Assicurazioni Generali, S.P.A. v. Kenneth Neil, Assicurazioni Generali, S.P.A. v. Kenneth Neil

OPINION DIANA GRIBBON MOTZ, Circuit Judge: An insurance company filed this diversity declaratory judgment action to resolve questions under a general liability policy. The company sought a declaration that it need not provide defense or coverage to its insured in two personal injury actions brought against the insured in Florida. The district court awarded summary judgment and attorneys’ fees to the insured. Because, under Maryland law, the policy’s pollution exclusion bars coverage for the injuries alleged in tbe underlying tort actions, we reverse. I. This controversy arises out of an accident at a Holiday Inn in West Palm Beach, Florida. A Maryland partnership in which Kenneth Neil was a general partner owned and operated the Holiday Inn for sometime; in December 1991, the partnership sold its interest in the hotel. Three months later, in March 1992, several hotel guests suffered carbon monoxide poisoning. The injured guests brought personal injury actions against the new owners and Neil. Neil’s general liability insurance policy, issued by Assicurazioni Generali, S.p.A. (Gen-erali)', an Italian corporation, covered a number of hotels owned by the partnership, including the West Palm Beach Holiday Inn. The stated policy period was August 1, 1991 to August 1, 1992. The policy provides comprehensive general liability insurance for“bodily injury” caused by an “occurrence.” The policy defines “occurrence” as “an accident, including continuous or repeated exposure to conditions, which results in bodily injury.” It defines bodily injury as “bodily injury, sickness or disease sustained by any person which occurs during the policy period.” The policy also provides coverage for injury arising from Completed Operations Hazards, which is defined to include: bodily injury ... arising out of operations ... but only if the bodily injury ... occurs after such operations have been completed or abandoned and occurs away from premises owned by or rented to the Named Insured. In addition, the policy contains an “absolute” pollution exclusion. This exclusion states: [T]he insurance provided in this Section of the Policy DOES NOT APPLY TO: (a) The contamination of any environment by pollutantsthat are introduced at any time, anywhere, in any way; (b) Any bodily injury, personal injury, property damage, costs or other loss or damage arising out of such contamination ...; or (c) Payment for the investigation or defense of any loss, injury or damage ... related to any of the above. It is hereby understood that the following meanings apply to various terms used in the foregoing: a. “CONTAMINATION” means any unclean or unsafe or damaging or injurious or unhealthful condition arising out of the presence of pollutants, whether permanent or transient in any environment. b. “ENVIRONMENT” includes any person, any manmade objects or feature ... land, bodies of water ... air and any other feature of the earth or its atmosphere, whether or not altered, developed or cultivated, including, but not limited to, any of the above, owned, controlled or occupied by the insured. c. “POLLUTANTS” means smoke, vapors, soot, fumes, acids, sounds, alkalies, chemicals, liquids, solids, gases, *1000 thermal pollutants and all other irritants or contaminations. Notwithstanding anything in the foregoing which may bestated in the foregoing to the contrary, it is hereby understood and agreed that this Pollution Exclusion does not apply to “Bodily Injury ... caused by heat, smoke or fumes arising from a hostile fire.” The policy requires locations covered under the policy to be identified and kept on file with the insurer. Consistent with this provision, in January 1992, Neil notified Gen-erali that he no longer owned or managed the West Palm Beach Holiday Inn and sought a refund of a portion of the policy to reflect this fact. Generali subsequently issued a refund and an endorsement providing that coverage for the West Palm “location” was “deleted” effective January 15,1992; the endorsement also stated that “All Other Terms and Conditions Remain Unchanged.” After Neil sought, and Generali denied, defense and coverage in the Florida suits, Generali brought this declaratory judgment action, seeking to establish that its policy provided no coverage to Neil. The district court granted summary judgment to Neil. The court held that (1) the deletion of the West Palm Beach location did not deprive Neil of coverage under the general liability provision, (2) in addition to the general liability coverage, Neil was entitled to coverage under the completed operations hazard provision, and (3) the pollution exclusion did not bar coverage. Because determination of the applicability of the pollution exclusion dictates the holding of this case, we focus on that issue, assuming without deciding that the policy would otherwise provide coverage. We note as a preliminary matter that the parties agree that Maryland law applies in this diversity action involving an insurance contract issued in Maryland. See Continental Cablevision of New England, Inc. v. United Broadcasting Co., 873 F.2d 717, 720 (4th Cir.1989); Traylor v. Grafton, 273 Md. 649, 332 A.2d 651 (Md.1975). II. Neil contends that the Generali policy’s pollution exclusion eliminates coverage only for injuries resulting from environmental pollution and therefore does not bar coverage for injuries arising from carbon monoxide poisoning inside a hotel, as alleged in the Florida cases. The policy language providing for the pollution exclusion is, however, quite expansive. It excludes from coverage “[t]he contamination of any environment by pollutants that are introduced at any time, anywhere, in any way.” (Emphasis added). Moreover, the policy defines “contamination” as “any” injurious condition “arising out of the pi’esence of pollutants, whether permanent or transient in any environment.” (Emphasis added). The definition of “pollutants” includes“smoke, vapors, soot, fumes, acids, sounds, alkalies, chemicals, liquids, solids, gases, thermal pollutants and all other irritants or contaminations.” Thus, carbon monoxide—whether considered a “fume,” “vapor,” or “gas”—plainly falls within this policy definition of “pollutant.” The policy also defines “environment” broadly to include “any person, manmade objects or feature” and “any other feature of the earth or its atmosphere, whether or not altered, developed or cultivated.’The policy goes on to explain that the pollution exclusion does not bar coverage for any bodily injury “caused by heat, smoke or fumes arising from a hostile fire.” This provision clearly applies to accidents that occur within a building and that do not result from what is commonly considered industrial environmental pollution. If, contrary to its broad language, the policy intended “environment” to be confined to geological features, this “exclusion to the exclusion” would be unnecessary. Neil does not suggest that this broad policy language is either ambiguous or generally inapplicable to injuries resulting from carbon monoxide poisoning. His sole argument, as to the policy language, is that the pollution exclusion only applies to property “owned or controlled” by him. He relies on the definition of environment in the policy: “Environment” includes any person, any manmade objects or feature, animals, crops or vegetation, person, any land, bod *1001 ies of water ... air and any other feature of the earth or its atmosphere, whether or not altered, developed or cultivated, including, but not limited to, any of the above, owned, controlled or occupied by the insured. (Emphasis added). At the time of the carbon monoxide poisoning Neil had sold the hotel, and thus he no longer “owned or controlled” it. He contends that the pollution exclusion, therefore, does not apply here. This is an exceedingly odd claim for Neil to assert. He asks us to hold that general liability and completed operations hazards coverage for the hotel extended after he sold it, but the pollution exclusion ended at that very time. His interpretation would provide Neil with more insurance coverage for less money. It would mean that after he sold the hotel and received premium refunds, he obtained coverage that had been barred by a pollution exclusion when he owned the hotel and paid the full premiums. Surely no reasonable insured or insurer would enter into a contract that provided limited coverage for full premium payments and far greater coverage for partial premium payments. Nevertheless, Neil maintains that the policy language quoted above makes the pollution exclusion inapplicable to any “environment” not “owned, controlled, or occupied by the insured.” But the language Neil relies upon certainly does not compel such a nonsensical interpretation. Rather, its more natural reading is to clarify that the pollution exclusion includes, but is not limited to, an environment that is “owned, controlled or occupied by the insured.” This more natural reading not only accords with common sense and normal commercial practice, but also is the only construction consistent with the obvious purpose of the absolute pollution exclusion: to exclude from coverage injuries arising from “contamination” of “any environment” by “pollutants,” which are “introduced at any time, anywhere, in any way.” Perhaps for these reasons, Neil chiefly relies on an alternative argument to which we now turn. III. Neil’s principal contention, which the district court found persuasive, is that regardless of the policy language, a Maryland court would limit the applicability of any pollution exclusion to instances of environmental pollution. As we noted at the outset, Maryland law governs this case. Hence, if a federal court determines that Maryland’s highest court would hold that the pollution exclusion only applies to industrial environmental pollution and thus does not bar coverage in this case, we must so hold. Faithful application of Maryland law would require this result even if, as here, the policy’s plain language and ordinary meaning indicate a different result. Accordingly, we examine Maryland law on this subject. A. Generally, Maryland law does not compel a court to ignore an insurance policy’s plain language and ordinary meaning. Quite the contrary, Maryland’s highest court has repeatedly noted that “[i]n the interpretation of the meaning of an insurance contract,” it “aceord[s]a word its usual, ordinary and accepted meaning unless there is evidence that the parties intended to employ it in a special or technical sense.” Cheney v. Bell Nat’l Life, 315 Md. 761, 556 A.2d 1135, 1138 (Md.1989); see also Chantel Associates v. Mt. Vernon, 338 Md. 131, 656 A.2d 779, 784-85(Md.1995). These general principles would seem to require that a Maryland court defer to the policy’s plain meaning, and so in this case find the pollution exclusion applicable. In fact, the only Maryland appellate case interpreting a pollution exclusion in a similar context has held that the exclusion bars coverage for injuries resulting from carbon monoxide poisoning. See Bernhardt v. Hartford Fire Ins. Co., 102 Md.App. 45, 648 A.2d 1047 (Md.Ct.Spec.App.1994), cer t. granted and volunt. withdrawn, 337 Md. 641, 655 A.2d 400(1995). In Bernhardt, the Maryland Court of Special Appeals detailed the changes made by the insurance industry over time in pollution exclusion clauses. Id. at 1049-50. It recognized that some courts have interpreted this drafting history as demonstrating that the industry may have originally intended to limit the exclusion’s applicability to industrial environmental pollution. Id. at 1050. Nevertheless, after *1002 careful application of the interpretative principles set forth above to the specific language of the insurance policy before it, the intermediate appellate court ruled that the policy’s pollution exclusion clearly barred coverage for injuries resulting from carbon monoxide poisoning that had occurred inside a house. Id. at 1052. The Bernhardt court acknowledged that “the title of the endorsement—‘pollution exclusion’—is, standing alone, ambiguous” but noted that the language of the policy provision was “quite specific.” Id. at 1051. Quoting that language, the court explained that carbon monoxide “was a ‘gaseous ... irritant or contaminant’ and constituted ‘fumes’ and ‘chemicals’ within the clear language of the definition of ‘pollutant.’ ” Id. The court pointed out that “nowhere” in the exclusion before it did “the word ‘industry’ or ‘industrial’ appear” nor did the policy draw any distinction “between intentional and non-intentional discharge of pollutants” or “suggest that only chronic commission ... is excluded from coverage.” Id. at 1051-52. For these reasons, the Bernhardt court held that whatever the insurance industry’s original intent with regard to such clauses, the pollution exclusion in the policy it was construing unambiguously barred coverage for injuries resulting from the carbon monoxide poisoning. Because the Bernhardt analysis is equally applicable here—in fact, the pollution exclusion before us, on its face, even more clearly covers injuries resulting from carbon monoxide poisoning—Bernhardt would seem to require us to hold that the pollution exclusion bars coverage. If the Bernhardt holding were contained in a state statute or a decision of the state’s highest court, the Maryland Court of Appeals, we would, absent extraordinary circumstances, be required to follow it. See Bernhardt v. Polygraphic Co. of Am., 350 U.S. 198, 204-05, 76 S.Ct. 273, 100 L.Ed. 199 (1956). It is axiomatic that in determining state law a federal court must look first and foremost to the law of the state’s highest court, giving appropriate effect to all its implications. A state’s highest court need not have previously decided a case with identical facts for state law to be clear. It is enough that a fair reading of a decision by a state’s highest court directs one to a particular conclusion. Only when this inquiry proves unenlightening, as we find it does in this case, should a federal court seek guidance from an intermediate state court. When seeking such guidance we defer to a decision of the state’s intermediate appellate court to a lesser degree than we do to a decision of the state’s highest court. Nevertheless, we do defer. Indeed, the Supreme Court has specifically directed: [w]here an intermediate appellate state court rests its considered judgment upon the rule of law which it announces, that is a datum for as certaining state law which is not to be disregarded by a federal court unless it is convinced by other persuasive data that the highest court of the state would decide otherwise. West v. A T & T, 311 U.S. 223, 237, 61 S.Ct. 179, 85 L.Ed. 139 (1940); accord, Hicks v. Feiock, 485 U.S. 624, 630 n. 3, 108 S.Ct. 1423, 99 L.Ed.2d 721 (1988); see also Stoner v. New York Life Ins. Co., 311 U.S. 464, 467, 61 S.Ct. 336, 85 L.Ed. 284 (1940) (“[Federal courts, under the doctrine of Erie ... must follow the decisions of intermediate state courts in the absence of convincing evidence that the highest court of the state would decide differently.”). Thus, a federal court must “present” persuasive data when it chooses to ignore a decision of a state intermediate appellate court that is directly on point. United States v. Little, 52 F.3d 495, 498 (4th Cir.1995). What a federal court, sitting in diversity, cannot do is simply substitute its judgment for that of the state court. Id. B. The experienced district judge undoubtedly understood the above principles. However, we believe he miscalculated the persuasiveness of the data upon which he relied in concluding that the Maryland Court of Appeals would decline to follow Bernhardt. This may be because we have never articulated what constitutes “persuasive data,” permitting a federal court to disregard an intermediate appellate court’s holding on a question of state law. We take this opportunity to set forth some guidelines. *1003 Generally, only if the decision of a state’s intermediate court cannot be reconciled with state statutes, or decisions of the state’s highest court, or both, may a federal court sitting in diversity refuse to follow it. When an intermediate state court decision is at odds with an existing statutory scheme or one amended after issuance of the intermediate court’s decision, a federal court may justifiably surmise that the statute presents persuasive data that the state’s highest court would not follow the intermediate court’s decision. See, e.g., J & R Ice Cream Corp. v. California Smoothie Licensing Corp., 31 F.3d 1259, 1271-73(3d Cir.1994); Ground Air Transfer, Inc. v. Westates Airlines, 899 F.2d 1269, 1275 (1st Cir.1990) (alternative holding). Similarly, the holdings of a state’s highest court that undermine the rationale of an intermediate appellate court decision may constitute such persuasive data, see Dimidowich v. Bell & Howell, 803 F.2d 1473, 1482-83 (9th Cir.1986); this is so even if a state supreme court holding antedates the inferior court’s opinion. See Humphrey v. C.G. Jung Educ. Ctr., 624 F.2d 637, 643 (5th Cir.1980) (alternative holding). However, a federal court must closely scrutinize the assertedly conflicting statutory schemes or supreme court decisions to satisfy itself that they truly do undermine an intermediate appellate court decision directly on point. See, e.g., Gates Rubber Co. v. USM Corp., 508 F.2d 603, 606-15 (7th Cir.1975). A federal court can depart from an intermediate court’s fully reasoned holding as to state law only if “convinced” that the state’s highest court would not follow that holding. West, 311 U.S. at 237, 61 S.Ct. 179; cf. United States v. Ramos, 39 F.3d 219, 221-22 (9th Cir.1994) (an intermediate court’s totally unexplained holding may merit less deference). Accordingly, a federal court cannot refuse to follow an intermediate appellate court’s decision simply because it believes the intermediate court’s decision was wrong, bad policy, or contrary to the majority rule in other jurisdictions. West, 311 U.S. at 237 (federal courts have a duty to apply state law “rather than to prescribe a different rule, however superior [the different rule] may appear from the viewpoint of ‘general law and however much the state rule may have departed from prior decision of the federal courts”); Malone v. Bankhead Enters., 125 F.3d 535, 540 (7th Cir.1997) (“The mere fact that more courts have interpreted other states’ savings statutes differently does not persuade us that the Illinois appellate courts improperly interpreted Illinois’ statute.”). With these precepts in mind, we examine the data that the district court relied on in refusing to follow Bernhardt. C. The district court based its ruling on two factors. First, the court noted that the Maryland Court of Appeals had granted certio-rari in Bernhardt (which was withdrawn upon settlement), “indicating] that at least a number of the judges of that Court though[t] the matter ought to be taken up.” Second, the district court determined that the Maryland Court of Appeals’ decision in Sullins v. Allstate Ins. Co., 667 A.2d 617 (Md.1995), led to the inescapable conclusion that the high court “would reverse Bernhardt if that decision were to reach it for review.” As to the first factor, although a grant of certiorari surely does indicate that some of the judges on the high court wish to hear a case, it does not, in itself, constitute a decision on the merits. Indeed, more frequently than not, a grant of certiorari in Maryland does not even result in reversal by the high court. During the 1997 fiscal year the Maryland Court of Appeals reversed, dismissed, or remanded 36 of the civil cases it heard. Administrative Office of the Courts, Annual Report of the Maryland Judiciary 1996-1997 Table CA-7 (1997). By comparison, the court affirmed 40 cases. Id. Given these statistics and the principles set forth above governing federal courts in diversity cases, the Maryland Court of Appeals’ decision to grant certiorari in a case cannot constitute persuasive data that the high court would choose not to follow the decision of the intermediate appellate court in that case. Cf. Haugen v. Total Petroleum, Inc., 971 F.2d 124, 126 n.2 (8th Cir.1992) (Minnesota Supreme Court’s denial of review of Minnesota Court of Appeals decision does not “give the court of appeals decision any more or less precedential weight”); Giguere v. United States Steel Corp., 262 F.2d 189, 192 (7th *1004 Cir.1959) (Illinois Supreme Court’s grant of appeal to Illinois Appellate Court ease does not diminish precedential weight of the intermediate court’s holding). Thus, the district court’s reliance on the certiorari grant was misplaced. Sullins, upon which the district court principally relied, however, presents a more complicated question. In Sullins, the Maryland Court of Appeals held, a year after Bernhardt, that a pollution exclusion clause did not bar coverage for injuries resulting from exposure to lead paint chips. 340 Md. 503, 667 A.2d 617. The district court concluded that Sullins “essentially held that the standard pollution exclusion applies to what, in common sense, it ought to apply, viz., to pollution of the physical environment that is the contamination of the air and water that surround us all, not contamination of indoor living space.” To the extent the district court relied on “common sense” or what “ought to” be the law, it erred. These do not constitute persuasive data permitting a federal court to ignore the holding of an intermediate state court. See West, 311 U.S. at 237, 61 S.Ct. 179; Malone, 125 F.3d at 540. For this same reason, Neil’s citation to assertedly well reasoned cases construing the law of other states to find pollution exclusions inapplicable to carbon monoxide poisoning or other indoor contaminations, does not assist him. Such cases do exist. See, e.g., Western Alliance Ins. Co. v. Gill, 426 Mass. 115, 686 N.E.2d 997 (Mass.1997); Stoney Run Co. v. Prudential-LMI Ins. Co., 47 F.3d 34 (2d Cir.1995) (New York law); Regional Bank of Colorado v. St. Paul Fire & Marine Ins. Co., 35 F.3d 494 (10th Cir.1994) (Colorado law); Thompson v. Temple, 580 So.2d 1133 (La.App.1991). They may even state the “better view.” But see State Farm Fire & Casualty v. Deni Assoc., 678 So.2d 397 (Fla.App.1996); League of Minn. Cities Ins. Trust v. City of Coon Rapids, 446 N.W.2d 419 (Minn.Ct.App.1989); see also New Castle County v. Hartford Accident and Indem. Co., 933 F.2d 1162, 1195 nn. 60, 61 (3d Cir.1991) (listing cases and concluding “about half of the cases hold[ ] that the clause bars coverage, and ... the other half hold[ ] that it does not”). But they do not provide a basis for ignoring the Court of Special Appeals’ statement as to Maryland law. Although neither “common sense” nor the law of other states assists Neil, if Sullins does hold that pollution exclusions can only apply to environmental pollution, then the district court was correct. In that case, Sul-lins would provide persuasive data that the Court of Appeals would, if given the opportunity, overrule Bernhardt. The question is: does Sullins hold this? The “bottom line” of Bernhardt and Sul-lins obviously differs—Bernhardt held the pollution exclusion applicable, resulting in a denial of coverage for the insured, while Sul-lins held the exclusion inapplicable, resulting in a finding of coverage for the insured—but just as clearly, Sullins did not expressly overrule the Bernhardt holding. Indeed, Sullins principally relied on Bernhardt in recounting the history of the pollution exclusion, citing Bernhardt no less than four times. See Sullins, 667 A.2d at 622. This is noteworthy because the Maryland Court of Appeals does not indiscriminately or frequently cite opinions from the Court of Special Appeals; for example, although Sullins relied on multiple cases both in Maryland and throughout the country, it referred to only two other cases from the Court of Special Appeals, and cited each a single time. It seems unlikely that Sullins would have so heavily relied on Bernhardt, without any qualification, if it disapproved of Bernhardt. However,because Sullins depended on Bernhardt only for historical purposes, this alone does not indicate approval of the holding in Bernhardt. To determine the vitality of Bernhardt, we must examine Sullins ’ rationale. Like Bernhardt, Sullins relied heavily on general principles of Maryland law governing interpretation of insurance policies, and,again like Bernhardt, Sullins carefully examined the language of the insurance policy before it in applying these principles. In Sullins, the pollution exclusion clause provided in its entirety: We do not cover bodily injury or property damage which results in any manner from the discharge, dispersal, release, or escape of: *1005 (a) vapors, fumes, acids, toxic chemicals, toxic liquids or toxie gasses; (b) waste materials or other irritants, contaminants or pollutants. 667 A.2d at 618 (quoting policy language). Sullins first recognized that when interpreting insurance policies, words are to be “given their customary, ordinary and accepted meaning, unless there is an indication that the parties intended to use the words in a technical sense” and that “[a] term which is clear in one context may be ambiguous in another.” 667 A.2d at 619. Applying these principles, the Sullins court found the terms “contaminants” and “pollutants” ambiguous in the context of the case before it, i.e., in determining whether lead paint chips constitute a contaminant or pollutant as defined by the policy. Id. at 620. For this reason, the Court of Appeals in Sullins determined that “the policy must be construed against [the insurance company] as the drafter of the policy” and thus the pollution exclusion would not bar coverage. Id. This analysis, of course, does not compel the conclusion that the Court of Appeals would not follow Bernhardt. The intermediate court there found that carbon monoxide was clearly a “ ‘gaseous ... irritantor contaminant’ and constituted ‘fumes’ and ‘chemicals’ within the clear language of the definition of ‘pollutant’.” Bernhardt, 648 A.2d at 1051. Thus, the exclusion in Bernhardt did not present any ambiguity to be construed against the insurer. Accordingly, if the Sul-lins opinion stopped after its analysis of the policy language, notwithstanding its ultimate holding, nothing in Sullins would suggest disapproval of or disinclination to follow the intermediate court’s decision in Bernhardt in a case, like the present one, with unambiguous policy language. But Sullins does not stop with analysis of the policy language. Rather, the Sullins court went on to state that the “history of the pollution exclusion supports this conclusion.” Id. at 622. The court then recounted that history, which it found demonstrated an intention by the insurance industry “to exclude only environmental pollution damage from coverage.” Id. Undoubtedly, it is these observations that led the district court to hold that Maryland’s high court would “reverse the decision in Bernhardt.” On close examination, however, the Sul-lins ’ statements and review of the insurance pollution’s drafting history provide no persuasive data that Maryland’s high court would limit the applicability of all pollution exclusions to injuries resulting from environmental pollution. The Sullins court, although initially somewhat unclear on the point, ultimately explained the precise “conclusion” that it found supported by the insurance industry’s intent to confine the pollution exclusion to environmental pollution. That “conclusion” is not a general one, i.e., pollution exclusions apply only to injuries resulting from environmental pollution and not to contamination of indoor living space. Rather, Sullins found the intent of the insurance industry, as demonstrated by the history of the pollution exclusion, to support a very specific “conclusion,” ie. that the language of the pollution exclusion in the case before it was ambiguous: It appears from the foregoing discussion [of the drafting history of the pollution exclusion] that the insurance industry intended the pollution exclusion to apply only to environmental pollution. That supports our conclusion that a reasonably prudent layperson may interpret the terms“pollution” and “contamination” in the circumstances of the case now before us, as not encompassing lead paint.... Id. at 623 (emphasis added). This “conclusion” does not undermine Bernhardt’s holding that when a court examines language in a pollution exclusion that unambiguously applies to a cause of alleged injuries, the exclusion bars coverage of those injuries, regardless of what the drafting history may suggest about the original intent of such exclusions. Although Sullins does reason, as Neil contends, that the insurance industry originally intended pollution exclusions to be limited to environmental pollution, the Sullins court’s holding rests upon the ambiguity in the exclusion language before it and its resolution of that ambiguity against the insurer. In contrast, the language of the pollution exclusion in Bernhardt and in the case at hand unambiguously applies to injuries resulting from carbon monoxide poisoning. *1006 We note that Bernhardt itself questioned whether the broad, but clear, pollution exclusion before it was “beyond that which might be required to meet the industry’s legitimate aims,” but decided that it was “not the function of courts to rewrite the contracts of parties, even when the enforcement of the contract as written may result in a hardship.” Bernhardt, 648 A.2d at 1052. This conclusion is dictated by along series of Maryland Court of Appeals’ cases. See, e.g., State v. Attman/Glazer, 323 Md. 592, 594 A.2d 138, 146 (Md.1991); Canaras v. Lift Truck Services, Inc., 272 Md. 337, 322 A.2d 866, 873 (Md.1974); Compania de Astral, S.A. v. Boston Metals Co., 205 Md. 237, 107 A.2d 357, 372 (Md.1954). Given this principle and that Sullins expressly follows the same contract interpretation analysis as Bernhardt, it seems unlikely that the Maryland Court of Appeals would rewrite an unambiguous pollution exclusion either. For these reasons, we believe Sullins does not provide persuasive data that the Maryland Court of Appeals would refuse to follow Bernhardt. Since Neil offers no other persuasive data that Bernhardt does not accurately state Maryland law, we must follow it and hold that the pollution exclusion bars coverage for the injuries allegedly caused by carbon monoxide poisoning. IV. We reverse the district court’s grant of summary judgment to Neil, vacate its awards of attorneys’ fees to Neil for services rendered this action and as in the underlying Florida tort suits, and remand the case so that the district court can enter an order granting summary judgment to Generali. REVERSED AND REMANDED