cubic yards of contaminated soil, we estimated that EPA would be willing to spend (or have PRPs spend) $12 million (in 1987 dollars) more to incinerate the soil than to cap it. If a PCB site were to contain 100,000 cubic yards of contaminated soil, we estimated that the agency would be willing to spend (or have PRPs spend) $36 million more to incinerate the soil than to cap it. It is interesting that EPA appears to be willing to spend (or have PRPs spend) almost as much for off-site incineration as for on-site incineration at PCB sites, suggesting that it finds little difference in the utility assigned to the two alternatives at PCB sites.
For wood-preserving sites, as for PCB sites, we estimated that the agency was willing to spend (or have PRPs spend) more for on-site incineration than for capping. At a typical wood-preserving site, we estimated that EPA would be willing to spend an additional $12 million (in 1987 dollars) to incinerate the soil. However, EPA did not demonstrate a willingness to spend nearly as much for off-site incineration as for on-site incineration at wood-preserving sites. Its willingness to pay for off-site incineration at these sites was not significantly different than its willingness to pay for capping.
Interpreting the results
From the perspective of public policy, it is important to ask whether the benefits of Superfund cleanups justify the amount of money EPA is willing to spend (or to have PRPs spend) on cleanups. Because the information collected in RI/FSs is insufficient to conduct a conventional benefit-cost analysis of EPA's cleanup decisions, we attempted to determine whether benefits and costs were correlated with the agency's choice of cleanup goal (target risk level) and cleanup option (permanence of cleanup) at wood-preserving sites and PCB sites for which Records of Decision had been signed.
As we have seen, there is no evidence that EPA balanced benefits against costs in selecting a target risk level for these sites. However, it would be incorrect to infer from this that the agency did not do its job. Under CERCLA, the target risk level chosen for a Superfund site must be one that protects the health of the population, regardless of cost. Thus, by ignoring cost in its selection of target risk levels, EPA acted in accord with its mission.
CERCLA does allow EPA to balance the cost of cleanup against other goals, including the permanence of cleanup, once a safe target risk level has been cho-sen. At the sites we studied, the agency did this. Other factors being equal, the cheaper the cost of a cleanup option, the more likely it was that EPA chose it. However, the agency was clearly willing to pay (or have PRPs pay) more for more permanent remedies than for less permanent remedies. As noted above, it was willing to pay more to incinerate contaminated soil than to cap it.
Variables associated with the benefits of cleanup, such as the location of the site (size of exposed population) and the health risk associated with the site before cleanup (baseline risk), were only weakly associated with the target risk level chosen, and were unrelated to the permanence of the cleanup that the chosen cleanup option would achieve.
In assessing the 127 cleanup decisions considered in our study, we found that EPA followed its mandate: it reduced baseline health risks without regard to cost, but it was willing to balance the costs of cleanups against the permanence of cleanups in choosing a cleanup option. What must be asked is whether the benefits of more permanent cleanups, such as those achieved by the incineration of contaminated soil, are worth the amount that the agency is willing to pay (or have PRPs pay) for them. To answer this question, it will be necessary to first define and then value the benefits of alternative waste disposal technologies. In view of the size of the resources devoted to Superfund cleanups, this is research that deserves the very highest priority.
Shreekant Gupta and George Van Houtven are doctoral candidates in the Department of Economics at the University of Maryland. Maureen L. Cropper, a senior fellow in the Center for Risk Management at Resources for the Future, is a professor in the department.
A version of this article appeared in print in the May 1993 issue of Resources magazine.
A version of this article appeared in print in the May 1993 issue of Resources magazine.
