Estimating Environmental Liability: Quantifying the Unknown

An extensive psychological literature documents how fallible human judgment under uncertainty can be. See Russo J.E. Schoemaker P.J.H. , Decision Traps (New York, NY: Doubleday, 1989). In this case, the inability to imagine all pathways to trouble requires careful analysis, mental simulation, and categorization. Kahneman D. Tversky A. , “The Simulation Heuristic,” in Kahneman D. Slovic P. Tversky A. , eds., Judgment Under Uncertainty: Heuristics and Biases (New York, NY: Cambridge University Press, 1982), pp. 201–210. Whereas people are normally overconfident in their estimates [ Russo J.E. Schoemaker P.J.H. , “Managing Overconfidence,” Sloan Management Review, 33/2 (Winter 1992): 7–18], systematic attempts to generate reasons and challenge beliefs tend to lead to better calibration of subjective probabilities. See Koriat A. Lichtenstein S. Fischhoff B. , “Reasons for Confidence,” Journal of Experimental Psychology: Human Learning and Memory, 6 (1980): 107–118; or Lichtenstein S. Fischhoff B. Phillips L.D. , “Calibration of Probabilities: The Slate of the Art to 1980,” in Kahneman , eds. (1982), pp. 306–334.

See Surma J.P. Vondra A.A. , “Accounting for Environmental Costs: A Hazardous Subject,” Journal of Accountancy (March 1992), pp. 51–55. The legal reporting requirements are governed by various SEC and FASB regulations. In general, the SEC requires the reporting of any events or conditions that may materially affect the firm's value or solvency (see Regulation S- K of the Securities Exchange Act, specifically items 101 and 103). Disclosure without recognition as to financial liability (e.g., on the balance sheet or income statement) is consistent with Generally Accepted Accounting Principles (GAAP) if the probability and estimability of loss are ambiguous. SFAS 5 of GAAP only requires that a liability be accrued and expensed if the loss is “probable” and the amount of loss can be “reasonably estimated.”

See Barth Mary E. McNichols Maureen F. , “Estimation and Market Valuation of Environmental Liabilities Relating to Superfund Sites,” Harvard and Stanford University Working Paper, April 1994. The authors show that in the following multiple regression, Market Value = a₀ + a₁ Assets + a₂ Liabilities + a₃ NPL, the coefficient a₃ was significantly (p<.005) negative. The variable NPL is a proxy for environmental liabilities that were not quantified in the firm's balance sheet but were in the public domain via the EPA's published list of firms notified as PRPs. A measure as simple as the number of sites in which a firm is a PRP, without any estimates of its share of remediation costs, produced a significant negative correlation. Overall, the number of sites, their composition and the extent of soil contamination proved to be the strongest predictors of total cleanup costs, but the site specific characteristics explained only a small portion of the variance (adjusted R²≤<.2).

For example, when Robins A.H. disclosed on March 19, 1975 that it had been named in 186 products liability lawsuits over the Shield Dalkon (an intrauterine device implicated in pelvic inflammatory disease), its market value dropped by $18.3 million (or 6.2%) in one day. A second announcement on February 19, 1976 of a rise in lawsuits to 547, entailing total claims of $444 million, knocked down another $49.2 million of its market capitalization. See Viscusi W. K. , Reforming Products Liability (Cambridge, MA: Harvard University Press,. 1991).

See Kleindorfer P.R. Kunreuther H.C. , eds., Insuring and Managing Hazardous Risks: From Seveso to Bhopal and Beyond (Berlin: Springer-Verlag, 1987).

Roger Marzulla cites the cost of Superfund enforcement actions alone to be 24% to 44% of total cleanup costs. Marzulla R. , “Superfund ‘91—Congress’ Chance to Clean Up Its Act,” Risk Management (April 1990), pp. 32–40. A RAND Corporation study found that between 1986 and 1989, insurance companies spent $1.3 billion on Superfund litigation—of which, $1.2 billion went directly to lawyers.

Superfund Report, “Treasury Department Suggests Scrapping Current Liability Scheme,” September 8, 1993, pp. 3–5; see also “New Superfund Law Proposed,” EPA Journal, 20 (1994): Pp. 2–3.

See Russell M. Colglazier E.W. English M.R. , Hazardous Waste Remediation: The Task Ahead, Waste Management Research and Education Institute, University of Tennessee, Knoxville, TN, December 1991; Colglazier E.W. Cox T. Davis K. , Estimation of Resource Requirements For NPL Sites, Waste Management Research and Education Institute, University of Tennessee, Knoxville, TN, December 1991; and Waste Management Research and Education Institute, “State and Private Sector Cleanups” and “Estimation of Resource Requirements for NPL Sites,” University of Tennesse, Knoxville, TN, December 1991.

Office of Technology Assessment, “Coming Clean: Superfund Problems Can Be Solved,” Washington, D.C., October 1989.

This may be due to internal bias, the age of the study (the OTA study is two years older than the Tennessee study, when less data were available), and/or because the OTA did not consider multiple scenarios regarding cleanup technologies and levels of funding.

Environmental Protection Agency, “Superfund: Focusing on the Nation at Large—1991 Update,” EPA/540/8–91/016, September 1991.

The issue of cost growth is quite uncertain and debatable. On the one hand, improvements in technology, economies of scale, and learning curve effects could actually make cost growth negative. Numerous cases can be cited in which per unit costs declined over time, ranging from automobile manufacturing to home building (when controlling for changes in raw material and labor costs). On the other hand, costs in many walks of life (from medical treatment to legal representation) have grown faster than inflation. In the absence of a clear direction, we decided to ignore cost growth.

The real rate of interest was stable around 2% from 1959–1965 and turned negative from 1974–1980 (while nominal rates skyrocketed due to oil price hikes and inflation). After 1980, the real rate turned strongly positive but has been gradually declining ever since. See Shapiro A.C. , Modern Corporate Finance (New York, NY: Macmillan Publishing Co., 1989), p. 50.

See Frantz R.W. Weisshaar M.G. , “The Next Generation of Hazardous Wast Site Cases—Toxic Tort Suits,” Natural Resources and Environment, 2 (1987): 34–35.

See National Research Council, Environmental Epidemiology-Public Health and Hazardous Wastes. (Washington, D.C.: National Academy Press, 1991).

Weber O.J. , “Toxic Torts—Current Developments,” Federation of Insurance and Corporate Counsel Quarterly, 36 (1986): 189–208.

Viscusi W.K. , “Compensating Workplace Toxic Torts,” Proceedings of the Academy of Political Science, 37/1 (1988): 126–138; and Viscusi (1991), op.cit.

Kraus N. Malmfors T. Slovic P. , “Intuitive Toxicology: Expert and Lay Judgments of Chemical Risks,” Risk Analysis, 12/2 (June 1992): 215–232.

Our analysis focused especially on one of the dioxins known as 2,3,7,8-tetrachloro-dibenzo-p-dioxin, which has received a large amount of scientific scrutiny. In August of 1994, the EPA released its most recent, reassessment of dioxin's health (EPA/600/BP-92/001a-c) and exposure data (EPA/600/6-88/005Ca-c) for external public review.

See Agency for Toxic Substances and Disease Registry, Toxicological Profile for Chloroform, January 1989.

Good sources of information on health effects of chemicals are the Agency for Toxic Substances and Disease Registry(ATSDR), National Technical Information Service(NTIS), and the National Library of Medicine's Toxicology Information Program (TOXNET).

See Bertazzi P.A. Pesatori A.C. Consinni D. Tironi A. Landi M.T. Zochetti C. , “Cancer Incidence in a Population Accidently Exposed to 2,3,7,8-tetraehlorodibenzo-papa- dioxin,” Epidemiology, 4 (1993): 398–406.

On average, the chance of a random product liability lawsuit resulting in a plaintiff court victory is only 1.5% since most lawsuits (>95%) are settled prior to reaching a jury. Viscusi (1991), op. cit. We assumed 3% in our calculations. Whether big cases may have different odds is hard to judge without knowing how the size of stakes affects the risk-attitudes of both plaintiffs and defendants.

Viscusi (1991), op. cit.

Willging T. E. , Trends in Asbestos Litigation, Federal Judicial Center, 1987.

See Fingerhut M.A. Halperin W.E. Marlow D.A. Piacitelli L.A. Honchar P.A. Sweeney M.H. Greife A.L. Dill P.A. Steenland K. Suruda A.J. , “Cancer Mortality in Workers Exposed to 2,3,7,8-tetrachlorodibenzo-p-dioxin,” New England Journal of Medicine, 324 (1991): 212–218.

See Marshall E. , “Toxicology Goes Molecular,” Science, 259 (1993): 1395.

See Barrett J.C. Lamb P.W. Wiseman R.W. , “Multiple Mechanisms for the Carcinogenic Effects of Asbestos and other Mineral Fibers,” Environmental Health Perspectives, 8 (1989): 81–89.

If the conditions are not fully independent, conditional probabilities should be used. In this case, conditions 1 and 2 are positively correlated whereas conditions 2 and 3 are negatively related (i.e., the longer the latency, the more difficult it becomes to establish causation in theory and in fact). Condition 4 seems independent of the others, whereas condition 5 is positively correlated with 1 and 2. Our probability estimates reflect these dependencies and should be interpreted as conditional probabilities. Also, note that our final joint probability estimate is rather conservative, since it requires the occurrence of all five conditions that gave rise to the asbestos calamity. In fact, just a subset of these five conditions could generate significant liability. A more detailed decision tree analysis is needed to capture the weight to be assigned to the various subsets of branches.

See Lippmann M. , “Formaldehyde and Other Aldehydes,” in Environmental Toxicants, Human Exposures and Their Health Effects (New York, NY: Van Nostrand Reinhold, 1992), pp. 299–322.

At least one case of alleged industry cover-up has been reported regarding dioxin. See Roberts Leslie , “Dioxin Risks Revisited.” Science, 251 (1991): 626. It involved the credibility of industry studies on the health effects of dioxin exposure on workers at a Monsanto pesticide factory in West Virginia. Although the EPA initiated an investigation to determine whether there was any criminal intent in Monsanto's interpretation of these studies, the EPA eventually dropped the investigation.

The asbestos estimate of $47 billion was arrived at as follows. Between 1970 and 1982, about $1 billion had been paid to compensate and litigate claims. Estimates of future costs range from $8 to $87 billion, with a “most likely estimate” of $38 billion. See Kakalik J.S. Ebener P.A. Felstiner W.L.F. Shanley M.G. , Costs of Asbestos Litigation (Santa Monica, CA: Rand, Institute for Civil Justice, 1983). However, by 1992 the actual liability level had already risen to $9 billion. Wall Street Journal, July 15, 1992, p. A3. In view of this, we subjectively increased the future mean estimate to $50 billion as of 1982 (in undiscounted 1952 dollars). Assuming a 30-year time frame, a constant cost profile, a 3% inflation rate since 1982, and a future discount rate of 3%, this yields a combined cost of $47 billion in 1993 dollars.

Readers may wonder why this 90% confidence range is smaller than the range shown in Table 2, which only concerns hazardous waste sites. The reason is that the wider range in Table 2 (from $76 to $1,421 million) is not a 90% range. It was obtained by adding six component ranges; if each of these were viewed as a 90% confidence range, their sum would be closer to a 100% range.

See Schoemaker P.J.H. , “When and How to Use Scenario Planning: A Heuristic Approach with Illustration,” Journal of Forecasting, 10 (1991): 549–564.

Greenberg Maurice R. , chief executive officer of American International Group Inc., one of the largest insurance companies, has argued that Superfund spends too much money on apportioning liability and too little on cleanup. To spread the cost of paying for pollution cleanup, he proposed a broad-based fund that would be paid for by businesses, insurers, and government.

For instance, if two variables are inversely correlated (e.g., employment and inflation in today's economy) certain outcome combinations are implausible (e.g., full employment and zero inflation). We examined the major correlations that constrained the possible combinations. As it turned out, the extreme scenarios (obtained by grouping all positive and negative outcomes into two maximally opposite worlds) contained no contradictory pairs of outcomes.

See Schoemaker P.J.H. , “Scenario Planning: A Tool for Strategic Thinking,” Sloan Management Review, 36/2 (Winter 1995): 25–40.

See Avila J. Whitehead B. , “What is Environmental Strategy?” The McKinsey Quarterly, 4 (1993): 53; see also, Elkington J. , “Towards the Sustainable Corporation,” California Management Review, 36/2 (1994): 90–100.

They involve questions of information sharing, linkages across risks and insurers, the time value of money, a dwindling pool of insurers, and reinsurance issues that are beyond our scope.

Figure 4 was adapted and updated from Balzhiser Richard E. , “Meeting the Near-Term Challenge for Power Plants,” in Technology and Environment, National Academic Press, 1989, pp. 95–11 3. Although we think that legislative environmental action may have peaked in the U.S., regulatory action (to interpret and enforce the laws) may not have peaked, given the customary lag of many years plus likely amendments.

See Ward K.M. Duffield J.W. , Natural Resource Damages, Law and Economics (New York, NY: John Wiley and Sons, Inc., 1992).

Menell P. , “The Limitations of Legal Institutions for Addressing Environmental Risks,” Journal of Economic Perspectives, 5 (1991): 93–113.

Another important part of Superfund established liability for damages or injuries to natural resources resulting from the release of hazardous substances. These statutes, however, are seldom fully enforced.

Abraham K.S. , Environmental Liability Insurance Law (Englewood Cliffs, NJ: Prentice Hall Law and Business, 1991).