WOLLMAN, Circuit Judge.
This case is once again before us.
See BJC Health Sys. v. Columbia Cas. Co.,
348 F.3d 685 (8th Cir.2003) (reversing dismissal for failure to state a claim). The plaintiffs, BJC Health System (BJC) and ATG Assurance Company Limited (ATG),
ap
peal from the district court’s grant of a judgment as a matter of law, a motion for partial summary judgment, and a motion to strike ATG’s prayer for punitive damages. They also contend that the district court erred in excluding certain evidence. We affirm in part, reverse in part, and remand for further proceedings.
I.
BJC, a network of hospitals, is the sole shareholder of ATG, a captive insurance company that provides insurance for BJC. In September 1998, BJC and defendant-appellee Columbia Casualty Company, doing business as CNA Health Pro (Columbia), entered into a three-year contract, pursuant to which Columbia agreed to provide reinsurance to ATG for a fixed yearly premium.
Under the contract, Columbia agreed to reinsure ATG for individual claims that exceeded $1 million or $8 million, depending on the hospital, and aggregate claims over $22 million. One clause (“the incurred loss condition”) provided that continued coverage would be conditioned upon an incurred loss ratio of less than 75%. Although the parties contest how incurred loss should be defined, they apparently agree that incurred loss refers to the costs associated with claims made against the insured party. The incurred loss ratio, in turn, is determined by dividing BJC’s incurred losses by the premium paid by BJC. The incurred loss condition is triggered if an incurred loss ratio exceeding 75% occurs with respect to either aggregate claims or with regard to any individual claim.
Columbia provided reinsurance for the first two years of the program. On September 22, 2000, only days before the end of the current policy year, Columbia informed BJC that it had determined that BJC had exceeded the aggregate incurred loss ratio. Columbia also stated that the incurred loss ratio had been exceeded on an individual claim, the “Baby C” claim, as well. BJC had set a claim reserve
of below $3 million on the Baby C claim, but Columbia assessed a reserve of $5 million, a sum which would lead to an incurred loss ratio well in excess of the 75% set forth in the contract. Columbia informed BJC that it would not continue with the third year of the program unless the premium was increased to $3.1 million and the aggregate retention level was increased to $31.1 million.
Columbia extended the policy so that BJC could obtain replacement insurance, which BJC acquired from Zurich. There were differences, however, between the policy offered by Zurich and the policy that BJC had with Columbia. As noted earlier, Columbia’s coverage commenced at an aggregate loss of $22 million. Zurich, however, only offered coverage starting at an aggregate loss of $25 million. This $3 million gap in coverage means that BJC faces an additional $3 million in potential liability that it would not be facing had Columbia continued with the third year of the program.
Although Zurich declined to provide reinsurance starting at $22 million, it stated that if it did offer coverage commencing at that level, the premium for the additional $3 million in coverage would be $1.2 million.
BJC filed suit, alleging breach of contract, and sought compensatory damages for the cost of the Zurich replacement policy as well as compensation for the additional risk exposure BJC had to assume with the Zurich replacement policy. BJC claimed that this latter amount was $1.2 million' — 'the price Zurich would have charged for meeting the $3 million coverage gap. BJC also sought punitive damages, which were based on Columbia’s allegedly fraudulent conduct.
Much of this case revolves around the actuarial work Columbia presented to BJC to justify Columbia’s determination that the incurred loss ratio had exceeded 75%. On October 10, 2000, the parties met in Chicago to discuss their differences. At this meeting, Columbia distributed an actuarial analysis that concluded that the incurred loss ratio had been exceeded. That report was time-stamped at 10:23 that morning. This analysis contained errors that were not present in an earlier version of the report that had been printed at 8:33 that morning, but which had not been distributed. The errors in the distributed version contributed to a higher incurred loss ratio. On October 19, in response to unrelated concerns voiced by BJC,
Columbia distributed a third version of the report that addressed BJC’s concerns but also corrected the incorrect figures from the second version. This third version, however, contained a new error: one of the figures was about $2.4 million too high, an error which, as described below, may have resulted in the calculation of an incurred loss ratio in excess of 75%.
John Pierce, a witness for BJC, reviewed all three versions of the actuarial report and testified that Columbia did not provide any explanation regarding the manner in which Columbia converted accident year data to report year data. An accident year to report year conversion was necessary because, while BJC’s loss data was compiled on an accident year basis, the reinsurance policy was priced on a report year basis. In his report, Pierce remarked that this conversion was “the critical final step” in the analysis and that it was “unreasonable” to make this conversion without an explanation as to how it was performed. Pierce also noted that without knowing how Columbia had made the transition from accident year to report year figures, it was difficult to estimate the impact that the new error in the third version of the actuarial report might have had. He did, however, attempt to calculate the effect and estimated that if this error was corrected the incurred loss ratio would fall below 75%. Pierce also testified that Columbia unreasonably selected the same average experience ratio — an important figure — m all three versions of the actuarial analysis, despite the fact that the underlying data in these versions was different. He stated that the average experience ratio had a significant effect on the accident year figures, but could not state with exactitude how it might affect the report year data because Columbia did not explain the accident year to report year conversion.
Columbia’s actuaries either did not know or could not remember how the accident year to report year conversion was done, and Columbia was unable to produce the computer models it used to conduct its 1999 and 2000 actuarial analyses of the BJC account. There was also evidence that Columbia’s actuaries recommended a premium of nearly $13.5 million for the third policy year, a substantial increase that could only be justified by a dramatic change in BJC’s losses or a by significant change in the assumptions Columbia used in assessing the BJC account. Columbia’s actuaries did not know whether BJC had
experienced a serious change in losses between 1998 and 2000 and did not know whether there was a significant change in Columbia’s methodologies. BJC presented testimony that there were no significant changes in BJC losses.
BJC also presented testimony pertaining to the Baby C claim. Angela Standish, BJC’s director of claims and litigation, testified that there was “absolutely” no reason why a reserve above BJC’s $3 million self-insured retention had to be set and that Columbia’s reserve of $5 million was neither proper nor reasonable. She also outlined various considerations that she said would tend to lessen BJC’s potential losses on the claim. She also stated that Columbia had never expressed concerns about her reserving practices. BJC also presented testimony that after a claims audit in August 2000, the “general consensus” was that “things were fine.”
The district court granted Columbia’s pre-trial motion for partial summary judgment on the issue of compensatory damages (as it pertained to the $1.2 million price quoted by Zurich), as well as Columbia’s motion to strike BJC’s prayer for punitive damages. The district court also granted Columbia’s motion to exclude evidence pertaining to Columbia’s financial and human resources practices and goals. This evidence included testimony suggesting that Columbia sought to shed or increase premiums for medical malpractice insurance clients. Following the presentation of BJC’s case, the district court granted Columbia’s motion for judgment as a matter of law.
II.
In granting a judgment as a matter of law against BJC, the district court concluded that, under the contract, Columbia had the right to determine incurred loss and that BJC had offered no evidence that Columbia’s incurred loss determination was made in bad faith.
We turn first to BJC’s claim that judgment as a matter of law against BJC was improper because the incurred loss condition is a condition subsequent. BJC contends that because the incurred loss condition is a condition subsequent, it was Columbia — not BJC — that bore the burden of presenting evidence pertaining to incurred loss. In other words, BJC argues that incurred loss was not an element of its case and thus any infirmities in BJC’s evidence could not serve as a proper basis for a judgment as a matter of law against it.
“Courts should not construe contract provisions to be conditions precedent unless required to do so by plain, unambiguous language or by necessary implication.”
James E. Brady & Co., Inc. v. Eno,
992 F.2d 864, 869 (8th Cir.1993) (quoting
Kansas City S. Ry. Co. v. St. Louis-San Francisco Ry. Co.,
509 S.W.2d 457, 460 (Mo.1974) (internal quotation marks omitted)). Here, the contract’s language does not plainly indicate a condition precedent. In fact, the language suggests that the incurred loss condition is a condition subsequent.
Accordingly, we agree with BJC that the incurred loss condition is a condition subsequent. We also agree that defendants typically have the burden of proving the occurrence of conditions subsequent.
Peabody Holding Co., Inc. v.
Costain Group PLC,
813 F.Supp. 1402, 1416 (E.D.Mo.1993);
see also Truck Ins. Exch. v. Hunt,
590 S.W.2d 425, 427 (Mo.Ct.App.1979) (noting that if an insurer seeks to avoid coverage because of the occurrence of a condition subsequent, the insurer has the burden of proof). When the contractual provision in question reposes discretion in the defendant, however, the plaintiff must show that this discretion was exercised in bad faith.
See Brozo v. Oracle Corp.,
324 F.3d 661, 667-68 (8th Cir.2003) (judgment as a matter of law was warranted where contract accorded discretion to the defendant and there was no allegation that the discretion was exercised in bad faith).
In this case, the contract allocated to Columbia the right to determine incurred loss. The contract specifies that Columbia would utilize a subdivision of Columbia, CNA HealthPro Claim Consultants (CNA), “[f]or the purposes of determining incurred losses.” (Exhibit 113). This amounts to a grant of discretion because, as the evidence indicates, the determination of incurred loss is not an exact science and requires actuaries to exercise a certain measure of actuarial judgment. Although one could debate the scope of Columbia’s discretion, we will assume that Columbia had broad discretion in determining incurred loss.
Any latitude Columbia might have had under the contract, however, was constrained by the covenant of good faith and fair dealing that Missouri law finds implicit in every contract.
Cf. Farmers’ Elec. Coop., Inc. v. Missouri Dep’t of Corr.,
977 S.W.2d 266, 271 (Mo.1998) (“Missouri law implies a covenant of good faith and fair dealing in every contract.”). This good faith requirement extends to the manner in which a party employs discretion conferred by a contract.
Mo. Consol. Health Care Plan v. Cmty. Health Plan,
81 S.W.3d 34, 45 (Mo.Ct.App.2002). To establish a violation of the covenant of good faith and fair dealing, it is not enough for a plaintiff to show that a party invested with discretion made an erroneous decision.
Id.
at 48. Instead, the plaintiff must show that the party exercised its discretion “in such a manner as to evade the spirit of the transaction or so as to deny [the other party] the expected benefit of the contract.”
Id.
at 46. The covenant of good faith and fair dealing does not establish a “general reasonableness requirement.”
Schell v. LifeMark Hospitals of Missouri,
92 S.W.3d 222, 230 (Mo.Ct.App.2002). Unreasonableness can serve only as “evidence of subjective intent to undermine fulfillment of the contract.”
Id.
at 231.
To establish a submissible case of bad faith, then, BJC was required to do more than offer evidence that Columbia’s incurred loss determination was flawed or unreasonable. Instead, BJC had to present evidence that Columbia’s determination of incurred loss was directed toward evading the spirit of the contract or denying BJC the benefit of its bargain. In this case, the benefit for which BJC bargained was three years of reinsurance coverage guaranteed at a fixed premium. Accordingly, it would have been bad faith, for example, for Columbia to have selected a particular method for calculating incurred loss or to have made other judgments in the process of determining incurred loss that were specifically designed to trigger a contractual provision enabling Columbia to
avoid its third-year obligation. This appears to be the theory advanced by BJC.
Columbia contends that it is not bad faith for a party to act on the basis of financial self-interest. Financial self-interest, however, will not excuse a party from performing its contractual obligations in good faith and fair dealing.
See Venture Associates Corp. v. Zenith Data Sys. Corp.,
96 F.3d 275, 279-80 (7th Cir.1996) (noting in a case where the parties contracted to negotiate a final agreement that although it would not be bad faith for a party to act in its financial self-interest and demand a higher price, it would constitute bad faith if the party’s purpose in demanding a higher price is to “scuttle the deal”). We have observed that “the implied duty of good faith and fair dealing does not extend so far as to undermine a party’s general right to act on its own interests in a way that may incidentally lessen the other party’s anticipated fruits from the contract.”
Countrywide Services Corp. v. SIA Ins. Co.,
235 F.3d 390, 393 (8th Cir.2000) (quoting
M/A-COM Security Corp. v. Galesi,
904 F.2d 134, 136 (2d Cir.1990)) (internal quotation marks omitted). The operative word in that passage, for purposes of this case, is “incidentally.” It would not necessarily constitute bad faith, for instance, for Columbia to employ a risk-averse method of calculating incurred loss that might have the “incidental” consequence of triggering the incurred loss condition. As John Pierce wrote in his report, the incurred loss provision functions as a kind of “escape clause” in the “unusual event that the results on the BJC Excess policy prove[] to be much worse than expected.” (Exhibit 61). A certain measure of caution, then, would be consistent with the bargain struck by the parties and not constitute an act of bad faith, even if it incidentally resulted in the calculation of a higher incurred loss ratio. It would not, however, be permissible for Columbia to exercise its discretion in a manner designed to “scuttle the deal,”
Venture Associates Corp.,
96 F.3d at 279, however much in Columbia’s best financial interest it might have been to reduce its profile in the area of medical malpractice insurance and shed clients in that line.
Because Columbia’s broader business motivations and goals bear on how and for what purposes Columbia employed its discretion, evidence that Columbia was seeking to shed medical malpractice clients like BJC was relevant to the question of good faith. In excluding this evidence, the district court remarked that the relevant issue was whether Columbia breached the contract and not the alleged reasons, good or bad, it might have done so. The excluded evidence relates to the question of Columbia’s good faith, however, and thus is relevant to the question whether the contract was breached. The extent to which the excluded evidence should be admitted is a matter to be evaluated pursuant to Fed.R.Civ.P. 402 and 403.
Columbia claims that the incurred loss ratio was breached with respect to both aggregate claims and a single claim (the Baby C claim). BJC’s evidence of bad faith as it pertains to the aggregate claims centers primarily on the actuarial analysis conducted by Columbia. Considered as a whole, the evidence permits an inference that the purpose of the analysis was to reach a desired and predetermined conclusion,
i.e.,
that the incurred loss ratio had exceeded 75%. There was testimony that Columbia unreasonably selected the same average experience ratio in all three versions of the actuarial analysis despite the fact that the underlying numbers were changing. This evidence, if credited, could indicate that the actual data were not as important to Columbia as arriving at a particular conclusion. BJC also presented evidence of errors in the various versions of the actuarial analysis, most of which
worked in Columbia’s favor. In addition, Columbia appeared largely incapable of explaining how it made a crucial transition between accident year and report year data and was unable to produce certain records. BJC also put on evidence from which a jury could infer that the actuarial assumptions and models it had used to evaluate the third policy year of the BJC account were more stringent than those it utilized in the previous policy years. Finally, there was the excluded evidence that indicated that Columbia was seeking to reduce its profile in the medical malpractice insurance business. From this evidence, a jury could conclude that Columbia wanted to shed BJC as a client (or exact from BJC a higher premium), that Columbia’s actuarial analysis was driven by Columbia’s wish to arrive at an incurred loss figure that would justify cancelling the third year of the contract rather than by the dictates of actuarial judgment, and that Columbia’s determination of incurred loss as it pertained to aggregate claims was therefore made in bad faith.
BJC presented sufficient evidence of bad faith on the Baby C claim as well. First, BJC presented testimony that Columbia’s $5 million reserve was unreasonable. Although unreasonableness is not the same as bad faith, it may circumstantially support the inference that a party sought to undermine the contract.
Schell,
92 S.W.3d at 231. There was also evidence suggesting that Columbia had not previously voiced any concerns about BJC’s reserving practices. In addition, there was the aforementioned evidence that Columbia sought to shed medical malpractice insurance clients. Considering this evidence as a whole, we conclude that BJC presented sufficient evidence for a jury to conclude that Columbia’s determinations regarding the Baby C claim were made in bad faith.
Columbia contends that BJC cannot establish bad faith because BJC’s witnesses expressed only a professional disagreement with Columbia’s incurred loss determination and because none of the witnesses testified that the incurred loss figure assessed by Columbia was unreasonable. This argument is unavailing. As set forth above, there was testimony that the manner in which Columbia calculated incurred loss was unreasonable and evidence that the calculation was made in a manner designed to trigger the incurred loss provision, thus denying BJC the third year for which it had bargained. There was evidence, then, that the incurred loss figure arrived at by Columbia was, for lack of a better word, pretextual. Although the figure might at first glance appear reasonable to a neutral observer if considered out of context (without reference, for example, to the manner in which it was calculated), that will not suffice to preclude a jury from determining that Columbia had acted in bad faith. In sum, then, because BJC presented evidence from which a reasonable jury could have concluded that Columbia acted in bad faith, the district court erred in entering judgment as a matter of law.
III.
The district court, pursuant to Federal Rule of Civil Procedure 12(f), struck ATG’s prayer for punitive damages because it
concluded that ATG’s complaint did not allege fraud with the particularity required by Federal Rule of Civil Procedure 9(b). We agree.
Courts may strike “from any pleading any insufficient defense or any redundant, immaterial, impertinent, or scandalous matter.” Fed.R.Civ.P. 12(f). Judges enjoy liberal discretion to strike pleadings under Rule 12(f).
Nationwide Ins. Co. v. Cent. Mo. Elec. Coop., Inc.,
278 F.3d 742, 748 (8th Cir.2001). We review the allowance of a 12(f) motion for abuse of discretion.
Id.
Striking a party’s pleading, however, is an extreme and disfavored measure.
Stanbury Law Firm, P.A. v. IRS,
221 F.3d 1059, 1063 (8th Cir.2000).
Although Missouri law generally does not permit the recovery of punitive damages for a breach of contract, punitive damages may be permissible if the breach also amounts to an independent tort.
Peterson v. Continental Boiler Works, Inc.,
783 S.W.2d 896, 902-03 (Mo.1990). ATG bases its prayer for punitive damages on a claim of fraudulent misrepresentation. ATG’s fraud claim must therefore comply with the heightened pleading standards of Rule 9(b), which require plaintiffs to plead “the circumstances constituting fraud ... with particularity.” Fed.R.Civ.P. 9(b). Under Rule 9(b), a plaintiff must plead “such matters as the time, place and contents of false representations, as well as the identity of the person making the misrepresentation and what was obtained or given up thereby.”
Abels v. Farmers Commodities Corp.,
259 F.3d 910, 920 (8th Cir.2001) (quoting
Bennett v. Berg,
685 F.2d 1053, 1062 (8th Cir.1982),
adhered to on reh’g,
710 F.2d 1361 (8th Cir.1983) (en banc)). In other words, the party must typically identify the “who, what, where, when, and how” of the alleged fraud.
United States ex rel. Costner v. URS Consultants, Inc.,
317 F.3d 883, 888 (8th Cir.2003). This requirement is designed to enable defendants to respond “specifically, at an early stage of the case, to potentially damaging allegations of immoral and criminal conduct.”
Abels,
259 F.3d at 920. The level of particularity required depends on,
inter alia,
the nature of the case and the relationship between the parties.
Payne v. U.S.,
247 F.2d 481, 486 (8th Cir.1957). “Conclusory allegations that a defendant’s conduct was fraudulent and deceptive are not sufficient to satisfy the rule.”
Commercial Property Investments Inc. v. Quality Inns Int’l Inc.,
61 F.3d 639, 644 (1995). Rule 9(b) should be read “in harmony with the principles of notice pleading.”
Schaller Telephone Co. v. Golden Sky Sys., Inc.,
298 F.3d 736, 746 (8th Cir.2002) (quoting
Abels,
259 F.3d at 920).
The district court determined that ATG had not pled sufficient facts to show that Columbia had acted fraudulently. In light of the record, we cannot conclude that this determination was erroneous. The allegations of fraud are essentially recounted in two paragraphs in the complaint. Neither of these paragraphs, however, alleges fraud with the particularity required by 9(b).
According to paragraph 17, Columbia deleted a number in its second actuarial analysis so that BJC and ATG would not discover that there were errors in Columbia’s first analysis. It is difficult to see, however, how errors in the first actuarial analysis — or Columbia’s attempt to conceal the fact that they were made — could support a fraud claim. After all, as paragraph 18 alleges, Columbia intended for BJC to rely on the second, rather than the first, actuarial analysis.
Later, in paragraph 20, the complaint states that “[t]he second actuarial analysis was false in its conclusion and Columbia Casualty knew that the second actuarial analysis was false in this conclusion at the
time it provided that analysis to BJC and ATG.” In the circumstances of this case, we do not believe that this conclusory assertion adequately alleges the “how” of the alleged fraud. As this case illustrates, actuarial decisions are complex and often require some measure of judgment and discretion on the part of the actuary conducting the analysis. ATG’s complaint expresses disagreement with the conclusion of the second actuarial analysis, but it does not specify where and how the analysis falls short. It does not state, for example, whether the report relies on flawed data, makes incorrect assumptions, or employs suspect methodology. Although such specificity may not be warranted in other cases, we believe that more detail was called for here.
In sum, ATG’s complaint lacks the particularity required by Rule 9(b). We therefore affirm the district court’s order granting partial summary judgment.
IV.
We turn now to the district court’s order precluding BJC from recovering compensatory damages for the additional risk it bore as a result of Columbia’s decision to terminate the contract. BJC is seeking $1.2 million — a sum that reflects what Zurich claimed it would have charged BJC had Zurich offered coverage commencing at $22 million. Although the district court agreed that BJC faces additional risk due to the $3 million coverage gap between the Columbia policy and the Zurich replacement policy and acknowledged that “[t]here may well be ascertainable costs associated with that risk,” the district court concluded that BJC had failed to properly quantify these costs. We agree.
There is no evidence that the Zurich quote adequately mirrors the costs associated with BJC’s additional risk. Although the price quoted by Zurich is at least partly based on Zurich’s estimate of BJC’s future losses, it is unclear whether this is the only component of the Zurich quote. The Zurich quote, for example, may reflect not only the costs associated with BJC’s additional risk, but other factors — such as a profit margin for Zurich — as well. Moreover, there is little in the record regarding how Zurich’s forecast of BJC’s future losses was conducted. Because BJC’s calculation of damages is conclusory and insufficiently supported by the record, the district court’s order precluding BJC from recovering compensatory damages relating to the gap between the Columbia and Zurich policies was proper.
The case is affirmed in part, reversed in part, and remanded for further proceedings.