Full opinion text
LEVIN H. CAMPBELL, Senior Circuit Judge. This is an appeal from a final judgment of the district court in an action brought by a number of foreign reinsurance syndicates, companies and pools against a domestic reinsurance company and related parties. At issue are reinsurance contracts (or “treaties,” as they are known) under which plaintiffs, Compagnie De Reassurance D’lle de France, et al., agreed to reinsure portions of risks selected, and also reinsured, by defendant New England Reinsurance Corp. (“NER-CO”). After sustaining heavy losses under these Treaties, plaintiffs sued defendants NERCO, First State Insurance Company (“First State”), and Cameron and Colby Co., Inc. (“Cameron & Colby”), alleging that they had been induced to enter into the reinsurance treaties by fraud, and further claiming breach of contract, violations of Mass.Gen.L. eh. 93A, § 2, and violations of the Racketeer Influenced and Corrupt Organizations Act (“RICO”), 18 U.S.C. §§ 1961-1968. Defendants counterclaimed, alleging breach of contract and violations of Mass.Gen.L. ch. 93A, § 2. Following a 30-day bench trial, the district court found for the plaintiffs on all but the RICO claims. The court ordered rescission of the challenged reinsurance Treaties and ordered defendants to pay plaintiffs $38,118,940.07, representing all sums plaintiffs had previously paid out on losses incurred under the Treaties with credit for premiums received, plus prejudgment interest at 12 percent. Defendants estimate that the net cost to them of the court’s decision, adding together the court’s judgment and the sums plaintiffs have been excused from paying out as reinsurers of various losses, is approximately $106 million. Defendants have appealed from the judgments for plaintiffs on the fraud, contract and Mass.Gen.L. ch. 93A claims. Plaintiffs have cross-appealed from the district court’s dismissal of their RICO claim. For the reasons set forth below, we sustain the district court’s findings and rulings on certain matters; reverse others as being clearly erroneous or legally incorrect; and identify still others that require the district court to make findings and rulings now absent. We, therefore, vacate the district court’s judgments and remand for further proceedings consistent herewith. Our specific dispositions are summarized on pages 93-94 of this opinion. 1. Background The following is an overview. More specific facts will be related as needed in our discussion of the various issues. The defendants are all subsidiaries of the Hartford Group of Insurance Companies (“the Hartford”). First State, based in Boston, Massachusetts, was a primary insurer. NERCO was a Boston-based reinsurer. Cameron & Colby, also based in Boston, provided management, marketing, underwriting, and other services to both First State and NERCO. Neither First State nor NER-CO had employees of its own; their businesses were carried on by employees of Cameron & Colby. Graham Watson, Inc., not a party, was created in 1979 as an unincorporated division of Cameron & Colby; it became the latter’s wholly owned subsidiary in mid-1980. Graham Watson’s role was to provide marketing and underwriting services in the facultative reinsurance venture that is the subject of this litigation. The underlying casualty and property risks germane to this case were located in North America. Individuals and entities wishing to insure against these risks procured policies of insurance from primary insurers. The latter then purchased reinsurance from NERCO in order to indemnify themselves in whole or in part against losses sustained under the primary policies they had issued. Not wanting to keep all the exposure that it had assumed as a reinsurer, NERCO itself — often acting with and through Graham Watson — sought reinsurance on the London insurance market, resulting in the arrangements with which this lawsuit is concerned. Under these reinsuring agreements — the so-called System and Non-System (“SANS”) Treaties — many syndicates at Lloyd’s of London and other overseas reinsurance entities (some of whom are the plaintiffs in this case) agreed to provide continuing reinsurance to NERCO on a portion of each risk it reinsured. In industry terminology, NER-CO, having been “ceded” the risks by the primary insurers, became a “retrocedent,” the plaintiffs became “retrocessionaires,” and the agreements between them were “retro-eessional” treaties. The plaintiff retrocessio-naires agreed to indemnify NERCO for a portion of any losses NERCO might sustain in its reinsurance of primary insurers. In return, NERCO promised to acquire (“produce”), evaluate (“underwrite”), and price (“rate”) the risks and to share with plaintiff retrocessionaires, subject to its retention of certain commissions, a portion of the premium it received. A. Signing the Treaties In 1979, NERCO retained a U.S. broker, G.L. Hodson, to assist it in arranging for this reinsurance on the London market. Towards this end, Graves Hewitt, the CEO of Cameron & Colby, and his associates drafted and circulated in late 1979 a document known as the Placing Information. This document stated that Cameron & Colby had established the Graham Watson division after studying facultative reinsurance operations in North America and after receiving the approval and support of the Hartford and ITT. The stated purpose of the division was: 1. To participate in the property and casualty facultative reinsurance business which is currently dominated by the direct writers. 2. To rationalise [sic] the facultative placements of both the Hartford and the First State not only from an administration [sic] point of view but also to provide the retrocessionaires with a broad cross section of facultative reinsurance emanating from these two companies. According to the Placing Information, Graham Watson was charged with penetrating the “non-brokered ... direct professional reinsurance market,” leaving “[fjacultative reinsurance emanating from reinsurance intermediaries ... [to] continue to be written separately through NERFAC,” the latter being an existing in-house entity that had, for some time, been writing reinsurance for the defendants. The Placing Information was circulated to, among others, several European sub-brokers retained by Hodson to act on NERCO’s behalf in seeking potential retro-cessionaires. In late 1979, Hewitt traveled to London accompanied by Thomas Hearn, a Hodson employee. Aided by employees of sub-broker Sedgwick Payne, they approached Ralph Bailey, the head underwriter for plaintiff Terra Nova Insurance Company Limited, and described to him the proposed reinsurance plan. Sedgwick-Payne’s brokers thereafter negotiated with Bailey the “slips” spelling out the terms of the treaties. With Bailey agreeing to act as “lead underwriter” for the London market companies, the brokers approached Ron Kellet, head underwriter for plaintiff B.P.D. Kellet & Others, a Lloyd’s syndicate, with the request that he act as lead underwriter on behalf of all other Lloyd’s syndicates. After the leads had stamped and initialed the slips, indicating the proportion of the total risk they were bound to accept, the slips were separately presented for approval to the underwriters for each of the plaintiffs, each of whom indicated his or her acceptance of a portion of the risks by initialing the slip. These slips constituted, in abbreviated form, the contracts between the cedent NERCO and the various retrocessionaires. Briefly summarized, the slips provided that the subject matter of the Treaties was “Business classified by the Reassured [NERCO] as Property and Casualty Facultative Assumed business pi'odueed and underwritten by the Graham Watson division of Cameron & Colby Co., Inc.” They also stated that the Lead Underwriter had authority to require exclusion of certain types of risks, and to agree to the final wording of the formal contract. NERCO was to retain a minimum of $250,000 of each risk ceded, and as respects system business (i.e., risks written by First State and other Hartford entities, infra ), was not to cede more than 50 percent of the original reinsurance limit of any given risk to the Treaties, and was to eo-reinsure for 10 percent participation on each such risk. The slips also specified the commission structure and various other conditions of the Treaties. The slips did not incorporate the Placing Information as such. Each underwriter subsequently signed Treaty Wordings, formal contracts containing a more elaborate statement of the parties’ agreements. These were based on the slips, and the parties agreed that, in the event of any inconsistency, the slips would control. The first set of SANS Treaties ran for the eleven month period from February 1 through December 31, 1980. Thereafter, those plaintiffs who desired to continue for another year indicated their willingness to join by initialling new slips and ultimately executing new Treaty Wordings for 1981. Successive Treaties were entered into- for 1982 and for 1983. Some of the plaintiffs entered into Treaties for each of the four years; others were parties to the Treaties for only one, two, or three of those years. The Treaties were open to renegotiation each year, and certain changes, e.g., relating to commission structure and the like, were in fact made. For each Treaty year there were actually two slips prepared, one for property business and one for casualty business. The business covered by each slip was further divided into “system business” and “non-system business.” “System business” denoted risks written by member companies of the Hartford, and included, among others, First State and the Hartford itself. “Non-system business” referred to risks written by any other primary insurer. As a condition of participation, although not included in the written contracts, Bailey insisted that no non-system business be ceded to the SANS Treaties for the first year. He testified in his deposition that this was because he felt that the system business was “the steadier, better part of the portfolio.” B. Performance of the Treaties Once the Treaties were fully placed for the 1980 treaty year, NERCO began retroceding to the plaintiffs portions of the risks it was reinsuring. Central to the plaintiffs’ present complaint, and to the district court’s finding of liability, are the source and nature of NERCO’s business as ceded to them. In the first year, over 95 percent of the business so ceded was system business. However, few, if any, of the risks reinsured were from Hartford companies other than First State. In the ensuing three years, the proportion of system business declined in favor of non-system business to less than 50 percent. There was evidence that defendants had hoped that system business would grow and that NERCO and its retrocessionaires would obtain more reinsurance business directly from the other Hartford companies, in addition to First State, but that these hopes were not realized. The proportion of non-system business rose steadily after the first year, but the non-system business was of a kind which plaintiffs contend, and the district court found, was different from that represented in the Placing Information. The court construed the Placing Information as representing “that Graham Watson would produce ‘non-system’ reinsurance business directly from primary insurance companies without the use of intermediaries.” In support of the court’s construction, plaintiffs point to representations in the Placing Information that Graham Watson did not intend to seek reinsurance “on a wholesale basis from all and sundry” but rather to develop a close working relationship “with selected primary companies.” The Placing Information stated that non-brokered business “placed significantly with the direct professional reinsurance market” characterized over 80 percent of United States facultative reinsurance. The Placing Information also stated that Graham Watson was “charged with the responsibility of penetrating this business.” Notwithstanding these announced intentions in the Placing Information, most of defendants’ growing non-system -business after the end of 1980 was, in fact, obtained from intermediaries— to wit, brokers and Managing General Agents (“MGAs”). MGAs serve as agents of primary insurance carriers with authority to underwrite and place certain business on the insurers’ behalf. Defendants received the majority of their non-system business, portions of which were then ceded to the plaintiffs under the SANS Treaties, from Baccala & Shoop Insurance Services, an MGA representing a variety of primary insurance companies. Baccala & Shoop worked closely with the broker, G.L. Hodson; in fact, they were owned by the same entity. 1. Semi-Automatic and Automatic Facilities Another key issue in the present litigation stems from the fact that, during the annual periods covered by the Treaties, almost all of the non-system business that defendants produced, and shared with the plaintiffs, was underwritten using what are called “semiautomatic facilities.” (A “facility” is an agreement setting out, among other things, the rules under which a reinsurer will rein-sure risks ceded by the other party.) Defendants insist that semi-automatic facilities were perfectly consistent with the representations in the slips and Treaties that the reinsurance to be ceded to plaintiffs would be “business classified by the Reassured [NER-CO] as Property and Casualty Facultative Assumed business.” (Emphasis supplied.) Plaintiffs sharply dispute this. Calling facul-tative underwriting the “fundamental material term in the SANS Treaties,” the district court agreed with plaintiffs that the term “facultative” included only reinsurance that a reinsurer underwrites and negotiates with the primary insurer on a risk-by-risk individual certificate basis in advance, i.e., a certificate of reinsurance is issued for each risk after the reinsurer has first looked into and approved reinsuring that particular risk. Under the semi-automatic method that defendants mostly used in underwriting non-system risks, defendants’ underwriters did not evaluate risks one at a time in advance of the issuance of a policy of reinsurance on each risk. Instead, in contracts called “Master Facultative Certificates” (“MFCs”), NERCO agreed with an MGA, broker, or primary insurer that the latter entity could issue reinsurance upon risks of described types, and upon certain conditions and with certain limits, prior to defendants’ underwriters’ scrutiny and approval of the risk. After the reinsurance attached to each risk, however, the agent or ceding company would send to Graham Watson a “risk bordereau” — a document identifying and providing a summary of information as to that, and any other, risks reinsured within the reporting period. Graham Watson then had a brief period after receipt of the bordereau, for example 72 hours, within which to cancel the reinsurance on a particular- risk if it so desired, cancellation to take effect within a specified period, say, 14 days. Defendants contend, and presented evidence at trial, that the semi-automatic facility is commonly classified in the industry today as a form of facultative reinsurance. They concede that, in an earlier era, “facultative” was a term applied only to reinsurance individually underwritten on a risk-by-risk basis in advance of binding. But while accepting that the reinsurer’s right to reject individual risks remains a general feature of facultative reinsurance, defendants contend that this feature is adequately preserved in the more economical and streamlined semi-automatic facility. Defendants also used, in a few instances, a variation known as an “automatic facility.” Under this type of facility, rather than having the right to cancel an individual risk, the reinsurer has the right to cancel the entire facility on very short notice. Even without the right to cancel a particular risk, defendants argue that this was “facultative,” since the reinsured would, as a practical matter, agree to cancel individual risks rather than face cancellation of the entire facility. Moreover, the reinsured retained the freedom to cede or not to cede a particular risk, which is not the case in treaty reinsurance. Automatics comprised only a small portion of the non-system business, most of which was underwritten using semi-automatics. 2. The First State Business With regard to system business (which was almost exclusively with First State), the defendants did not use a risk bordereau, nor did they ever enter into a formal contractual arrangement spelling out First State’s and NERCO’s relationship in respect to the latter’s reinsuring of risks later assigned under the SANS Treaties. There was evidence, however, indicating how matters worked in practice. In practice, First State’s underwriters had the power initially to commit NERCO and the SANS Treaty signatories to the reinsuring of individual risks primarily insured by First State. The reinsurance was evidenced by a layoff sheet that First State prepared; each layoff sheet identified a First State risk that NERCO and the Treaty signatories were to reinsure, and provided a brief summary of information about that risk. A packet containing many of these layoff sheets was periodically provided by First State to Graham Watson, whose underwriter could study the risks and would have the right to cancel the reinsurance at will. Defendants contend that this method was “facultative” because each risk was individually evaluated in due course by a Graham Watson underwriter based on the information provided on the layoff sheets — and by follow-up phone and face-to-face inquiries, as well as by means of microfiches which reproduced First State’s entire underwriting file for a risk, and were available upon request— and it was understood that the reinsurance was subject to cancellation at will by Graham Watson. They point out that because First State’s and Graham Watson’s employees were under the same roof and answerable to the same bosses, the latter’s underwriters could informally influence First State not to cede business the latter did not wish, as further evidence of their facultative control. Notwithstanding the absence of a written understanding between Graham Watson and First State, the district court found, after hearing the evidence, that, “Graham Watson underwrote all ‘system business’ ... by the ‘automatic’ and/or ‘semi-automatic’ method of underwriting.” Since practically all system business was with First State, this finding grouped that underwriting with the explicit semi-automatic and automatic facilities used in non-system business. 3. Further Performance At trial, plaintiffs made much of the absence of proof of particular occasions when defendants had ever actually rejected a risk listed in a bordereau or layoff sheet. Plaintiffs also sharply questioned whether the information in the bordereaux and layoff sheets was sufficient to allow for adequate underwriting (evaluation) of individual risks. Defendants responded by emphasizing that, whether or not used, the right to reject at all times existed, and by pointing to evidence that its underwriters adequately reviewed the risks and had other means — personal inquiries, telephone calls, inspection of First State files, and so on — to make inquiry in doubtful cases. Defendants’ evidence also indicated that Graham Watson conducted periodic audits of the underwriting practices of MGAs and others in order to assess compliance with the terms of the various facilities. The district court found no evidence that defendants had rejected any risks and found that Graham Watson’s underwriting of individual risks was inadequate. In any case, while defendants wrote some small percentage of reinsurance under the SANS Treaties that was facultative in the traditional sense of advance risk-by-risk underwriting, most of the reinsurance produced under the SANS Treaties was underwritten either under some variety of the semi-automatic facility or, in the case of First State system business, under the informal in-house procedures previously described. And, as mentioned above, over the four years of the SANS Treaties, one MGA, Baccala & Shoop, furnished almost all of the non-system business to defendants. Non-system business was found by the court to constitute approximately one-half of the treaty business during the four year period. The agreement with Ralph Bailey to avoid non-system business for the first year was not to the liking of Graham Watson, whose employees felt that non-system business would be a steadier source of income for the Treaties. Bailey also made known his dislike of MGAs and his reluctance to allow MGA business to be ceded to the SANS Treaties. Bailey testified that, because MGAs did not themselves bear any risk, they did not underwrite as carefully as did underwriters on the payrolls of the primary companies, and hence the business produced through them was of a lower quality. Again, this was not to the liking of Graham Watson; one internal memorandum, dated December 11, 1980, stated that “Ralph Bailey has an aversion to MGAs and he will have to be approached rather delicately because a good deal of the business going into this facility will be on business which is designed to provide a real flow of business from a single source.” This memorandum also noted that Tom Hearn, of Hod-son, would travel to London on December 15, 1980 to attempt to overcome this aversion. Responding to repeated requests from Graham Watson employees, Bailey agreed at some time during the first year to begin allowing non-system business to be ceded to the Treaties. While he expressly agreed to the cession of certain MGA business during the first year (from an MGA known as the London Agency), he did so only with great reluctance. However, he did not request or insert an exclusion for MGA business in the slips or Treaty Wordings for subsequent years, as he could have done. No such express exclusion was ever inserted. 4. Renewal of the Treaties The SANS Treaties were continuous contracts subject to cancellation “upon 120 days prior written notice at December 31, 1980 or any subsequent December 31st.” This allowed any desired adjustments to be made in the terms of the Treaties on a yearly basis. In practice, all of the retrocessionaires can-celled during the 120-day period preceding December 31, 1980 and then initialled new slips for the next calendar year. In order to induce renewal, Graham Watson, again through Hodson and the European sub-brokers, disseminated a document referred to as the 1981 Anniversary Information. In addition to listing losses in excess of $50,000 reported through September 30, 1980, and providing a summary of the business ceded thus far, the Anniversary Information included the following statements: To date, the preponderance of the business has been assumed from First State Insurance Company and written on a pro rata basis. Non-System business represents a relatively small proportion of the total and what has been written is limited to Casualty business on an excess of loss basis emanating from Baccala and Shoop Insurance Services. Because of the competitive climate in the United States, Non-System business will develop more slowly than originally anticipated. It continues to be the posture of Graham-Watson not to seek business on a wholesale basis but rather to develop close working relationship [sic] with selected primary sources. On March 23, 1981, a meeting was held in Boston to discuss the performance of the SANS Treaties. In attendance were Ralph Bailey and several employees of Hodson and Graham Watson. One major topic of conversation was the inclusion of MGA business. Bailey asked the Graham Watson underwriters for their opinion of Baccala & Shoop, and was told by Bob Wright, the property underwriter, that Wright knew most of Baccala & Shoop’s home office people and was comfortable with them. Later in the meeting, however, Bailey stated that he would not consider any new MGA business for the facility. He did not, however, make this a contractual requirement by inserting an exclusion for MGA business in the slip at the next renewal. At the close of the second year, a 1982 Anniversary Information was disseminated, which again provided a list of losses and a summary of the business. This document also included figures as to overall loss experience through September 30, 1981, which disclosed that the SANS Treaties were losing money. Indeed, the loss ratio for the 1980 Treaties was an alarming 248.65 percent. In addition, the 1982 Anniversary Information included the following statements: The rating basis of these treaties is being amended with effect from 1st January 1982 to more accurately reflect the basis used by Graham-Watson. All business other than that assumed from the First State which is a “system” company is being written on a net rated basis in that Graham-Watson is quoting their price and if a ceding commission is required by the original company, this is then added to the premium required by Graham-Watson The current sources of business is [sic] as follows:— FIRST STATE INS. CO. TWIN CITY per Baccala and Shoop Insurance Services ST. PAUL FIRE & MARINE NORTHBROOK CRUM & FORSTER CNA ROYAL INS. CO. CHUBB AND SON AETNA CASUALTY & SURETY Plaintiffs point out defendants’ failure to mention, other than in the case of Twin City, that certain of the listed primary insurers acted through Baccala & Shoop or other intermediary. It appears that no formal anniversary information was prepared for 1983, the last year of the SANS Treaties, although letters were sent to the retrocessionaires containing a list of losses, a summary of the business, and notification of various changes that had been made over the past year, none of which are material here. However, the retrocessio-naires were told that the treaties were “continuing for 1983 basically as before.” Following the placing of the SANS Treaties, the plaintiffs at first accepted their shares of the premiums and paid their shares of corresponding losses incurred by NERCO. The losses were considerable, as they were throughout the insurance industry at this time. Beginning as early as the fourth quarter of 1982, however, certain of the plaintiffs ceased paying losses. There was evidence that some of the plaintiffs (in addition to Terra Nova, through Bailey, as related above) began to inquire as early as 1982 about the use of MGAs to obtain business (rather than through the formation of direct relationships with primary insurers), and about the underwriting methods used by the defendants. C. The Present Lawsuit In 1985, some of the plaintiffs retained counsel and sought to conduct a preliminary inspection of NERCO’s books pursuant to a provision in the Treaty Wordings allowing a right of inspection “at all reasonable times for the purpose of obtaining information concerning this contract or the subject matter thereof.” NERCO allowed a seven day preliminary inspection in the fall of 1985, but a dispute then arose between the parties concerning the conditions under which any further inspection was to be conducted. Evidently dissatisfied with the results of this inspection, and concerned about the growing loss ratios of the SANS Treaties, a group of sixteen plaintiffs (of whom seven are no longer parties) commenced this action against NERCO on January 6, 1987. They alleged that they had a contractual right under the treaties to inspect NERCO’s records, and that although they had previously conducted a seven day inspection, a further, more exhaustive evaluation was needed. They sought an order compelling a new inspection. On February 13, 1987, the parties entered a voluntary stipulation allowing, and establishing procedures for, a further inspection to be conducted by Roy T. Ward and four of his employees. On February 27, 1987, the district court entered an order allowing the inspection to continue pursuant to that stipulation. Meanwhile, in late 1986 a second group of reinsurers that had continued to pay losses to NERCO while the parties discussed the possibility of a commutation retained a reinsurance inspection firm, Palange & Associates, to inspect NERCO’s books and records. The inspection was conducted in Boston in 1987. Again, disputes arose as to the scope and methods of this inspection; however, no court action was required to resolve these disputes, and Mr. Palange completed his inspection in the spring of 1988. Following these inspections, on July 12, 1988, the original plaintiffs, now joined by the remainder of the present plaintiffs, moved to amend their complaint. The new complaint omitted the substantive allegations of the original complaint, and deleted the plaintiffs’ request that the treaties be enforced. Instead, the plaintiffs asserted that the treaties had been induced by fraud and should be rescinded. They also asserted claims for breach of contract, violation of Mass.Gen.L. ch. 93A, § 2, and the Racketeer Influenced and Corrupt Organizations Act (“RICO”), 18 U.S.C. §§ 1961-1968. Denying these allegations, defendants asserted the statute of limitations as a defense and counterclaimed for recovery under the challenged treaties, and for treble damages, costs and attorneys fees under Mass.Gen.L. ch. 93A, § 2. Prior to trial the defendants moved for summary judgment on the statute of limitations issue, as well as on a variety of other grounds not important here. The district court held a hearing on the matter on January 15,1992. In a written order dated January 16,1992, the court denied summary judgment on the statute of limitations ground, with no explanation. A jury-waived trial began on April 5, 1993. The statute of limitations was raised again during trial. The court delayed ruling until it could “find out what the facts were.” On the twenty-second day of trial, the judge stated simply that “[i]t seems that there is no problem with Statute of Limitations.” However, the court appeared to entertain the issue again two days later, accepting a deposition into evidence after defendants argued it was relevant to the statute of limitations. The trial concluded on May 19, 1993, and the district court entered a memorandum and order on June 7, 1993. See 825 F.Supp. 370 (D.Mass.1993). The court held that the defendants had induced the plaintiffs to enter into the treaties by means of fraudulent misrepresentations, had breached their contracts with plaintiffs, and had engaged in unfair and deceptive trade practices in violation of Mass.Gen.L. ch. 93A. The RICO count was rejected. The court made no mention of defendants’ statute of limitations defenses. By way of relief, the district court ordered rescission of the challenged reinsurance contracts and ordered defendants to repay to plaintiffs all sums plaintiffs had previously paid' out on losses incurred under those contracts with credit for premiums paid to the plaintiffs, plus prejudgment interest at 12 percent. Judgment was entered for the plaintiffs in June 30, 1993 in the amount of $37,501,701.12 plus postjudgment interest and costs. Following several motions to amend this judgment, a new judgment was entered on September 21,1993 in the amount of $38,118,940.07, which listed the amount due to each individual plaintiff, as opposed to the lump sum stated in the first judgment. The defendants’ filed their notice of appeal from the fraud, contract, and ch. 93A claims on October 19,1993; the plaintiffs’ filed their notice of appeal from the adverse RICO finding on November 2, 1993. Motions relating to the plaintiffs’ requests for attorney’s fees and costs are still pending in the district court. D. The District Court’s Findings In its memorandum and order of June 7, 1993, the district court found that the defendants made, and the plaintiffs relied upon, “four material representations” to secure the plaintiffs’ participation in the SANS Treaties. These were: 1. That Graham Watson would produce and underwrite property and casualty fa-cultative reinsurance. This representation mean[t] that Graham Watson would underwrite reinsurance on an individual, risk-by-risk, certificate basis. 2. That Graham Watson would produce such reinsurance directly from system and non-system original insurers without the use of any intermediaries. This representation mean[t] that Graham Watson would be a direct writer of reinsurance from the original insurer, which reinsurance cessions would not be brokered. 3. That the Hartford Companies, along with First State, would be the “system business” original insurers. This representation mean[t] that the Hartford Insurance Group would be the source of “system business.” The Hartford Insurance Group is made up of the so-called Hartford Companies and First State, an excess and surplus line carrier. 4. That Graham Watson would seek fa-cultative reinsurance business from selected primary companies, rather than on a wholesale basis. This representation meanft] that Graham Watson would assume reinsurance from selected insurance companies, not reinsurance companies or Managing General Agents, that is, from risk-bearing insurance entities. 825 F.Supp. at 376-77 (emphasis in original). The district court found that although business had been assumed from several of the Hartford Companies, including First State, Hartford Specialty Company, Nutmeg Insurance Company, and Twin City Insurance Company, all of this business with the exception of the First State business had been classified as non-system business. The court listed, as sources of non-system business, a number of primary insurance companies, but also a number of brokers and MGAs, and found that “[t]he majority of ‘non-system business’ emanated from Baccala and Shoop, a Managing General Agent, through the intermediary G.L. Hodson.” Id. After a further discussion of Graham Watson’s underwriting practices, the court stated: Upon a review of the evidence, the Court finds that Graham Watson did not facultatively underwrite, that is, underwrite on an individual risk certificate basis as represented, any of the “system business” nor virtually any of the “non-system business”; Graham Watson underwrote all “system business” and virtually all “non-system business” by the “automatic” and/or “semi-automatic” method of underwriting. The Court also finds, on the basis of the evidence, that most of the “non-system business” emanated from MGAs through the use of intermediaries and from intermediaries themselves, and was not produced from primary risk-bearing insurance entities directly. Although the plaintiff reinsurers were aware that Baccala and Shoop, an MGA, had ceded to the SANS Treaties approximately five percent of the total business during the first year, 1980, they were never apprised that, during the ensuing three years, Baccala and Shoop would cede the majority of “non-system business” and that other MGAs and intermediaries would cede, in conjunction with Baccala and Shoop, most of the “non-system business” to the SANS Treaties. “Non-system business” constituted, over the course of the SANS Treaties, approximately one-half of the total business ceded to the Treaties. 825 F.Supp. at 379 (emphasis in original). With respect to the plaintiffs’ fraud claims, the district court stated that the plaintiffs understood that the term “facultative,” as used in the SANS Treaties, was being used in its “standard and traditional sense, namely, underwriting on a risk-by-risk certificate basis.” It found that NERCO was aware of this understanding on the plaintiffs’ part, “and was well aware that it, itself, was secretly using the term in a special sense without ever disclosing such special meaning” to the plaintiffs, and that this was therefore a knowing misrepresentation. As to the breach of contract claims, the district court found that NERCO did not keep, and never intended to keep, its contractual promise to underwrite risks obtained directly from selected primary sources on an individual risk-by-risk certificate basis. II. Preliminary Matters A. Standard of Appellate Review When reviewing the findings of a district court sitting without a jury, “ ‘the court of appeals cannot undertake to decide factual issues afresh.’ ” Jackson v. Harvard Univ., 900 F.2d 464, 466 (1st Cir.1990) (quoting Reliance Steel Prod. Co. v. National Fire Ins. Co., 880 F.2d 575, 576 (1st Cir.1989)), cert. denied, 498 U.S. 848, 111 S.Ct. 137, 112 L.Ed.2d 104 (1990). We may set aside findings of fact by the district court, whether based on oral or documentary evidence, only if “clearly erroneous,” and with due regard “to the opportunity of the trial court to judge of the credibility of witnesses.” Fed.R.Civ.P. 52(a). A finding is clearly erroneous when, ‘“although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed.’ ” See Anderson v. City of Bessemer City, 470 U.S. 564, 573, 105 S.Ct. 1504, 1511, 84 L.Ed.2d 518 (1985) (quoting United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 542, 92 L.Ed. 746 (1948)), reh’g denied 333 U.S. 869, 68 S.Ct. 788, 92 L.Ed. 1147; accord Brown Daltas & Assoc., Inc. v. General Accident Ins. Co. of Am., 48 F.3d 30, 36 (1st Cir.1995). Review of legal rulings is, however, de novo. “[I]f the trial court bases its findings upon a mistaken impression of applicable legal principles, the reviewing court is not bound by the clearly erroneous standard.” Inwood Lab., Inc. v. Ives Lab., Inc., 456 U.S. 844, 855 n. 15, 102 S.Ct. 2182, 2189 n. 15, 72 L.Ed.2d 606 (1982) (citing United States v. Singer Mfg. Co., 374 U.S. 174, 194 n. 9, 83 S.Ct. 1773, 1784 n. 9, 10 L.Ed.2d 823 (1963)); accord Cumpiano v. Banco Santander Puerto Rico, 902 F.2d 148, 153 (1st Cir.1990). “[T]o the extent that findings of fact can be shown to have been predicated upon, or induced by, errors of law, they will be accorded diminished respect on appeal.” Dedham Water Co., Inc. v. Cumberland Farms Dairy, Inc., 972 F.2d 453, 457 (1st Cir.1992) (citing RCI Northeast Servs. Div. v. Boston Edison Co., 822 F.2d 199, 203 (1st Cir.1987)). Application of these principles is complicated in the present case by the district court’s disregard, in several key areas, of Rule 52(a)’s further injunction that, “[i]n all actions tried upon the facts without a jury ..., the court shall find the facts specially and state separately its conclusions of law thereon.” Fed.R.Civ.P. 52(a). Rule 52(a) imposes on the district court “an obligation to ensure that its ratio decidendi is set forth with enough clarity to enable a reviewing court reliably to perform its function.” Touch v. Master Unit Die Products, Inc., 43 F.3d 754, 759 (1st Cir.1995). The court made no findings and rulings whatsoever on the important statute of limitations issues discussed infra, nor, in general, did it make subsidiary findings resolving disputed evidence. Thus in finding that defendants had committed fraud in promising “facultative” reinsurance, the court stated that all parties understood that term to mean risk-by-risk, individual certificate underwriting, but made no attempt to distinguish or explain the great body of evidence indicating a broader meaning. Its finding that all plaintiffs relied on the Placing Information is similarly bereft of explanation as to how this could be, given the absence of proof of reliance in a number of instances. These omissions have required us to remand for certain additional findings. Where possible, however, we have disposed of key issues or, if that was impossible, have set out a guiding legal standard for use on remand. In sum, we have endeavored to dispose of as much of the appeals as we properly can at this juncture. B. Choice of Law We dispose first of certain contentions raised with regard to legal standards. Plaintiffs challenge defendants’ assertion that the SANS Treaties contain an express choice of law provision providing for the application of Massachusetts law to plaintiffs’ common law fraud and contract claims. In fact, Article XVIII of the SANS Treaties merely provides that if a dispute is litigated, plaintiffs will submit to the jurisdiction of any court of competent jurisdiction in the United States, and “all matters hereunder shall be determined in accordance with the law and practice of such court.” But while plaintiffs’ point is well taken, they go on to concede that “[i]n this case, Massachusetts choice of law principles dictated the application of Massachusetts substantive law to plaintiffs’ common law claims.” Given the parties’ (and the lower court’s) general acceptance of Massachusetts law, albeit on different theories, and in the absence of a preferable choice, we shall apply Massachusetts law except as otherwise noted. See Bird v. Centennial Ins. Co., 11 F.3d 228, 231 n. 5 (1st Cir.1993) (accepting parties’ agreed choice of law where there was a “reasonable relation” between the litigation and the forum whose law had been selected). C. The Burden Required to Prove Fraud The defendants argue strenuously, and the district court stated, that the plaintiffs were required to prove fraud by “clear and convincing evidence.” The plaintiffs respond that under applicable Massachusetts law fraud need not be shown by anything more than the ordinary preponderance of the evidence standard applicable to civil cases in general. Review of Massachusetts law indicates that the plaintiffs are right. In Callahan v. Westinghouse Broadcasting Co., Inc., 372 Mass. 582, 363 N.E.2d 240 (Mass.1977), the Massachusetts Supreme Judicial Court (“SJC”) commented on the burden of proof applicable to a libel action governed by Gertz v. Robert Welch, Inc., 418 U.S. 323, 94 S.Ct. 2997, 41 L.Ed.2d 789 (1974) and New York Times Co. v. Sullivan, 376 U.S. 254, 84 S.Ct. 710, 11 L.Ed.2d 686 (1964). Recognizing that the Supreme Court required “clear and convincing proof’ in a libel case, the SJC nonetheless noted that, the words “clear and convincing proof’ had not been discussed in our cases [other than in the libel context] because the phrase had not been used theretofore in this Commonwealth. Indeed, because of the vagueness of an intermediate standard of proof, we have not looked with favor on the use of such a standard. Callahan, 372 Mass. at 583, 363 N.E.2d at 241. We have not found any Massachusetts case stating that a “clear and convincing” standard should be applied in a common law fraud case, nor have we found any indication that the SJC has, since Callahan, looked with greater favor on introducing a “clear and convincing” standard of proof to cases where none otherwise exists. See Paul J. Liacos, Handbook of Massachusetts Evidence 38-39 (5th ed. 1981) (stating that the burden of proof in Massachusetts civil cases is “by a preponderance of the evidence” and listing those few issues, not including fraud, where a higher standard is required, including proof of a gift causa mortis, contents of a lost will, irregularity of official proceedings, and malice in a defamation action); see also 9 John Henry Wigmore, Evidence in Trials at Common Law § 2498 (Chadbourn rev. 1981) (noting that “clear and convincing” standard is commonly applied in cases of fraud, but failing to cite, in a comprehensive list of authorities, any Massachusetts case applying this standard). We conclude, therefore, that Massachusetts has not adopted a “clear and convincing” standard in cases of fraud. D. The Duty Owed to the Reinsurers The plaintiffs argue, and the district court found, that the defendants were under a duty to the plaintiffs of utmost good faith (“uberrimae fidei ”). The defendants refer to the same standard. We agree that a reinsurer like NERCO, having obtained by treaty the power to impose significant risks and liabilities upon plaintiff retrocessionaires, owed to them the utmost good faith in its dealings under the treaties. See generally Unigard Sec. Ins. Co., Inc. v. North River Ins. Co., 4 F.3d 1049 (2d Cir.1993). This means that, as the district court properly recognized, defendants owed plaintiffs a duty “to exercise good faith and to disclose all material facts.” In the non-marine context, however, a claim of fraud may not be founded on innocent misrepresentation and concealment. Thus, the district court properly required the plaintiff to prove that the defendant made a false representation of a material fact with knowledge of its falsity for the purpose of inducing the plaintiff to act thereon, and that the plaintiff relied upon the representation as true and acted upon it to his damage. Kennedy v. Josephthal & Co., Inc., 814 F.2d 798, 805 (1st Cir.1987) (quoting Danca v. Taunton Sav. Bank, 385 Mass. 1, 8, 429 N.E.2d 1129, 1133 (1982) (citations omitted)). The standard for fraudulent concealment is similar: Except with respect to marine risks, concealment exists and avoids the policy where the insured has knowledge of a fact material to the risk which honesty, good faith, and fair dealing require that he should communicate to the insurer but which he designedly and intentionally withholds. 9 George J. Couch, Cyclopedia of Insurance Law § 38:2 (2nd ed. 1985) (Couch). Massachusetts’ adherence to the same rule is indicated in Century Indem. Co. v. Jameson, 333 Mass. 503, 504-05, 131 N.E.2d 767, 769 (Mass.1956); see also Unigard, 4 F.3d at 1069 (holding that simple negligence in not disclosing a material fact does not constitute bad faith so as to avoid a policy of reinsurance). III. The Fraud Claims We turn now to the substantive issues, and, initially, to the fraud claim which is pivotal to all the district court’s findings. The district court saw two fundamental issues in the ease, both of them relevant to its finding of fraud. One was “whether Graham Watson did underwrite facultative reinsurance” on the system and non-system business. The other was “whether Graham Watson produced ‘non-system business’ by establishing direct relationships with primary, risk-bearing, insurance companies.” A. “Facultative” Underwriting Plaintiffs argued, and the district court found, that “the parties to the contract” (including, it would seem, defendants themselves) “understood the meaning of that term [“facultative”] in its standard and traditional sense, namely, underwriting on a risk-by-risk certificate basis, the classic meaning of the term.” 825 F.Supp. at 382. According to the court, “NERCO knew” that plaintiffs understood “‘facultative’ in its standard and traditional sense of risk-by-risk certificate underwriting and was well aware that it, itself, was secretly using the term in a special sense without ever disclosing such special meaning to the Plaintiff reinsurers.” Thus, the court concluded, the defendants’ representation that the business would be underwritten on a risk-by-risk individual certificate basis was “knowingly false when made.” Id. 1. No Express Misrepresentation We hold that these findings are clearly erroneous insofar as they attribute to defendants an implicit or express representation that they would engage exclusively in classic risk-by-risk, individual certificate underwriting. The record is without evidence from which a court could find that defendants represented to plaintiffs that the facultative business underwritten by Graham Watson would be limited to individual certificate, risk-by-risk underwriting. To be sure, as the court found, there was evidence that the overseas sub-brokers engaged to represent defendants by their broker, G.L. Hodson, understood “facultative” in the classic risk-by-risk individual certificate sense. Nigel Huntington-Whitely, the employee principally assigned to the SANS Treaty placements by defendants’ sub-broker, Sedgwick-Payne, testified to having this understanding. But he indicated that it “may have just been an assumption,” and could not identify the source of his understanding beyond his sense of what the term “facultative” might mean. It was not established that the sub-brokers were told this by defendants nor that prior to the initial (1980) Treaties the sub-brokers communicated this view to plaintiffs or to defendants during negotiations. To fill this gap, plaintiffs point to Huntington-Whitely’s letter of June 24, 1981 (well over a year after the Treaties were entered into), wherein he states in passing that “Graham Watson is underwriting each risk individually.” However, this statement clearly did not induce the plaintiffs to enter into the 1980 and 1981 SANS Treaties, given its timing. Moreover, it is arguable that the semiautomatic facilities, because they allowed Graham Watson’s underwriters to reject individual risks, were a form of individual risk underwriting and thus not necessarily inconsistent with this statement. Plaintiffs also point to the 1982 Anniversary Information, which states that, on non-system business, “Graham Watson is quoting their price and if a ceding commission is required by the original company, this is then added to the premium required by Graham Watson.” Plaintiffs argue that this specifically describes individual risk negotiation. However, it could just as easily be read to refer to Graham Watson quoting a price during the initial negotiations leading to the formation of a semi-automatic facility. Thus, it is not an explicit promise to perform individual risk-by-risk certificate underwriting. Moreover, this representation, like the statement in Huntington-Whitely’s letter, was made in 1981, and thus could not have fraudulently induced the plaintiffs to participate in the 1980 and 1981 SANS Treaties. Defendants’ chief executive, Graves Hewitt, testified at the trial to having told one of the lead underwriters, Bailey, in 1979, about his dissatisfaction with the method of using a separate certificate as to each risk, and his intention, in connection with the SANS Treaties, to use a single controlling facility for multiple risks, as was later done by means of the semi-automatic MFCs. Because the district court found — contrary to Hewitt’s testimony — that defendants had not disclosed their intent to use semi-automatics, we must assume that it did not credit that testimony, although nowhere in its findings did the court mention and reject the testimony. We do not, in any case, rely upon Hewitt’s testimony in determining that the court’s fraud findings premised on the term “facultative” were erroneous. Bailey himself did not attend the trial but was deposed and his deposition was read. He did not describe a meeting with Hewitt in 1979 nor any specific representations having been made to him prior to the execution of the Treaties on the character of the faculta-tive underwriting. He testified generally to “understanding” that Graham Watson would assess and underwrite each risk separately, but did not refer to any conversation or occasion where any defendant so promised. Asked if he would have considered semiautomatic binding authorities to be faculta-tive underwriting, he said that “is not what I had intended and not what I had been told from my own recollection.” This was the closest he came to suggesting that he was told by someone (defendants, brokers, or others?) that facultative meant what the judge found. We think this vague testimony, which makes no reference to specific sources, falls short of supporting a finding that defendants expressly promised to engage in risk-by-risk underwriting only or knew that plaintiffs misunderstood their intentions in this regard. 2. The Intended Meaning of the Term “Facultative” We similarly hold clearly erroneous the finding that defendants “knew” that plaintiffs understood “facultative” to be limited to risk-by-risk certificate underwriting. There is no evidence of statements or correspondence by plaintiffs or their representatives to defendants, prior to execution of the slips and treaties, informing defendants that the plaintiffs understood the meaning of facultative to be so limited. Of course, if the court properly could have found, on the basis of the evidence, that the term “facultative” was unambiguous, referring only to individual certificate, risk-by-risk underwriting, then defendants would be charged with knowledge of that ordinary meaning. However, as the evidence clearly showed, that term, both standing alone and as used in the Placing Information, slips, and Treaty Wordings, encompasses a variety of underwriting methods, about the propriety of which the parties and their experts disagree. Whether or not a term as used by parties to a contract is ambiguous is a question of law subject to plenary review. ITT Corp. v. LTX Corp., 926 F.2d 1258, 1261 (1st Cir.1991) (citations omitted); see also In re Navigation Technology Corp., 880 F.2d 1491, 1495 (1st Cir.1989) (“Contractual language is considered ambiguous where the contracting parties reasonably differ as to its meaning.”). However, where a term is ambiguous, its meaning presents a question of fact, see Commercial Union Ins. Co. v. Boston Edison Co., 412 Mass. 545, 557, 591 N.E.2d 165, 172 (1992) (citations omitted), a finding on which may only be reversed if clearly erroneous. Fed.R.Civ.P. 52(a). As noted, the district court found that the parties understood the meaning of the term “facultative” in its “standard and traditional sense, namely, underwriting on a risk-by-risk certificate basis.” If by this finding, and others like it, the district court meant that the term was legally unambiguous, being limited in meaning to only that one type of underwriting, it was wrong as a matter of law. Expert testimony and treatises presented by both sides support the view that the term, as used in the industry today, has been broadened beyond its classic roots, notwithstanding plaintiffs’ insistence that the classic method is alone the proper one. Most likely the court did not mean the term was unambiguous as a matter of law, but rather concluded, on the basis of all the evidence, that, as used in the present circumstances, it should be given the limited meaning ascribed. Yet the district court offered no reasons why it gave the term the limited reading it did. Nor did it explain why it believed defendants “knew” that plaintiffs so restricted the term. On the latter point, it may have been influenced by testimony from defendants’ principal, Graves Hewitt, who said he had as good an understanding of the London insurance market as any American. The judge may have felt that, possessing such insight, Hewitt “knew” that, as some English witnesses testified, “facultative” would be understood to mean risk-by-risk certificate underwriting in that market. But absent evidence that Hewitt actually knew and believed this, such a leap would be pure speculation given Hewitt’s own contrary testimony. Moreover, English treatises introduced into evidence by plaintiffs indicate that, notwithstanding plaintiffs’ witnesses, the reinsurance industry in England recognizes types of facultative reinsurance other than the risk-by-risk certificate variety. A leading English writer on reinsurance, Golding, describes in his authoritative treatise (introduced by plaintiffs) various types of faculta-tive reinsurance other than the risk-by-risk certificate variety. One variation Golding describes is the so-called “cover in course of post.” He states: It will be clear that much of the labour involved in the facultative method is connected with getting the necessary initials on the slips. In modern practice this can be largely avoided by the system of what may be called giving cover “in course of post” — though the term nowadays extends to the use of telex communications as much as to the mail. The reinsured will arrange facilities with a number of reinsurers, whereby it may issue request notes by post, for one or more lines of a risk to be reinsured, as may be agreed. The reinsur-ers will then hold covered each up to the amount of its agreed share and remains so bound, unless and until it has signified its declinature “in course of post”. As a rule a limit is fixed, within which this must be notified say 48 hours after receipt, though sometimes this is extended up to as much as a fortnight to allow for possible delays in transmission. If no declinature is made within the period, the reinsurer is bound in the ordinary way. The system does save a great deal of work, and is much favored by reinsureds accordingly. Yet it may be emphasized that it still remains facultative reinsurance, for the reinsurer is in no way deprived of its power to decline, even though it must accept responsibility in the meantime. C.E. Golding, Golding: The Law and Practice of Reinsurance 42 (K.V. Louw ed., 5th ed. 1987) (emphasis supplied); see also R.L. Carter, Reinsurance 234-35 (2nd ed. 1983) (detailing use of bordereau to report risks bound under the “cover in course of post” method, which he also classifies as faculta-tive). But even ignoring these indications that English custom and practice have gone beyond classic facultative methodology, it is the American, not the English, usage that seems to us key. The underwriters in London and Europe contracted in the slips with defendant NERCO, an American company, for reinsurance “classified by the Reassured [NERCO] as Property and Casualty Faculta-tive Assumed Business produced and underwritten by the Graham Watson division of Cameron & Colby, Inc.” (Emphasis supplied.) The facultative reinsurance NER-CO was to classify covered risks in the American, not the English, market. NER-CO was expressly delegated the right to “classify” the reinsurance. See supra note 20. In exercising that right, NERCO was, of course, held to a standard of reasonable classification. Salem Glass Co. v. Joseph Rugo, Inc., 343 Mass. 103, 106, 176 N.E.2d 30, 32-33 (1961) (where a contract leaves a certain discretion or power in the hands of one party, that party is under a duty to exercise that power reasonably); accord Johnson v. Educational Testing Serv., 754 F.2d 20, 26 (1st Cir.1985), cert. denied, 472 U.S. 1029, 105 S.Ct. 3504, 87 L.Ed.2d 635 (1985). Nonetheless, being an American company operating here, NERCO would obviously be expected to classify its business pursuant to American, not English, terminology. Hence, to the extent there is any difference between the prevailing English and American views of what kind of underwriting the market regards as “facultative,” the parties would have intended the American interpretation to control, absent evidence of some contrary intent. Cf. Hazard’s Adm’r. v. New England Marine Ins. Co., 33 U.S. 557, 564, 8 L.Ed. 1043 (1834) (“Underwriters are presumed to know the usages and customs of all of the places from or to which they make insurances.”). To be sure, plaintiffs’ experts gave testimony tending to show that the American market understood the term “facultative reinsurance” to mean risk-by-risk certificate underwriting. One might argue that the district judge was entitled to believe plaintiffs’ experts over defendants’ (who testified to the opposite understanding), and to infer that the ordinary meaning of the term “faculta-tive” was, therefore, the traditional one of risk-by-risk certificate underwriting. But the evidence that the term “faculta-tive,” within the American market, embraces more than just the individual risk certificate method is simply too extensive for the court to have rejected. Normally, of course, we are bound by the district court’s choice among competing experts. But it is hard to gainsay experts such as defendants’ expert James Inzerillo, see supra note 21, when even plaintiffs’ experts did not categorically deny the widespread use, within the faculta-tive operations of American reinsurers, of facilities like the semi-automatics and automatics here in issue. Plaintiffs’ experts did not, in faet, testify that the ordinary meaning of the term in the Americ