POSNER, Circuit Judge.
This appeal in a diversity suit governed by Illinois law requires us to interpret the pollution exclusion from coverage found in most general liability insurance policies. The most common policy is the "commercial general liability policy" drafted by the Insurance Services Office and purchased by businesses to insure against losses arising out of general business operations. 9A Steven Plitt et al., Couch on Insurance § 129:1, pp. 129-5 to 129-7 (3d ed.2005). The policies at issue in this case are "public entity general liability policies," which are issued to municipalities to cover analogous risks and contain the same pollution exclusion as the commercial general liability policy.
Crestwood is a small Chicago suburb (population 11,000) that supplies its residents with water obtained from both Lake Michigan and wells that it owns, and bills the residents for the water. According to the tort complaints, in 1985 or 1986 Crestwood's mayor and other Village officials learned from state environmental authorities that one of the wells was contaminated by perc (PCE—perchloroethylene, also known as tetrachloroethylene). A solvent widely used in dry cleaning, perc is a common contaminant of soil and groundwater and is more difficult to clean up than oil spills are; it is a carcinogen to boot. Agency for Toxic Substances & Disease Registry, U.S. Department of Health & Human Services, "Toxicological Profile for Tetrachloroethylene" 55-59 (Sept.1997), www.atsdr.cdc.gov/toxprofiles/tp.18.pdf (visited Jan. 31, 2012). Perc used by a nearby dry-cleaning establishment had leaked into the groundwater tapped by the well. Village officials promised the state authorities that the well would be used only in emergencies. But instead, for reasons of economy, the well continued to be used as a source of the daily Village water supply—without disclosure to the Village's residents. The well remained in use until 2007, and not until 2009 was it sealed. Illinois Department of Public Health, Division of Environmental Health, "Water Well Sealing Form," www.villageof crestwood.com/documents/ CRESTWOOD_-_WATER_WELL_ SEALING_FORM_04-22-2009_15-50-37-1.pdf (visited Jan. 31, 2012).
Hundreds of Crestwood residents, having learned of the contamination of their water supply from a series of articles in the Chicago Tribune, sued the Village and past and present Village officials in an Illinois state court seeking damages for injury to health. In a parallel suit the State of Illinois seeks an injunction requiring the Village to finance "a site inspection to determine the nature and extent of contamination" and take "all necessary steps to remediate the contamination." All these suits are pending.
The defendants' insurance policies (primary policies issued by Scottsdale and excess policies issued by National) exclude from coverage "`bodily injury,' `property damage,' or `personal injury' arising out of, or `wrongful act(s)' which result in the actual, alleged or threatened discharge, dispersal, seepage, migration, release or escape of `pollutants' at any time," and also exclude from coverage expenses arising from orders for "cleaning up . . . or in any way responding to, or assessing the effects of pollutants." "Pollutants" are defined as "any solid, liquid, gaseous or thermal irritant or contaminant, including smoke, vapor, soot, fumes, acids, alkalis, chemicals and waste." There are slight differences in the wording of the various policies but they are immaterial and we ignore them.
There is no doubt that perc is a "contaminant" within the meaning of the policies; and the tort plaintiffs are complaining about its "dispersal" by the Village from the contaminated well to their homes via the system of water mains that connects the well to the homes. The problem with stopping there and affirming the district
Suppose a tanker truck filled with perc crashes into a bridge abutment, spilling its liquid cargo, and another vehicle skids on the wet surface of the highway into the abutment, injuring the driver. Perc is both a contaminant and a cause of the bodily injuries in this example. But it would be absurd to argue—and the insurers do not argue—that a claim arising from such an accident would be within the pollution exclusion, since in no reasonable sense of the word "pollution" was the driver a victim of pollution. Our Pipefitters opinion, citing cases that hold the exclusion inapplicable to "injuries arising from an individual's ingestion of malathion during a municipal pesticide-spraying operation," "paint damage to vehicles which occurred during the spraypainting of a bridge," an "apartment-dweller's ingestion of lead paint," and a "release of asbestos particles during installation, handling and removal of insulation," noted that all these cases "involve injuries resulting from everyday activities gone slightly, but not surprisingly, awry. There is nothing that unusual about paint peeling off of a wall, asbestos particles escaping during the installation or removal of insulation, or paint drifting off the mark during a spraypainting job. A reasonable policyholder, these courts apparently believed, would not characterize such routine incidents as pollution." 976 F.2d at 1043-44.
Generalizing, the Supreme Court of Illinois has interpreted the pollution exclusion to be limited to harms arising from "traditional environmental pollution." American States Ins. Co. v. Koloms, 177 Ill.2d 473, 227 Ill.Dec. 149, 687 N.E.2d 72, 82 (1997), involved a furnace that leaked carbon monoxide, injuring several workers in the building that contained the furnace. As in the cases discussed in Pipefitters, there was no environmental damage and the court held the pollution exclusion inapplicable.
A more perspicuous formula than "traditional environmental pollution" would be "pollution harms as ordinarily understood." That formula would also exclude the case of the leaking furnace; for think of what a misuse of language it would be to say that the workers had been injured by pollution. If one commits suicide by breathing in exhaust fumes, is that death by pollution?
We can make further progress by thinking about reasons for exclusions from insurance coverage. The reasons do not include fear by insurance executives of losses by insureds. The business of insurance is covering losses. The more policies written, the better from the insurance company's standpoint—but this is provided the company can estimate within a reasonable range the size of the losses that it is likely to be required to reimburse the policyholders for. Otherwise it can't set premiums that will be high enough to compensate it for the risk of having to reimburse the losses it's insuring, without being so high that no one will buy its polices.
Insurance companies use statistical methodologies (actuarial science) to calculate "expected losses"—the sum of the insured losses, if they occur, discounted (multiplied) by the probability of loss. The higher that probability or the greater the
The effect of insurance on an insured's behavior and hence on the risk is called "moral hazard." A related problem, also illustrated by fire insurance, goes by the name of "adverse selection." ("Adverse self-selection" would be clearer.) Overinsurance would attract people who valued their property at less than its insured value, and the addition of such people to the insurance pool would increase the probability of losses and so drive up premiums. Legitimate insurance purchasers would respond to the increased premiums by shifting their business to companies that refused to overinsure, and this would raise the probability of losses to the (shrunken) insurance pool of the first company and drive its premiums even higher. It would be a death spiral. The fear of such spirals explains the exclusion of pre-existing medical conditions from health insurance policies, the requirement in the federal health-reform law that in exchange for the elimination of that exclusion everyone be required to have health insurance, and laws requiring all motorists to have liability insurance as otherwise premiums would be driven to high levels by the most careless drivers, causing the safer ones to drop insurance, although that is not the only reason for such laws.
Insurers of pollution liability would encounter difficulty in estimating the expected loss from pollution and calculating the premium accordingly. Environmental damage is often very difficult to detect until it has become extensive, let alone to predict, or estimate its likely extent, in advance; and the financial consequences can be horrific but again are unpredictable. Most of those consequences are felt at the clean-up stage, and the insureds in this case would like us to confine the pollution exclusion to those costs, noting (as remarked in the Koloms opinion, 227 Ill.Dec. 149, 687 N.E.2d at 81) that the creation of expansive clean-up liability under federal law triggered the modern compendious pollution exclusion at issue in this case.
The exclusion was first introduced into general commercial liability policies in 1970. But it was the passage in 1980 of CERCLA (Comprehensive Environmental Response, Compensation, and Liability Act, 42 U.S.C. §§ 9601 et seq.), the federal toxic-waste statute, and the threat and later the reality of government-ordered cleanup costs imposed by CERCLA, that prompted the industry to adopt the current, broader exclusion. MacKinnon v. Truck Ins. Exchange, 31 Cal.4th 635, 3 Cal.Rptr.3d 228, 73 P.3d 1205, 1210-11 (2003); Kent Farms, Inc. v. Zurich Ins. Co., 140 Wn.2d 396, 998 P.2d 292, 295 (2000); Doerr v. Mobil Oil Corp., 774 So.2d 119, 126 (La.2000). When CERCLA was first enacted, insurers couldn't calculate the costs it would impose on insureds, and while they could have dealt with this uncertainty by amending their liability policies to authorize retroactive calculation of premiums for pollution insurance, cf. Employers Ins. of Wausau v. Titan Int'l, Inc., 400 F.3d 486, 487-88 (7th Cir.2005), instead they limited their risk by excluding coverage of pollution harms in the broadest possible terms. To make assurance
Because as the opinion in Koloms notes the "predominate motivation" for excluding pollution coverage was "avoidance of the enormous expense . . . of environmental litigation," 227 Ill.Dec. 149, 687 N.E.2d at 81 (emphasis in original), which is a broader class of litigation than just CERCLA actions, the pollution exclusion cannot be limited to clean-up costs. Whittier Properties, Inc. v. Alaska National Ins. Co., 185 P.3d 84, 94 (Alaska 2008); Quadrant Corp. v. American States Ins. Co., 154 Wn.2d 165, 110 P.3d 733, 741 (2005); cf. Jeffrey W. Stempel, "Reason and Pollution: Correctly Construing the `Absolute' Exclusion in Context and in Accord with Its Purpose and Party Expectations," 34 Tort & Ins. L.J. 1, 5 (1998). And it is not. For remember that the policies at issue in this case contain separate exclusions for bodily-injury costs and clean-up costs.
The main reason for the broad pollution exclusion is the adverse-selection problem of which we gave examples earlier. It is true that there is adverse selection only where the adverse factor, such as a pre-existing medical condition or the very low value placed by a property owner on his property, is invisible to the insurer, who therefore can't adjust the insurance premium to the greater risk of loss from insuring those people—can't in other words separate its high-risk customers from its low-risk ones and charge different premiums to the different groups. But invisibility is a problem with pollution insurance too, as this case illustrates dramatically: deliberate concealment by the insureds of the pollution is alleged. If insurers can't determine how likely a would-be buyer of insurance is to pollute, coverage would force enterprises that have a slight risk of liability for causing pollution damage to subsidize the premiums of high-risk potential polluters.
Insurers could have excluded coverage just for knowing or deliberate polluting, which would have done the trick in this case but would not be a complete solution to the adverse-selection problem. A shopper for pollution insurance who knows that he has a high risk of accidentally polluting and being sued for it would, if able to buy the insurance at the normal premium, contribute to the premium spiral that we've described. Forcing him to self-identify as a potential polluter by buying a pollution-coverage rider to his general liability policy (as otherwise he will fall within the pollution exclusion) separates high- and low-risk polluters.
The concerns that animate the pollution exclusion were absent from the cases we discussed earlier, which were typical tort cases in the sense of involving a sudden occurrence the risks of which are well known and that injures one or a few persons. That the occurrence happens to be precipitated by a contaminant is incidental; its presence makes the risk or amount of loss no more uncertain than if, in our hypothetical case, the tanker truck had spilled milk rather than perc. Koloms along with many other cases makes clear that the type of injury illustrated by that case and the cases cited in the Pipefitters opinion fall outside the pollution exclusion. See, e.g., American States Ins. Co. v. Koloms, supra, 227 Ill.Dec. 149, 687 N.E.2d at 77-79; Pipefitters Welfare Educational Fund v. Westchester Fire Ins. Co., supra,
The defendants point out that they didn't originate the contamination. That is irrelevant. Housing Authority Risk Retention Group, Inc. v. Chicago Housing Authority, 378 F.3d 596, 604-06 (7th Cir. 2004) (Illinois law); Kim v. State Farm Fire & Casualty Co., 312 Ill.App.3d 770, 245 Ill.Dec. 448, 728 N.E.2d 530, 535-36 (2000); Town of Harrison v. National Union Fire Ins. Co., 89 N.Y.2d 308, 653 N.Y.S.2d 75, 675 N.E.2d 829, 832 (1996); but cf. Doerr v. Mobil Oil Corp., supra, 774 So.2d at 135-36 and nn. 17-18. The exclusion is of liability for harms resulting from the "dispersal," "migration," or "release" of contaminants, not their creation or just their first distribution. Initially the contamination by perc was confined to the groundwater drawn by the well. But by distributing the water to the residents of Crestwood the Village caused the perc to migrate throughout the Village and inflict (or so it is alleged) widespread personal injuries, along with contamination of soil or structures that is likely to be costly to eliminate. The insureds might as well be arguing that because the Village has never manufactured perc it is responsible for none of the harms that dispersing perc might cause. That would be like a murderer arguing that his victim was killed not by him but by his gun.
The Village "caused" the contamination of its water supply (it could have sealed the well a quarter of a century ago, when it discovered the well was contaminated) in a perfectly good sense of the word, though as in Nature Conservancy v. Wilder Corp., 656 F.3d 646 (7th Cir.2011) (see also Wilder Corp. v. Thompson Drainage & Levee District, 658 F.3d 802, 804 (7th Cir.2011)), the defendants did not introduce the contaminant into the soil or groundwater. The contamination had an infinity of authors, not only the Village and its officials but also the inventor of perc, the founder of Crestwood, and maybe Jean Baptiste Point du Sable, who built a farm at the mouth of the Chicago River in the 1780s and is thought to be the first permanent non-native settler of the Chicago area. None of the other authors could be thought to have caused the contamination and resulting injury in a sense of "cause" that is relevant to legal liability.
The pollution exclusion would mean little if the insured were required to have been the original author of the pollution in order to be within the exclusion. Suppose leaks in the Village's water mains caused by negligent maintenance had allowed the perc from the dry cleaner's establishment to contaminate soil throughout the Village, causing property values to plunge and necessitating heavy clean-up expenses. So long as the Village had notice (actual or constructive) of the groundwater contamination at its well, it would be liable in tort for having caused the pollution. Restatement (Second) of Torts §§ 165, 822(b) (1979); In re Resource Technology Corp., 662 F.3d 472, 475 (7th Cir.2011); Redevelopment
Groundwater contamination, with resulting contamination of drinking water, is extremely common and a fertile source of environmental litigation. See, e.g., Arrow Gear Co. v. Downers Grove Sanitary District, 629 F.3d 633, 635 (7th Cir.2010); Chico Service Station, Inc. v. Sol Puerto Rico Ltd., 633 F.3d 20, 23 (1st Cir.2011); United States v. Aerojet General Corp., 606 F.3d 1142, 1146 (9th Cir.2010); U.S. Bank National Ass'n v. EPA, 563 F.3d 199, 203-04 (6th Cir.2009); Smith v. Carbide & Chemicals Corp., 507 F.3d 372, 375-76 (6th Cir.2007); Olin Corp. v. Certain Underwriters at Lloyd's London, 468 F.3d 120, 123-24 (2d Cir.2006). But the Village and the other defendants in this case argue that when an insured's "core business activity" consists of the manufacture or distribution of the contaminated product, the pollution exclusion does not apply, as otherwise the liability insurance policy would not protect the insured against foreseeable risks run by such a supplier. West American Ins. Co. v. Tufco Flooring East, Inc., 104 N.C. App. 312, 409 S.E.2d 692, 697-98 (1991). This amounts to saying that there would be no adverse-selection problem because the risk of pollution liability would be obvious to the insurer, allowing the separation of high-risk and low-risk insureds that we noted earlier. But it would not be obvious, and the separation would not be feasible, just because the insured was known to sell a product (or use a production process) that could be contaminated and thus spread contamination; the risk would vary across producers. We're not surprised that the only pertinent Illinois case rejects the "core business activity" exception. Kim v. State Farm Fire & Casualty Co., supra, 245 Ill.Dec. 448, 728 N.E.2d at 535.
The insureds argue finally that this is not a pollution case at all, because the amount of perc in the Village's water supply was below the maximum level permitted by environmental regulations. But either the perc caused injuries, maybe because the relevant regulations are too lax, or it did not and the tort suits will fail. All that counts is that the suits are premised on a claim that the perc caused injuries for which the plaintiffs are seeking damages, and that claim triggers the pollution exclusion. Although an insurer's duty to defend his insured depends on what the complaint alleges rather than on the facts that emerge over the course of the litigation and that might or might not give rise to a duty to indemnify, Pekin Ins. Co. v. Wilson, 237 Ill.2d 446, 341 Ill.Dec. 497, 930 N.E.2d 1011, 1016-17 (2010); American States Ins. Co. v. Koloms, supra, 227 Ill.Dec. 149, 687 N.E.2d at 75; National Casualty Co. v. McFatridge, 604 F.3d 335, 338 (7th Cir.2010) (Illinois law), it is doubtful that the plaintiffs could have drafted complaints that did not reveal that this was a pollution case. Would an Illinois court accept as adequately stating a claim the bare assertion that the Village of Crestwood and its officials had by means unspecified caused cancer and other illnesses in hundreds of Crestwood's residents? The insurers conceded at oral argument that the duty to defend would be activated if so enigmatic a complaint were allowed. The complaints actually filed, however, describe in copious detail the conduct giving rise to the tort suits, and in doing so inadvertently but unmistakably acknowledge the applicability of the pollution exclusion.
AFFIRMED.