AXA Versicherung AG v Arab Insurance Group (B.S.C.)

The principal issue in this action is whether the claimant reinsurer (“Axa”) is entitled to avoid two reinsurance treaties entered into with the defendant reinsured (“Arig”) and to recover in consequence the net sum of about US $5.15 million paid to Arig under those two treaties.

The first treaty was entered into in 1996 by Axa’s predecessor in title, Albingia Versicherungs-AG (“Albingia”). It was a “first loss treaty” covering the first US $500,000 of losses for any one accident or occurrence on Arig’s book of inwards marine energy construction risks attaching during the period 1 January 1996 to 30 June 1997. There was no aggregate limit. It was “facultative/obligatory” in nature, meaning that Arig could choose which risks to cede to the treaties. Albingia’s line was 50%, the remaining 50% being written by another reinsurer, Rhine Re.

Axa seeks to avoid this treaty for non-disclosure of loss statistics relating to Arig’s existing book of inwards marine energy construction risks written during the period 1989 to 1995, alternatively for misrepresentation, the representation alleged being to the effect that there were no such statistics. It contends that disclosure of these statistics would have revealed a dreadful picture of past losses incurred by Arig in writing such risks and that, if these had been disclosed, Albingia would not have written the treaty.

The second treaty was the 1997 renewal of the 1996 treaty. It covered risks attaching during the period from 1 July 1997 to 30 June 1998. Axa seeks to avoid this renewal treaty on the same grounds and, in addition, for non-disclosure of three incidents which either had resulted or were likely to result in claims against Arig under the 1996 treaty. It contends that one of these incidents, a claim or potential claim by Clyde Petroleum, was material in its own right and therefore ought to have been disclosed, as was the cumulative effect of all three incidents together. Axa contends further that by reason of the Clyde Petroleum incident a representation by Arig to the effect that there had so far been only one claim under the 1996 treaty (as it happens a separate claim by Clyde Petroleum) was a misrepresentation which also entitles it to avoid the 1997 treaty.

Arig accepts that past loss statistics relating to insurance written by a proposed reinsured will generally, or at any rate may, be material as affecting the judgement of a prudent reinsurer deciding whether to accept the offered risk or upon what terms to do so and must therefore be disclosed. That this will generally be the case is well established as illustrated, for example, by the statement in Arnould’s Law of Marine Insurance & Average, (18^th Edition 2013), para 16.160:

“This issue is always a question of fact in all the circumstances and the materiality or otherwise of previous casualties will often be a function of the type of cover under consideration. Thus, previous casualties or other aspects of the assured's loss experience may be material for disclosure. In the case of reinsurance, the loss history will invariably be material.”

Carter on Reinsurance (5^th Edition, 2013) page 197 is to the same effect:

“Details of the loss experience of the account to be reinsured have always been required by reinsurers when negotiating the terms for a new treaty...”

However, Arig’s case is that in the particular circumstances of this case disclosure of its historic loss statistics would not have influenced the judgement of a prudent underwriter considering whether to write the 1996 or 1997 treaties. This is for two main reasons. The first is that energy construction risks are each unique, so that little or nothing is to be gained from considering past results achieved by the reinsured on the insurance of such risks. Energy construction reinsurance is therefore said to be an exception to the general position stated above. The second is that in mid-1991 there was a change of underwriter at Arig and that the new underwriter, Mr Lars Hylander, introduced a radical change of underwriting policy so that henceforth a much more rigorous approach to the selection of risks was adopted. Therefore, it is said, disclosure of loss statistics which in many cases related to risks written by a different underwriter, in different market conditions and pursuant to a different underwriting strategy, was unnecessary. Arig denies that any representation was made as to the non-existence of such statistics. It contends further that Axa has failed to prove that any non-disclosure or misrepresentation concerning these matters induced Albingia’s underwriter to write the treaties in question, contending that Albingia would have been written them on the same terms even if there had been a fair presentation of the full facts which as well as past loss statistics would have included Arig’s change in underwriter and underwriting policy.

As to the 1997 renewal, Arig contends that none of the three incidents complained about on renewal was material to Albingia’s 1997 decision to renew the risk which it had written and, once again, that Albingia has failed to prove its case on inducement.

Further, Arig contends that even if it would at one time have been possible for Albingia (or Axa as its successor) to avoid the treaties, it is now far too late to do so almost 20 years after the event.

There is in addition a separate dispute, which only arises if Axa’s avoidance case fails. It concerns one of the individual cessions to the 1996 treaty. Under the heading “Information N.L.O.W.”(which stands for “No Limitation Or Warranty”) in the 1996 treaty it was provided that “Arig will not cede any layered contracts which represent less than 20% of the full contract value.” The dispute is whether Arig did cede a layered contract which represented less than this 20% figure; if so, whether the provision in question was a contractual restriction on the risks which Arig was entitled to cede as distinct from a mere statement of intent; and if the former, whether Axa’s claim to recover the payment made on this risk is now time barred.

Finally, Arig counterclaims for sums which it says are due to it under the treaties. Axa contends that, even if its avoidance case fails, the counterclaim is (partly) time barred.

As will be apparent from this brief summary, most of the events with which this action is concerned took place almost twenty years ago or in some cases even more. While that may represent no more than the twinkling of an eye to those engaged in long tail reinsurance business, the inevitable consequences are that the available documents, voluminous as they are, are incomplete and that the witnesses cannot be expected to have any (let alone any precise or detailed) recollection of most of the matters about which they gave evidence. To their credit the witnesses did not pretend that they had any such recollection, but acknowledged that much of their evidence consisted of assertion as to what they said or did based upon their belief that it was what they would have done in the circumstances in view of their normal working practices. I have therefore approached such evidence with caution in making the findings set out below, recognising the danger that it is affected by hindsight knowledge of how matters have turned out and on occasion by an element of wishful thinking.

The need for caution applies with even greater force to hypothetical evidence as to what a witness would have done if circumstances had been different – in particular, on the issue of inducement, as to whether or not a risk would have been accepted if matters which were not in fact disclosed had been disclosed. Whether a reinsurance underwriter will be willing to write a risk or on what terms will depend on many considerations – not merely the bare facts relating to the risk itself, but such matters as the terms in which it is presented by the broker, the underwriter’s view of the reinsured company and its underwriter, the reinsurer’s book as a whole and the way in which the particular risk will fit into his overall strategy, his commercial relationships with the reinsured and the broker, and prevailing and anticipated market conditions. It will very often be difficult for a witness to think himself back into all of the circumstances as they existed at the time which would have had a bearing on his decision whether to write a particular risk in circumstances which are inevitably hypothetical since ex hypothesi there was no fair presentation of the risk. In a case such as the present which is litigated so long after the events in question that the witness has no recollection at all of the actual transaction, this difficulty is particularly acute.

Despite the usual limitations of witness recollection, exacerbated in this case by the length of time which has passed, I consider that all of the witnesses were doing their best to assist. As usual, however, where documents are available they represent much the best evidence not only of what the parties did, but also of what they were thinking at the time.

Axa’s principal witness of fact was Mr Thomas Holzapfel, who was Albingia’s head of inwards treaty business from 1985 (and later head of outwards treaty business as well) until he left the company in 2000. He has given evidence in several cases in which Axa has sought to avoid insurances written by Albingia (see e.g. Limit No.2 v AXA Versicherung AG [2008] EWHC 2321 (Comm), [2008] 1 Lloyd’s Rep 330). He retired in 2011. Although almost all of his evidence was of the nature indicated above as distinct from actual recollection of events, he was in my judgement a straightforward and fair witness. There is an issue whether Mr Holzapfel was the individual at Albingia who made the relevant underwriting decisions. That arises because, as it happened, he was out of the office for much of the time when those decisions were made and most of the relevant correspondence was between the brokers, Newman Martin & Buchan Ltd (“NMB”), and Mr Holzapfel’s assistant, Ms Sylvia Jerabek (now Rauser-Dittman). Ms Jerabek was not called as a witness although she is still employed by Axa. That was apparently because she declined to have any involvement with the case, partly because of her personal family circumstances and partly because the events concerned took place so long ago. Arig’s case is that it was Ms Jerabek and not Mr Holzapfel who made the relevant decisions and that in the absence of any evidence from her Axa has failed to prove its case on inducement.

Also a generally fair and straightforward witness was Mr Lars Hylander, Arig’s principal factual witness and the head of Arig’s energy underwriting department from 6 June 1991 until he left the company in mid-1999. Previously he had been with Skandia. He was an experienced energy underwriter, well regarded in the market, who had more experience of writing energy and specifically energy construction risks than either of the experts. He was understandably bemused to be asked questions about what he had meant by the precise wording of documents written many years before. Whether or to what extent Mr Hylander introduced a new and more conservative underwriting policy is one of the issues in the case.

Arig relied also on the evidence of Mr Robin Stow, the broker at NMB who was principally concerned with placing the reinsurances in issue in this case. He left NMB in September 2011. Although he too was a helpful witness, my impression was that, at least in the circumstances of these treaties, he was essentially reactive in passing on information provided to him as distinct from applying his mind to what might need to be provided to reinsurers. His evidence, for example, was that whether to provide past loss statistics would depend on whether the proposed reinsured thought they were relevant. I found that surprising. I would expect an experienced broker to have a view of his own about such a matter.

Both parties called expert evidence on the issue of materiality. Axa’s expert was Mr Stephen Hartigan, who had some 35 years experience of the London insurance and reinsurance market between 1971 and 2004, while Arig’s expert was Mr Richard Outhwaite, an underwriter in the London market from the 1960s until about 1994.

Mr Hartigan acknowledged that he had never written a direct energy risk and therefore had no personal experience as an underwriter in writing energy construction risks. Nor did he have experience in any capacity of the placing of a first loss facility covering such risks or of the specific reinsurance of energy risks. He did, however, have extensive experience of the reinsurance of risks more generally, including policies or treaties which included but were not limited to energy risks. Mr Simon Bryan QC and Mr Guy Blackwood QC for Arig submitted that Mr Hartigan did not have the expertise that would qualify him to give expert evidence on the underwriting of a first loss reinsurance of energy construction risks and associated questions of materiality, and that the evidence which he gave on the issue of materiality was therefore inadmissible in its entirety. I do not accept this. I consider that Mr Hartigan’s evidence was relevant, albeit that it was not the evidence of a specialist in energy (let alone energy construction) risks, a limitation which I have borne in mind in assessing his views.

Mr Outhwaite was a marine underwriter for many years. He did write energy risks, although these were only ever a fairly small part of his overall book. These may have included some energy construction risks, although he could not say whether he had written any such risks during the late 1980s or 1990s. If he did, they were very few in number. He also wrote excess of loss reinsurance covering energy accounts, although (in common with others in the market) without specifically distinguishing between operating and construction risks. He never wrote a first loss treaty, either of energy risks or any other kind of risk. Indeed, in response to one question as to how a prudent underwriter writing a first loss treaty would react to figures presented to him, Mr Outhwaite expressed the trenchant view that a prudent underwriter would not write a first loss treaty in the first place. This case might suggest the wisdom of that view. Nevertheless, it appears that such treaties were written, albeit comparatively rarely, and were written by underwriters whom it would be unfair to castigate as imprudent.

Thus while it is fair to say that Mr Outhwaite had more experience of energy insurance than Mr Hartigan, neither expert can be regarded as a specialist specifically in the direct insurance or reinsurance of energy construction risks or in first loss reinsurance at the relevant time. Energy construction insurance was, as it appears, a rather specialist field. Nevertheless both experts were highly experienced in the insurance and reinsurance market generally and their evidence was helpful.

The purpose of a first loss reinsurance facility of energy construction risks is to indemnify the reinsured up to the specified limit in respect of claims on the reinsured’s direct energy construction book. It is termed a first loss facility because it will indemnify the reinsured up to the specified limit for the entirety of the loss from the ground up. Thus, if there is no deductible in the direct insurance, every dollar of loss sustained by the original insured up to the limit of the first loss reinsurance will be passed on to the reinsurer. If there is a deductible in the direct insurance, all such losses in excess of the deductible will be passed on to the reinsurer up to the limit of the first loss reinsurance. The extent to which claims will be made on the reinsurance will therefore depend, very largely, on the composition of the energy construction risks written by the reinsured and the terms on which each original risk was written by the reinsured.

First loss reinsurance therefore has in common with quota share reinsurance that the reinsurer’s fortunes are closely tied to the original book of risks written by the reinsured. The principal difference, however, is that whereas the quota share reinsurer is liable to the extent of his line, typically with a retention by the reinsured (so that a quota share “is in the nature of a joint venture between the reinsured and his reinsurer” involving “a sharing of risks (premium and losses) between reinsured and reinsurer”: Bonner v Cox [2005] EWCA Civ 1512, [2006] 1 CLC 126 at [87]), the first loss reinsurer is liable for 100% of the losses up to the specified limit. Moreover, in contrast with excess of loss reinsurance where the reinsurer is only liable for losses in excess of whatever figure is agreed, the effect of first loss reinsurance is to pass on to the reinsurer each and every minor loss for which the reinsured is liable.

A further difference between quota share treaties and a facultative/obligatory treaty such as the first loss treaties in this case was explained by Lord Millett in Aneco Reinsurance Underwriting Ltd v Johnson & Higgins Ltd [2001] UKHL 51, [2002] 2 Lloyd’s Rep IR 91 at page 105:

"Under a quota share treaty, the insurer is obliged to cede to the treaty a fixed proportion of every risk which falls within the limits of the treaty. Under a fac/oblig treaty the insurer has a choice whether to cede any given risk to the treaty. He cannot cede it unless it falls within the limits of the treaty, but he is not obliged to cede it if it does. The reinsurer has no choice; he cannot insist on a risk being ceded, and cannot refuse to accept his share of a ceded risk … Fac/oblig treaties are naturally less attractive to reinsurers than quota share treaties. They are subject to the obvious risk that the insurer will retain good business for his own account and cede poor business to the treaty. There is, or at least is assumed to be, no obligation of good faith on the part of the ceding party when exercising his discretion whether to cede or retain a risk. The only constraint upon him is that he must exercise some restraint if he wishes to maintain a good reputation in the market and any hope of doing future business with existing and prospective reinsurers."

Accordingly the first loss reinsurer is heavily dependent on the success of the reinsured’s underwriting and is in particular vulnerable to a large number of low level claims.

Moreover, the availability of first loss cover may itself have an impact on the risks which the reinsured is prepared to write. For example, an insurer who would only be prepared to write a risk with a deductible of (say) US $200,000 without reinsurance might be willing to write it with a deductible of (say) US $20,000 with the benefit of first loss reinsurance. The effect would be that any losses between US $20,000 and US $200,00 (which ex hypothesi the insurer was not prepared to bear) would be passed on to his reinsurer. There is nothing wrong with this approach. As Bonner v Cox makes clear, there is no duty on a reinsured to act prudently in the reinsurer’s interest. If the reinsurer is prepared to write such business, the reinsured is under no legal duty not to take advantage of whatever terms the reinsurer is prepared to offer, provided always that he makes a fair presentation of the risk (see [105] to [111] of the judgment of Waller LJ in that case). Those engaging in first loss reinsurance must be taken to understand and accept this.

This is particularly relevant to first loss reinsurance of energy construction risks. The offshore energy business was by its nature a hazardous business and within that business construction was particularly hazardous. Often it involved high risk projects undertaken in adverse environmental conditions using cutting edge technology. It was a business in which claims, including low level attritional claims, the number of which could easily mount up, were to be expected.

In 1996 Albingia was a medium sized German insurance company based in Hamburg. It was part of the Guardian Royal Exchange group. Axa purchased this group in 1998 and so acquired Albingia. Arig was a well regarded insurance and reinsurance company domiciled in Bahrain, substantially backed by sovereign wealth funds from the Middle East region.

Between 1989 and 1993 Arig insured a substantial number of energy construction risks. These comprised, to adopt Arig’s own classification, rig building, pipeline construction, platform construction and construction liability risks. In 1989 some 40 such risks were written. In 1990 the figure was 39. Thereafter the numbers fell, although in 1993 Arig still wrote 19 such risks. In 1994 and 1995 Arig wrote only one and four such risks respectively.

Arig’s energy underwriting department during the period from 1989 to 1997 consisted of a team of five individuals all of whom had authority to write risks without seeking approval although in practice they would consult the head of the department if it was practicable to do so. Until he left Arig on 1 January 1991 that was Mr Bjorn Simenstand. After a gap of six months his replacement from 6 June 1991 was Mr Hylander. Not surprisingly it took Mr Hylander a little while to become familiar with the department and to exercise a degree of control over the way in which it wrote risks. His own assessment, which I accept, was that this familiarisation process took several months and that it was from about October 1991 that his influence on the department began to be properly felt. Mr Hylander reported to Mr Khwaja who was Arig’s Acting General Manager.

In the event (and ignoring any reinsurance recoveries) the underwriting results on the energy construction risks written by Arig during the 1989 to 1992 period were extremely poor. This was a period, speaking in general terms, of low premium rates and deductibles for the market generally. The 1989 and 1990 underwriting years were particularly bad years for energy insurers. Arig was not alone among energy construction insurers in experiencing a high level of claims and losses, even leaving to one side a number of high profile catastrophic losses. In a colourful phrase, Mr Outhwaite observed that during these years an energy construction underwriter who only wrote a small book of business “might just run between the raindrops, so to speak”, and thus avoid claims, but thought that in practice this was unlikely and that nobody writing such risks in those years would have made a profit.

Arig’s loss position on its inwards book as it had developed by the end of 1995 is set out in the following table:

These are the statistics which Axa says were required to be disclosed to Albingia before the conclusion of the 1996 treaty.

During 1996 the loss ratios for these underwriting years deteriorated further, as more claims came in.

To some extent these losses would have been mitigated by the reinsurance available to Arig. Between 1 October 1989 and 30 September 1992 Arig had the benefit of a first loss cover insuring construction risks to which it was therefore able to cede many of the inwards risks written during this period. It is likely that some of the risks written by Arig during the period from 1 October 1989 to 30 September 1992 would not have been written if that reinsurance had not been in place.

As indicated, during 1994 and 1995 Arig wrote very few energy construction risks, but by 1995 it was keen to re-enter this market. It sought to reduce the exposure which it planned to take on by obtaining first loss reinsurance. However, such reinsurance was difficult to obtain and Arig’s initial attempts to do so were unsuccessful. They appear to have begun in about February 1995 when Mr Masoud Bader, an underwriter in Mr Hylander’s department, indicated the broad scope of the cover which Arig was seeking in a fax to Mr Tony Rowe, a director of the reinsurance broker Swire Fraser. Arig was seeking a 12-month cover on a risks attaching basis for “primary/ground-up” offshore construction risks, with a limit of US $5 million. This limit was expressed to be the limit for 100% of the risk, though it was contemplated that Arig’s line would generally be for a lesser proportion and could be limited to a maximum of US $2 million. This could no doubt have been subject to negotiation if matters had proceeded further. Mr Bader indicated that the estimated annual premium income to Arig would be about US $500,000 to US $750,000.

There is no evidence of what, if anything, Swire Fraser did in response to this request until late October 1995 when it was discussed by Mr Bader and Mr Rowe at a meeting in London of which no record survives. Mr Bader gave evidence, but had no recollection of this meeting. However, on 31 October 1995 Mr Rowe wrote to Mr Bader identifying the information which Swire Fraser would require in order to market such reinsurance cover. The list included “triangulated figures on the overall account”, which is another way of referring to loss statistics on Arig’s existing book of inwards energy construction risks.

Mr Bader obtained the relevant underwriting statistics from Arig’s database and formulated the information which Swire Fraser had requested, sending it to Mr Rowe on 16 and 17 November 1995. This included a table of underwriting statistics for the underwriting years 1989 to 1993 showing the development of premium and losses at each year end and cumulatively at the end of October 1995. Mr Bader’s evidence was that Arig provided figures going back to 1989 and not merely from Mr Hylander’s appointment because it “wanted to show the full picture”.

The cumulative figures, all in US dollars, were as follows:

Although I will not set out all the figures, the figures presented showed that, year by year, the losses incurred on risks written well in the past were continuing to increase. For example, the total of paid claims and reported losses on risks written in 1989 had stood at US $6,175,121 at the end of 1994, but by the end of 1995 this figure had increased to US $7,918,638, the figure shown in the table above. Nor was there any reason to suppose that this process would not continue during 1996 (as in fact it did). The same was true for risks written during the 1990 and 1991 underwriting years.

As well as providing the actual figures, Mr Bader illustrated a variety of other ways in which they could also be presented in order to show the figures in a better light, referred to generally in evidence as “as if” (or less kindly by Mr Hartigan as “if only”) statistics. These included (a) omitting losses where Arig’s share of the loss was less than US $50,000, (b) applying higher deductibles which Arig regarded as more in line with current market conditions than the low deductibles which had been agreed in the earlier period, (c) applying an additional fixed Arig retention of US $1 million instead of a deductible, and (d) doubling the premium income achieved on risks written between 1989 and 1991 which Arig also regarded as more representative of the premium which could be achieved in 1995 market conditions. There was nothing wrong with these ways of presenting the figures alongside the actual losses so long as it was made clear, as it was, what each set of figures represented. On the contrary, to do so constituted a normal, legitimate and reasonable way of attempting to persuade prospective reinsurers to view the apparently disastrous underwriting results as unrepresentative of the outcome which might be expected on the new risks for which Arig was now seeking reinsurance. It also gave prospective reinsurers the opportunity to analyse the figures and, if they wished, to propose alternative terms (for example, a higher deductible or fixed retention) on which they would be prepared to offer the reinsurance which Arig was seeking.

I would at this stage make six observations about the information provided to Swire Fraser. First, the information was readily available to Arig. Second, Arig (at least in the person of Mr Bader and Mr Hylander who also saw and approved what was sent) was well aware of the extremely poor outcome of the risks written in (at any rate) 1989 to 1991. Third, there was no suggestion that any of this information (including that for the earlier years) did not need to be disclosed to prospective reinsurers, either because of the unique nature of each energy construction risk, or because a different underwriter pursuing a different underwriting policy was now in charge at Arig, or for any other reason. On the contrary it appears to have been regarded as obvious that this was information which prospective reinsurers being asked to write a fairly unusual form of reinsurance would wish to have. Fourth, it is clear that Mr Bader made considerable efforts to devise ways in which this information could be presented to prospective reinsurers in the most favourable light. Fifth, the positive points made by Arig with a view to persuading prospective reinsurers that these historic results did not reflect the outcome which might be expected on the new risks for which Arig was seeking reinsurance had nothing to do with the fact that there was a new underwriter now in place and that a radically different underwriting strategy was now being followed. Instead Arig focused on the higher deductibles and higher premiums which it suggested could be achieved in 1995 market conditions. Finally, the reinsurance terms which Arig was seeking to obtain through Swire Fraser were broadly similar to those which it was subsequently to seek through NMB, albeit that the Swire Fraser terms envisaged a higher limit.

Mr Rowe’s response, on 28 November 1995, was that he had spoken to a few potential underwriters and that the one showing the most interest had asked for more particulars of the rate rises for construction risks over recent years and more information about the deductibles which Arig had identified. Mr Bader appears to have had difficulty in providing convincing evidence of these improved market conditions, but provided such information as he could. It appears, however, that by about the end of January 1996 Arig’s efforts to obtain first loss reinsurance through Swire Fraser had petered out. Since this was reinsurance which Arig was at all times keen to obtain, and since as a broker Swire Fraser would have done its best to obtain such cover for Arig, the conclusion must be that no reinsurer which it approached was prepared to quote for the business. It is not possible on the evidence to say whether this was solely because of the poor loss statistics (and because reinsurers were unimpressed with Arig’s attempts to present them in a more favourable light) or even to say with any real confidence whether this was a significant factor in Swire Fraser’s failure to obtain a single quotation, although if it was a factor, that would not be surprising. It appears that Swire Fraser did not approach Albingia. At any rate, there is no record of any such approach.

From at least 1990 Arig had taken out quota share reinsurance, protecting its energy account as a whole. In November 1995 it sought to renew this quota share energy treaty in the market, using as broker NMB. For this purpose Mr Hylander sent a fax to Mr Chris Martin at NMB on 16 November 1995 setting out what Arig was seeking and including a risk profile by type and location of risk for the 1995 account which indicated that a total of 81 risks had been written by Arig of which only two were construction risks. A table of losses going back to 1988 indicated that it excluded certain types of risk which were no longer written, one of which was “concrete construction risks”.

Albingia had not previously participated in the Arig quota share treaty but on 27 November 1995, NMB sent a fax to Mr Holzapfel of Albingia offering him the opportunity to do so for the next policy year. The fax attached a draft slip and some (but not all) of the information which had been provided to NMB. It did include Arig’s risk profile breakdown. Thus information was at least available to Mr Holzapfel that Arig had written two energy construction risks in 1995.

At that time, Mr Holzapfel knew Arig to have a well-established base in the Middle East, with good access to business in that region and to be supported by sovereign wealth funds in the region. His initial response to NMB’s offer professed a lack of enthusiasm, although in general he was keen to participate in quota share reinsurance with high-quality reinsureds such as Arig. However, after further discussions with NMB which resulted in improved terms for excess of loss cover for Albingia’s participation and reinstatement premium protection, Mr Holzapfel eventually indicated Albingia’s willingness to write a 35% line on the Arig treaty for the 1996 year. This was a substantial line which indicated a significant degree of confidence in Arig by Mr Holzapfel, based on the information provided. Albingia was in effect offering to share Arig’s fortunes on its energy underwriting during the relevant period. In fact, the risk proved sufficiently popular with reinsurers that Albingia’s proposed 35% line was signed down to 19.57%.

As the premiums available for insuring energy operating risks declined and marine energy insurers turned their attention back to construction business in the mid 1990s, it was not only Arig that sought the protection of a first loss energy construction treaty. Mr Holzapfel at Albingia was introduced to this concept by NMB, acting for other principals, in October and November 1995.

Four such treaties were offered to Mr Holzapfel by NMB in addition to the Arig treaty. The first, for AIG, was discussed between late October 1995 and January 1996. After (among other things) a study of the loss records for construction risks written from 1991 to 1994 by AIG’s London office which showed limited losses incurred so far, with a loss ratio for 1991, the most mature year, of only 33.2%, Albingia confirmed a 50% line (“the AIG Oil Treaty”). AIG was rightly regarded by Mr Holzapfel as a “blue chip” company and a very attractive reinsured. In this respect Arig, although also a well regarded company and an attractive reinsured, was not quite in the same league.

The second such treaty was for the Copping syndicate at Lloyd’s. NMB first approached Albingia about this on 4 July 1996, shortly before it was to approach it in connection with the proposed Arig treaty. Extensive information was provided, including detailed loss records for construction risks written from 1991 to 1996 as well as a list of current construction exposures and a construction risk history. These records revealed that the 1991 and 1992 years had both been loss making, but it was explained that this was the consequence of two substantial losses in 1991 and one large loss in 1992. If those losses were taken out of the reckoning, the loss ratios on both years were around 30%. On 11 July 1996 Albingia agreed to take a line of 50%. Eventually Axa was held to be entitled to avoid this treaty for misrepresentation, the representation being that the syndicate’s policy was not normally to write construction risks unless the original deductible was at least £500,000 and preferably £1 million: see Limit No.2 v AXA Versicherung AG [2008] EWHC 2321 (Comm), [2008] 1 Lloyd’s Rep 330. It appears that Mr Holzapfel’s evidence in that action was that he would not have written the reinsurance if this representation had not been made. The judgment of the deputy judge (Mr Jonathan Hirst QC) records Mr Holzapfel’s evidence on this point at [54] as being that:

"I mean, the question of a high deductible of high involvement of a client in the first place was of absolute importance and distance ourselves from attritional losses and that is why the 500,000 or a million pounds was simply of utmost importance" … "the 500,000 threshold was an absolute essential feature of writing the business and distancing ourselves from attritional claims."

Subsequently, and after the conclusion of the 1996 first loss treaty with Arig, Albingia agreed a 100% line on an energy construction treaty with the Atkin syndicate at Lloyd’s. Statistics were presented showing every risk written for the previous four years, since the syndicate started, showing the original line, excess point and loss record and demonstrating what the result would have been if the premiums and losses achieved were allocated to the prospective treaty.

Finally, on 22 November 1996, Albingia was offered a further treaty with AIG. On this occasion no historic loss statistics were provided, but the risk was presented by NMB on the basis that Albingia already had a very good relationship with AIG, a “blue chip” reassured, and that the proposed treaty was very similar to the AIG Oil Treaty which Albingia already had with AIG. Mr Holzapfel agreed to take a 20% line, subject to suitable excess of loss protection. He did so on this occasion without asking to see historic loss statistics, probably because of the existing relationship with AIG.

Having been unsuccessful in placing a first loss energy construction treaty through Swire Fraser, Arig approached NMB in April 1996. Mr Bader and Mr Stow spoke on or about 24 April 1996, and on several occasions thereafter. There is no contemporaneous note of what was said, but both Mr Bader and Mr Stow asserted in their witness statements that they discussed Arig’s loss records, and that it was decided that the reinsurance should be broked without reference to or disclosure of the historic loss statistics on Arig’s inwards book of construction risks.

Mr Bader said in his witness statement that he believed that he had explained to Mr Stow that Arig's approach to underwriting had changed considerably, and that the energy construction loss information which existed would be unhelpful or misleading for a potential reinsurer because Arig’s past underwriting strategy did not reflect the way that it approached such underwriting in 1996. However, in cross examination Mr Bader made clear that he had no recollection of any such conversation with Mr Stow and that the reason for his belief that they would have discussed the past loss records was that this subject must have come up in the conversation because such records were “a part of underwriting, any underwriting discussion, you look at the information, loss records is part of the information” and that there would inevitably be some discussion of the historic loss records. His evidence in cross examination about whether he had discussed Arig’s change in underwriting strategy with Mr Stow was rather vague. He had no recollection of any such conversation and at times his evidence appeared to be rather to the effect that everyone knew about this because of Mr Hylander’s reputation in the market.

Mr Stow said in his witness statement that he recalled discussing during his early conversations with Mr Bader what loss records Arig had for construction risks, and being told that Arig's underwriting approach had changed considerably because, among other things, Mr Hylander had a more conservative approach to underwriting than those who had underwritten Arig's energy account before him, and that therefore Arig's historic loss records would not reflect its underwriting philosophy in 1996 or its future intentions. However, Mr Stow also acknowledged in cross examination that he did not recall any such conversation with Mr Bader, although he said that he had an impression that such a conversation must have taken place because it would form part of a wide-ranging discussion about all the information which the broker or the reinsurer would need to know. He did not recall being instructed to tell reinsurers about Arig’s change of strategy and was unable to say that he or anyone had explained this change of strategy to Albingia.

On this latter point of whether the change in strategy had been discussed with Albingia, what Mr Stow’s evidence came to was essentially that he would have expected Mr Holzapfel to want to see past loss statistics relating to Arig's energy construction account, that Mr Holzapfel was not provided with these and apparently did not ask for them, and that Mr Stow believed that therefore something must have happened to put him off from doing so. Mr Stow could not positively say what might have put Mr Holzapfel off, but thought that one possibility was that he had been persuaded by some conversation that these statistics were not relevant, for example because he had been told about Arig’s change of strategy.

This evidence represents a very flimsy foundation on which to make a finding that any conversation between Mr Bader and Mr Stow along the lines described by them in their witness statements took place. I think it much more likely that it did not. There is no contemporary indication of such thinking and clearly it had not occurred to Mr Bader when seeking to obtain this reinsurance through Swire Fraser that Arig’s past loss records were irrelevant. On the contrary he had gone to considerable effort to present them in a favourable light. Moreover, as such loss records generally were regarded as important to reinsurers, and were regarded by both witnesses as an important matter (in Mr Bader’s words, “loss records is part of the information”, or as Mr Stow accepted, part of the information which the broker or the reinsurer would need to know), I would have expected a competent broker participating in a decision that there were valid reasons not to disclose them on this occasion to ensure that there was a clear record of that decision, the reasons for it and the reinsured’s agreement to that course. Moreover, I have no doubt that if Mr Holzapfel or Ms Jerabek had been told that loss records existed but were not being disclosed because of a change of underwriter or strategy by Arig, questions would have been asked which would have left at least some trace in the surviving written records. I therefore reject the evidence of Mr Stow and Mr Bader on this point. In particular, I find that nothing was said to Albingia about any change of strategy or to indicate that any past loss statistics existed.

It is possible that in view of the unsuccessful experience with Swire Fraser, Mr Bader was reluctant to disclose the loss records and was not confident in his or NMB’s ability to explain them away if they had to be disclosed, recognising as Arig did that regardless of disclosure of the records first loss reinsurance would not be easy to obtain. Or it may be that he took the view that it was unnecessary to disclose them unless they were requested (as they had been by Swire Fraser). Ultimately, however, even if Mr Bader did succeed in persuading himself that the loss records did not need to be disclosed, materiality is an objective question.

Following the telephone conference between Mr Bader and Mr Stow on or about 24 April 1996, NMB and Arig began to prepare the draft slip.

Arig’s motivation in seeking this first loss reinsurance was set out in an internal memorandum from Mr Hylander to Mr Khwaja dated 7 July 1996 in which Mr Hylander sought authority to conclude such a treaty. He wrote:

“First Loss Construction Treaty

…

The main reason for providing such facility is at first hand to provide a bottom line protection for construction risks where we are expecting to accept about 25-30 policies in 1996. The loss development on protected basis is expected to be poor. The main reason being insufficient deductibles as accepted by the market. This facility will however enable us to protect the exposed primary areas and thereby standing a good chance of making an overall profit.

…

The treaty is backdated to 1^st January 1996 and the main reason for this is that the exposures are expected to develop first in the second half of the year and continuously depending upon the length of the construction period. Consequently we do not expect any losses for the first half of the year but if the policies are not ceded to the treaty we will not be able to recover claims which we expect will happen at a later stage.

The number of policies written so far in 1996 is 15 with an overall premium to the treaty of about USD230,000.

I am convinced that the facility as proposed will provide an excellent bottom line protection on business we by experience know is highly exposed. When discussing the issue with other underwriters and brokers they all agree that we are very lucky if we will be able to get a facility like this. It is still very difficult to find treaty capacity for these exposures and the only alternative is facultative reinsurance which is both difficult to find and expensive to buy.”

This is an important document. Mr Hylander sought in his witness statement to explain some of the things which he said in this memorandum, but it was apparent from his cross examination that he was understandably unable to remember what he had meant by particular words and phrases. In my judgment the memorandum speaks clearly for itself. I would make the following observations about it.

First, it is evident that Arig expected to achieve a poor loss development on the risks which it had already written and expected to write in 1996, principally because it had been and would be unable to achieve deductibles as high as it would have wished (“The main reason being insufficient deductibles as accepted by the market”). In that respect the position was comparable with that which had prevailed in the 1989 to 1991 period, albeit that there was no evidence to show any quantitative comparison between deductibles available in the two periods. Whatever Mr Hylander’s desire to obtain higher deductibles as part of any more conservative underwriting policy, he recognised that if Arig wanted to write such energy construction business it would be subject to the constraints imposed by the market and as a result could expect to be hit by a fairly large number of relatively minor attritional losses which would have been caught by a higher deductible but, if the proposed reinsurance were effected, could be passed to reinsurers. The result of such reinsurance would be to give Arig “a good chance of making an overall profit” which, without the benefit of such reinsurance, it was much less likely to achieve.

Second, the memorandum referred to Arig’s knowledge based on its own experience that energy construction business was highly exposed (“business we by experience know is highly exposed”). Mr Hylander suggested that this was a reference to his own personal knowledge and experience of energy construction risks generally as distinct from Arig’s corporate experience of such business as a result of the risks which it had written from 1989 onwards. I do not accept this. The natural meaning of the memorandum is that this refers to Arig’s experience of such business as set out in its loss records.

Third, Mr Hylander plainly regarded that experience as highly relevant to the decision to seek first loss protection. Hence his reference to it in this memorandum to his superior. But as a result of the decision made by Mr Bader not to disclose those records, that experience was not to be shared with prospective reinsurers including Albingia.

Fourth, there was no suggestion by Mr Hylander that Arig’s past experience was now irrelevant because a different underwriting strategy was being pursued.

Fifth, the memorandum recognised that it would be very difficult to obtain the first loss protection which Arig was seeking (“they all agree that we are very lucky if we will be able to get a facility like this”).

Mr Khwaja approved the order for the first loss treaty on 11 July 1996.

NMB first approached Albingia in connection with this proposed business on or about 17 July 1996. By then NMB had been active on Arig’s behalf for several months and had already approached a number of reinsurers who were not interested. It is to be inferred that Albingia was not Arig’s first choice of reinsurer. On that date Martin Stephenson of NMB sent a fax to Mr Holzapfel and his assistant Ms Jerabek, although it may be that the fax was preceded by a telephone call. The fax covering sheet stated, under the heading “Arig (Bahrain) Construction Treaty”:

“We are very pleased to be able to offer you an opportunity to participate in the captioned new treaty.

This is a new Treaty for the Reassured and as such does not have a corresponding loss record.

Note that this is anticipated to be a small proportion of the Reassured’s overall Energy Account Income and will comprise of approximately 30 to 40 risks. 15 have already been written since 1.1.96 and ARIG would like to attach this contract at 1.1.96 to include them. Obviously this is subject to no known or reported losses.

We have also included a schedule of benefiting XL reinsurances which leave Treaty Reinsurers with a manageable retention – currently these are only desk quotes.

Trust you find all in order and look forward to your advice.”

The statement in the second paragraph of this fax that there was no “corresponding loss record” is the basis of Axa’s claim for misrepresentation.

The fax included a draft reinsurance slip including a rating scale, additional clauses applicable to the treaty, a rating scale graph, the schedule of proposed excess of loss reinsurances and a list of 15 energy risks already written and intended to be ceded to the proposed treaty, setting out the name of the project, inception, expiry, and Arig’s exposure and premium. The offered treaty was a reinsurance declaration treaty protecting Arig for "Construction, Installation, Commissioning and Maintenance, plus Operating Exposures as advised”. The draft slip did not expressly refer to energy construction but the covering fax and excess of loss order form made it clear that the proposal related to energy construction business written by Arig. It was proposed to cover losses occurring on risks attaching during the period of 18 months from 1 January 1996, thus having retroactive effect back to the start of the 1996 year but subject to the qualification that there were "no known or reported losses at date of order”. The draft provided for a maximum original policy period of “36 months, plus discovery/maintenance period”: in other words, the direct insurance was to cover projects with a construction period of three years, together with a further period after completion in which defects might be discovered. The limit of the treaty was “US $500,000 any one unit and/or item and/or structure or currency equivalent each and every loss, any one accident or occurrence” and the offered treaty was a first loss treaty, meaning that Albingia would pay the first US $500,000 of each and every covered loss suffered by Arig. No information was provided about the level of deductibles already written or expected to be written by Arig and thus the level at which losses would begin to be suffered by Arig. There was no disclosure of any loss statistics for energy risks previously written by Arig.

The fax was received and reviewed by Ms Jerabek. Mr Holzapfel was on holiday from 5 July and returned to the office on 29 July 1996. On 17 and 18 July he was in Copenhagen with his family. Although he had no recollection of this fax or of having discussed it with Ms Jerabek and there is no documentary evidence of any such discussion, Mr Holzapfel’s evidence based on their normal working pattern was that Ms Jerabek would have discussed its contents with him by telephone before responding and would only have responded to NMB with his approval. I accept that evidence and find on the balance of probability that there was at any rate some discussion in which Ms Jerabek advised Mr Holzapfel of the approach from NMB. It may be that she read out to him the content of the covering fax quoted above including the statement that there was no “corresponding loss record”, but it is not possible to make any positive finding that she did or otherwise to reach any conclusion as to the detail in which the fax was discussed or the extent to which its contents registered with Mr Holzapfel at that stage.

On 18 July 1996, Ms Jerabek confirmed to NMB that Albingia was prepared in principle to write a 50% line on the treaty provided that Arig’s ceding commission of 2.5% was removed, the rates offered for Albingia’s excess of loss protections were improved, and RPP cover was available at an affordable price. She did so, as I find, having first obtained Mr Holzapfel’s approval for this course in the telephone conversation between them which I have found to have occurred. This was only an agreement in principle, not a firm commitment at this stage.

An amended version of the draft slip was sent by fax from Mr Stephenson of NMB to Ms Jerabek on 22 July 1996 while Mr Holzapfel was still away from the office. The following day Ms Jerabek reconfirmed in principle that Albingia would take 50% of the proposed treaty. It is likely that she did so without reference to Mr Holzapfel as the amendments were relatively minor and Mr Holzapfel was due back from his holiday the following week.

Mr Holzapfel returned to the office on 29 July 1996. His evidence was that on his return he and Ms Jerabek would have discussed what had happened while he had been away, including discussion of any new business. This would have involved going through the offer faxes and relevant correspondence for all such new business, including the proposed first loss treaty with Arig. Although he had no specific recollection of any such review, this would have been his usual practice upon returning to the office after a holiday. I accept that evidence and therefore find that it is more likely than not that Mr Holzapfel did conduct such a review, and that it included a reading of the fax dated 17 July 1996 with its statement that there was no “corresponding loss record”. He must have appreciated, if he had not already done so, that no loss records for energy business written by Arig had been provided. He must also have appreciated, if it was a matter of any significance to him, that no information at all had been provided concerning Arig’s policy (if any) as to the deductibles which would be applicable to the inwards energy construction risks which Arig proposed to cede to the new first loss treaty.

On 3 September 1996, Mr Stephenson of NMB sent a fax to Ms Jerabek including a revised slip and order for excess of loss reinsurance for US $350,000 in excess of US $150,000. He confirmed that the other 50% of the treaty had been placed with Rhine Re. There is no surviving documentation or evidence about the way in which the business was broked to Rhine Re.

The slip was signed by Ms Jerabek on 4 September 1996 confirming Albingia’s 50% line. She also signed the order for excess of loss protection. She probably did so without needing to consult further with Mr Holzapfel, who was in London at the time.

I accept Mr Holzapfel’s evidence that although Ms Jerabek signed and stamped the first loss treaty on behalf of Albingia and it is her name that appears on the contemporary documents, the underwriting decision to participate in this treaty was Mr Holzapfel’s. Ms Jerabek acted under his authority and subject to his approval throughout. However, it is impossible to say how detailed Mr Holzapfel’s review of the proposed treaty was, what factors weighed with him, or the extent to which he relied on the initial review work undertaken by Ms Jerabek.

On 23 September 1996 NMB sent a fax to Albingia referring to the fact that inclusion in the treaty of the 15 risks written by Arig prior to the July placement was subject to a warranty that there were no known or reported losses on those risks. It advised that some of the policies included on the original schedule had expired, while in other cases losses had been advised and that in order for Arig to comply with the “WNKORL” provision a new schedule had been drafted. Seven of the originally proposed risks were replaced on this new schedule. The fax made clear, therefore, that several of the risks written by Arig earlier in 1996 had suffered losses, albeit that these would no longer be covered under the treaty. The new schedule to the 1996 first loss treaty was signed and scratched by Albingia on 23 September 1996.

On 27 June 1997 NMB invited Albingia to renew the 1996 treaty for risks attaching during a further period of twelve months from 1 July 1997. The terms were amended to allow policies with a duration of up to four years (increased from three) to be covered. NMB attached a table of the accounts declared to the 1996 treaty with details of inception and expiry dates, exposure to Arig and premium to the facility. Some 49 risks had been declared, which was more than the 30 to 40 policies which had originally been anticipated, of which only five were operating accounts. Finally, the fax stated that:

“There is only one claim at this stage, details of which are as follows:

Declaration : Clyde

Date of Loss : 1^st September 1996

Claim Amount : US$55,283 to the facility.

Description : Pipeline damage by a fishing vessel.”

As this fax was sent on a Friday evening it was probably only seen by Ms Jerabek on the morning of 30 June 1997, the following Monday. Mr Holzapfel was away from the office on holiday with his son. Ms Jerabek responded to NMB’s offer on the same day, expressing a lack of enthusiasm about the extension of policy periods to 48 months, but nevertheless confirming the renewal of Albingia’s 50% line, subject to excess of loss protection as offered. Mr Holzapfel’s evidence was that he would have discussed this renewal with Ms Jerabek by telephone. I accept that evidence. However, I can make no finding as to whether Ms Jerabek told Mr Holzapfel about the Clyde Petroleum claim referred to in the NMB fax or about the statement that this was so far the only claim which had been made.

On 9 July 1997, Mr Bilborough sent to Mr Holzapfel and Ms Jerabek a copy of the signing slip and excess of loss reinsurance order for them to sign and return confirming Albingia’s agreement and written line. Ms Jerabek signed and stamped the slip for Albingia’s 50% line on 10 July 1997.

In fact, the Clyde claim referred to in the renewal fax of 27 June 1997 was not the only incident which had occurred at the time of this renewal. Three other incidents had occurred, although there is an issue whether they were material to be disclosed or such as to render false or misleading the representation that there was at this stage only one claim. They were as follows.

First, on 13 November 1996 damage had been caused to the legs and jacking system of a Clyde Petroleum jacking platform near Rotterdam. Arig was only informed of this on 7 May 1997 by a fax from Marsh & McLennan, Clyde Petroleum’s brokers. The fax attached a preliminary report from Matthews-Daniel International (London) LLP, loss adjusters appointed by the leading underwriter. This was dated 21 February 1997. After setting out the circumstances in which the incident had occurred, the report indicated that there might be an issue as to the extent to which loss would be covered under the original policy, concluding with a recommendation that:

“In the meantime, we suggest that a precautionary nett reserve of DFL 2,500,000 is adopted on the assumption that any adjusted cancellation/standby costs fall below the applicable policy deductible.”

The recommended precautionary reserve was equivalent to about US $1.3 million. Arig’s line on the policy was 10%, which would have meant a reserve of about US $130,000, although (assuming a valid claim in this amount) the loss to the first loss treaty after taking account of Arig’s quota share treaty would have been US $97,500.

There is no evidence of any action taken in response to this notification before 17 August 1997 (i.e. after the renewal of the first loss treaty). On that date Arig allocated a claim number to this incident. Subsequently a reserve was made, in fact for a higher amount in the light of more up to date information which by then had been received. While it is possible that Arig’s claims department decided to seek further information in view of the delay in receipt of the Matthews-Daniel report before deciding what action to take, the more likely explanation is that the matter was simply overlooked.

The second incident concerned a policy in favour of National Petroleum Construction Company (“NPCC”) and related to loss and damage to equipment during pipeline installation work in the Tapti Offshore Field, India which had occurred on 7 March 1997. Once again the appointed adjuster was Matthews-Daniel. The adjuster’s preliminary report, dated 15 April 1997, was received by Arig on or about 23 April 1997. It explained that NPCC had not yet provided any claim details, but had advised verbally that it intended to pursue recovery of all costs incurred as a result of the incident which it estimated might amount to about US $800,000. Matthews-Daniel commented, however, that it was not clear from discussions with NPCC how this estimate was derived, while any claim would be subject to a deductible of US $300,000. Moreover, policy coverage was not straightforward. The report concluded:

“Based upon our discussions with the Assured, we do not feel it appropriate at this stage to recommend Underwriters adopt a reserve against this loss. We shall however include our recommendations in respect of reserve in early course once we have clarified a number of issues including the notified costs associated with this incident and the application of appropriate deductibles.”

In accordance with this recommendation Arig did not establish any reserve. Nor did it allocate a claim number until 14 August 1997, after renewal of the first loss treaty.

Finally, on 18 June 1997 Arig received a fax from the broker Alexander Howden notifying it of a loss on a project at Ras Laffan. The fax stated that:

“No loss reserve has been given, as it is difficult to estimate the potential loss at this stage, but expect that the loss is likely to exceed the policy deductible”.

There was also attached a fax from the original insured which stated:

“At this stage it is very difficult to estimate the potential loss, but certainly it looks as through it will exceed the policy deductible of $250,000. Until we are aware of the scope of work necessary to rectify the damage it is very difficult to quantify (maybe up to US$ 1 million) but the Project see [sic.] confident of being able to raise the roof on 26^th June which is only 15 days later than planned, and the Project was in any event ahead of schedule.”

Arig did not establish any reserve for this incident. Even assuming an overall loss of US $1 million, because Arig had only a small line on this insurance the potential loss to the 1996 first loss treaty was only some US $11,775.

When it came to the 1998 renewal of the first loss treaty, Albingia decided not to renew. By then there had been two major losses, both to the 1996 treaty. Claims under the 1996 treaty exceeded US $1.1 million and even with the benefit of the excess of loss protection, the treaty was already making a loss only two years after it had been written. Although the 1997 treaty had so far suffered losses of only US $95,000, the premiums on risks declared to the treaty after three quarters were less than US $70,000 on a net basis, barely covering the cost of the excess of loss cover. Taken together the two treaties were showing a current loss of just under US $150,000. Mr Holzapfel and Ms Jerabek told NMB that they were “quite worried about the development of the account” and saw no point in continuing.

As time went by, losses continued to mount. Between 26 February 1998 and 6 November 2006, Albingia and subsequently Axa paid total claims to Arig of over US $2.5 million under the 1996 treaty and over US $3.2 million under the 1997 treaty. As losses continued, Axa began to question entries and figures in Arig’s treaty statements and made various requests for information and clarification relating to claims for which payment was sought. The process of obtaining information proved extremely slow, mainly due to delays by Arig in responding to Axa’s requests. It is unnecessary for the purpose of this judgment to recite the detail. Axa’s principal concern was that certain claims made by Arig under the treaties appeared to relate to risks which had not been ceded to the treaties prior to notification of the loss. It had no suspicion or grounds for suspicion at this time that there might have been a failure to disclose material facts or misrepresentation at the time when the treaties had been concluded.

Eventually, in an e-mail of 14 July 2010, Axa reserved all its rights. It followed this with further requests and in August 2011 sought to exercise its right to inspect Arig’s records. Owing to delays in agreeing the conditions on which the inspection was to be permitted and arguments as to the scope of the material to be made available to Axa’s inspectors, the inspection did not commence until 7 October 2012. It then took place in Arig’s offices in Bahrain between 7 and 11 October 2012, its primary purpose being to determine the manner in which Arig declared risks to the treaties. A further inspection of NMB’s files took place in London on 5 December 2012, at the offices of Holman Fenwick Willan LLP, Arig’s solicitors.

In the course of these inspections Axa learned of Mr Hylander’s memorandum of 7 July 1996 (see [59] above), as well as the reports received by Arig concerning the three incidents which had occurred before the 1997 renewal of the treaty (see [81] to [89] above). On 5 April 2013 Axa commenced this action claiming (among other things) a declaration that it had validly avoided the treaties.

As already noted, energy construction projects are often undertaken in harsh environmental conditions offshore and are, by their very nature, inherently risky. It can take several years for claims to emerge as risks are placed covering the full period of the construction project and it may be some years before a project is completed. Although claims can emerge at any time, claims in the later period of construction when most of the infrastructure is in place are generally (though not always) more significant financially as there is greater value exposed to risk. There can also be some lag in respect of loss reporting because it can take time for the consequences of an incident to become apparent. The “tail” on an energy construction insurance portfolio can be as long as five years or sometimes even longer. This would be understood by any energy underwriter or reinsurer.

For example, risks could be declared under the 1996 treaty to cover losses occurring on risks attaching during a period of 18 months, with a maximum original policy period of “36 months, plus discovery/maintenance period”. Thus a loss occurring right at the end of the contractual construction period might not even happen until June 2000 and even then it might take some time for Arig to learn about it. It would be less common for a construction loss to occur during the discovery/maintenance period but this was not impossible (such a claim was described by Mr Outhwaite as “a very rare bird”), in which case the tail would be longer. For the 1997 treaty the periods were longer, with a period of 12 months for risks to attach, but a longer construction period of four years rather than three.

Mr Hylander’s predecessor was Mr Simenstand. His approach to underwriting had focused mainly on generating premium income. During his tenure Arig was seeking to grow its business by increasing the volume of risks underwritten, including energy risks, and to maximise premium in order to take advantage of the then prevailing high interest rate environment. Mr Hylander’s evidence was that he adopted a more selective approach, focusing more on the quality of the risk and avoiding some of the higher risk projects which Mr Simenstand had been prepared to write.

This change was presented by Arig in this litigation as if it were a tangible and demonstrable change in the kind of risks being written – as Mr Outhwaite put it, a sea change from an underwriting philosophy which was “unselective, motivated by premium income rather than balance of risk, and without direction” to one which was “cautious and conservative” with clear principles established which “were adhered to in the vast majority of cases”.

In reality I consider that the position was less clear-cut. For example in his witness statement Mr Hylander stated that he would seek adequate deductibles and gave examples of the kind of risks which he would have preferred not to write. These included deep water construction risks, risks exposed to particularly harsh weather conditions, risks involving new and untried technology and risks involving large bore pipelines. Although he acknowledged that despite these preferences he was prepared to consider any kind of construction project, he also stated, apparently without qualification, that “If I was not keen on a risk I would not underwrite it, notwithstanding that this might reduce the total premium income to the account”.

However, Mr Hylander agreed in cross-examination that the position was more nuanced than this evidence might suggest. Under his tenure Arig had indeed continued to write risks with low deductibles where that was all that was available in the market and the only alternative was to write no business at all. Indeed, as noted at [61] above, this was one of the factors which led it to seek first loss protection. Likewise Arig continued from time to time to write risks with each of the characteristics which Mr Hylander would have preferred to avoid, albeit not necessarily on the same scale as before. There was never any absolute prohibition on the writing of risks with any of these characteristics. Realistically, in cross examination Mr Hylander described his list of the risks to be avoided as “more like an intention, a wish list, but you will always, you must always be prepared to make exceptions”. He acknowledged that whether he was prepared to accept such risks would inevitably depend upon a wide variety of circumstances existing at the time they were presented to him, including market conditions, the terms available, the reputation and standing of the insured, whether it was necessary to accept an energy construction risk in order to obtain a share of other more attractive business, and the maintenance of good relationships with an insured or a broker. These are all legitimate commercial considerations and are exactly what would be expected, but they mean that it is not practicable to point to a clear cut or easily demonstrable change in the kind of energy construction risks written by Arig before and after Mr Hylander took charge of Arig’s energy underwriting department.

In January 1995 Arig was in the process of renewing its energy Quota Share reinsurance treaty. This covered mainly operating rather than construction risks. Mr Hylander wrote a memorandum, approved by Mr Khwaja, which was intended to form part of the placing documents. It explained Arig’s change in approach and why it meant that Arig was now generating less premium income than before (emphasis added: Mr Charles Kimmins QC for Axa relied on the words emphasised):

“As you will see there has been a constant improvement in treaty results over the years. Partly due to the restriction imposed by the Marine Energy market in general but mainly due to changes in our portfolio structure and underwriting philosophy, we do not anticipate any major changes for 1995 in the energy market nor have we planned any material restructuring of the portfolio or underwriting guidelines.

As you are aware, we adopted a slightly different underwriting policy in 1993, in order to achieve an improved portfolio and a wider spread, by implementing the following:

a)

With effect from Jan. 1993 we stopped underwriting liabilities for US based assureds except when forming an integral part of an exploration drilling or production programme.

b)

With effect from 1^st November 1993 we have limited our acceptance on construction risks to some areas, mainly Middle East and on excess levels in Europe, Africa and the Far East with maximum risk period of 3 years. You will see from the attached that in 1994 we accepted only one construction risk that runs for 15 months only.

Although the above philosophy has improved our results, they understandably had a negative effect on our premium income.”

I consider that this is a fair summary, although by 1996 Arig was once more seeking to increase its book of construction risks as the premiums available for operating risks were becoming more competitive. The reality was that a balance (or as Mr Hylander described it, a compromise) had to be struck between maintenance of Arig’s overall market position and the underwriting strategy that Mr Hylander would in ideal circumstances have wished to pursue. I conclude from his evidence that there was a genuine change in direction, but that it was the kind of intangible and gradual change which although real would have been difficult to demonstrate in any quantifiable way as part of any broking exercise. Moreover other members of his team had authority to write risks without reference to Mr Hylander and would sometimes do so. Although he was the main influence on Arig’s underwriting strategy, the strategy was properly to be regarded as that of Arig and not of any individual underwriter who, after all, might not necessarily remain with the company.

Mr Bryan relied on Arig’s 1993 underwriting guidelines as demonstrating a change in strategy but these in my judgment are of no assistance in determining the extent to which there was such a change. The guidelines included the following objectives for Arig’s marine energy department:

“To underwrite within the prescribed limits an Oil/Gas Offshore account including certain other Marine Energy related business applying a consistent underwriting approach aimed at making an underwriting profit.

To pursue a selective underwriting policy with a focus on maintaining and further developing a well spread and balanced book of business.”

Under the heading “General Philosophy” the guidelines stated among other things:

“The Marine Energy Department shall maintain a conservative underwriting philosophy and not support aggressive underwriting attitudes.

The Marine Energy Department should function and operate as a truly professional underwriting unit by providing absolutely first class service including prompt settlement of claims.

The Marine Energy Department will establish flexibility of approach and use it within the broad framework of this document.”

With all due respect to those whose task it is to write such documents, these are guidelines at a high level of generality. It is hard to imagine such guidelines saying anything else. Moreover there is no evidence to show what any previous guidelines stated. It is therefore impossible to determine whether or to what extent the 1993 guidelines demonstrate a change of policy from what had gone before. If, as seems entirely possible, the paragraphs quoted above had survived substantially unchanged from previous versions, and had been applicable during Mr Simenstand’s tenure, that would only demonstrate that Arig’s underwriters were sometimes capable of ignoring whatever guidelines were in place.

For the purpose of this case Mr Hylander was asked by Arig’s lawyers to review each of the 21 energy construction risks written during the six months before his arrival at Arig, as well as those written afterwards during 1991, 1992 and 1993. He provided his comments upon many of those risks and in some cases expressed the opinion that he would not (or in some cases that he would) have written those risks if they had been presented to him, although in most cases he did not positively express an opinion either way. Based upon these comments, Arig’s lawyers prepared a series of schedules designed to show which of the risks written from 1 January 1991 onwards would have been written by Arig if Mr Hylander’s more conservative approach had been in place from that date. This was said to lead to the conclusion that loss statistics covering only the risks which Mr Hylander would have written would have shown a much more favourable outcome than the actual results achieved. It showed loss ratios of less than 30% for the 1991 and 1992 years (or, using slightly different assumptions, about 40% for each of these years) which contrast with the actual figures of 109% and 65% in the table at [32] above. I shall refer to the result of this exercise as the “Arig reconstruction”.

I consider, however, that this attempted reconstruction, which is essentially a matter of submission with only limited evidential support, is of very little value. So much depends upon the detailed circumstances of the individual risks and the way in which they were (or would have been) presented to Arig, none of which can now be known or evaluated, as well as the market conditions prevailing at the time. The exercise, however well intentioned, is highly speculative. Indeed Mr Hylander accepted that although he had done his best to give comments where he could, it had not been a straightforward exercise to do this some twenty years after the event. That is not surprising. I consider that Mr Holzapfel’s reaction when first told about this exercise in cross examination was telling. It was:

“Let me guess, without having gone through that, he would only have written the risks with the best performance and ceded them to the first loss treaty, is that right?”

Mr Blackwood for Arig described that reaction as cynical. However, while I accept that the exercise was carried out in good faith (i.e. in a genuine attempt to reconstruct what Mr Hylander was likely to have done), it seems to me that this is likely to have been at least the first reaction of a reasonable reinsurer presented with anything like this kind of hindsight exercise.

At first sight I had thought it doubtful whether anything of this sophistication and detail would ever form part of an actual broke. Mr Stow’s evidence, however, was that there had been occasions when he had done a similar detailed exercise as part of a broke in order to demonstrate what effect a new underwriter’s new approach would have had on a proposed reinsured’s historical results, although (as he said) this would require a client with very good systems as well as staff with the time to undertake such a complex exercise. I accept Mr Stow’s evidence on this point, but it is important that, as he made clear, on those occasions he would always present to the reinsurer the original unvarnished statistics together with the reconstructed statistics by way of comparison and would not think it right to proceed on the basis that if the reconstructed statistics looked good enough, nothing at all need be disclosed:

“MR JUSTICE MALES: Is what you are describing to me an exercise where the end result is a presentation to the insurer which says ‘Well, yes, we have got these past statistics which, on the face of them, look pretty bad but actually when you apply the new strategy or whatever to them, it looks a lot more promising’?

A. Exactly.

MR JUSTICE MALES: As distinct from a case where you do the exercise and then decide not to pass any of that on to the reinsurer?

A.

I think in that particular case the exercise would fail because it would be evident to the cedant and to the broker that the risk wasn't placeable.”

Nothing like the result of the reconstruction exercise carried out by Arig’s lawyers for the purpose of this case was prepared at the time or disclosed to Albingia. It does not appear ever to have occurred to Mr Bader or Mr Stow to undertake such an exercise.

There were in addition other flaws in the Arig reconstruction. It made no attempt to assess whether risks accepted by Mr Hylander’s predecessor in 1989 and 1990 would have been accepted under Mr Hylander’s new approach. I was told that Arig had not disclosed any documents which would have enabled the exercise to be attempted for these earlier years. Moreover, it assumed that Mr Hylander would not have written a small number of the risks which he actually did write in 1993 after his new and more conservative strategy was in place. Although these were few in number and did not in themselves have a major impact on the overall results, I consider that this assumption calls into question the validity of the reconstruction exercise more generally.

Thus, while I accept that Mr Hylander did seek to introduce a more rigorous and selective underwriting policy than his predecessor had followed and that he was successful in doing so, it is not possible now to demonstrate in any concrete or quantifiable way what impact his new policy would have had if it had been in place during the period from 1989 to 1992. Nor is there any valid reason to conclude that this would ever have been possible. Certainly it was never attempted at any material time.

As set out in section 18 of the Marine Insurance Act 1906 which in this respect codifies the common law, the assured must disclose to the insurer, before the contract is concluded, every material circumstance which is known to the assured. If the assured fails to make such disclosure, the insurer may avoid the contract. Every circumstance is material which would influence the judgement of a prudent insurer in fixing the premium, or determining whether he will take the risk. However, it is not necessary to disclose any circumstance which diminishes the risk, which is known to the insurer or which an insurer in the ordinary course of his business ought to know, or as to which information is waived by the insurer.

As Hobhouse J put it in Iron Trades Mutual Insurance Co Ltd v Companhia de Seguros Imperio [1991] 1 Re LR 213, the assured’s duty is essentially a duty to make a fair presentation of the risk to the insurer. In this regard a minute disclosure of every material circumstance is not required; the assured complies with the rule if he discloses sufficient to call the attention of the underwriter to the matter in such a way that, if he desires further information, he can ask for it; accordingly a presentation will be fair and accurate if it would enable a prudent insurer to form a proper judgement, either on the presentation alone, or by asking questions if he was sufficiently put upon enquiry and wanted to know further details (see Arnould (18^th Edition, 2013) at para 16-81 and Garnat Trading & Shipping (Singapore) Pte Ltd v Baominh Insurance Corporation [2010] EWHC 2578 (Comm), [2011] 1 Lloyd’s Rep 366 at [135]).

Likewise, as section 20 of the 1906 Act provides, every material representation made by the assured or his agent to the insurer during the negotiations for the contract, and before the contract is concluded, must be true. If it is not, the insurer may avoid the contract. A representation is material which would influence the judgement of a prudent insurer in fixing the premium, or determining whether he will take the risk.

Whether a circumstance or a representation is material is a question of fact. The question is objective, in the sense that what matters is the judgement of a prudent insurer and not that of the particular insurer. The circumstance not disclosed or the representation made need not be decisive. It is material if it would “influence the judgement” of the prudent insurer even if, in the end, the prudent insurer would still accept the risk on precisely the same terms as to premium and other matters as those on which he did in fact accept it (Pan Atlantic Insurance Co Ltd v Pine top Insurance Co Ltd [1995] 1 AC 501). Materiality therefore covers a wide spectrum, from matters which (if disclosed) would mean that no prudent insurer would under any circumstances have anything to do with the proffered risk to matters which the prudent insurer would wish to take into account but which would ultimately make no difference to what he would actually do.

As already mentioned, there is no duty to disclose circumstances as to which information is waived by the insurer. The applicable principle here is as stated in MacGillivray on Insurance Law, 12^th Edition (2012) at para 17-088 in a passage which, as it appeared in an earlier edition, was approved by Longmore and Peter Gibson LJJ in WISE (Underwriting Agency) Ltd v Grupo Nacional Provincial SA [2004] EWCA (Civ) 962, [2004] 2 Lloyd’s Rep 483 at [110] and [130] and (despite the different approach of Rix LJ in the latter case) was regarded by Flaux J in Synergy Health (UK) Ltd v CGU Insurance Plc [2010] EWHC 2583 (Comm), at [172] as binding on this court:

“The assured must perform his duty of disclosure properly by making a fair presentation of the risk proposed for insurance. If the insurers thereby receive information from the assured or his agent which, taken on its own or in conjunction with other facts known to them or which they are presumed to know, would naturally prompt a reasonably careful insurer to make further enquiries, then, if they omit to make the appropriate check or enquiry, assuming it can be made simply, they will be held to have waived disclosure of the material fact which that enquiry would necessarily have revealed.”

Rix LJ, on the other hand, stated the position as follows at [64] of the WISE case:

“Ultimately, it seems, the question is: Has the insurer been put fairly on inquiry about the existence of other material facts, which such inquiry would necessarily have revealed? The test has to be applied by reference to a reasonably careful insurer rather than the actual insurer, and not merely by reference to what such an insurer is told in the assured's actual presentation but also by reference to what he knows or ought to know, i.e. his s. 18(3)(b) knowledge. The reasonably careful underwriter is neither a detective on the one hand nor lacking in common-sense on the other hand. Mere possibilities will not put him on inquiry, and very little if anything can make up for non-disclosure of the unusual or special. Overriding all, however, is the notion of fairness, and that applies mutually to both parties, even if the presentation starts with the would-be assured.”

As established by the Pan Atlantic case, a material misrepresentation or non-disclosure will not entitle the underwriter to avoid the policy unless the misrepresentation induced the making of the contract, using "induced" in the sense in which it is used in the general law of contract. Here the question is subjective, as it is concerned with what the particular underwriter who wrote the contract would have done if a fair presentation had been made to him. Generally, therefore, the insurer or reinsurer seeking to avoid on whom the burden of proof of inducement lies must call the underwriter concerned to give evidence, although on occasion he may be able to rely on a “presumption of inducement”. The position was summarised by Clarke LJ in Assicurazioni Generali SpA v Arab Insurance Group (BSC) [2002] EWCA Civ 1642, [2003] Lloyd’s IR 131:

“59.

It seems to me that the true position is that the misrepresentation must be an effective cause of the particular insurer or reinsurer entering into the contract but need not of course be the sole cause. If the insurer would have entered into the contract on the same terms in any event, the representation or non-disclosure will not, however material, be an effective cause of the making of the contract and the insurer or reinsurer will not be entitled to avoid the contract. …

60.

Those principles seem to me to be consistent with the approach of this Court in St Paul Fire & Marine v McConnell: see per Evans LJ (with whom Rose and Nourse LJJ agreed) at pp 124-5, where he discussed the general principles, and at p 127, where he held that, if the three underwriters who gave evidence had been told the truth, on no view would they have underwritten the insurance at the same premium on terms which included subsidence risk. Evans LJ also considered the role played by presumption in this class of case. He did so in the context of a fourth underwriter who was not called to give evidence, no doubt because the trial took place before the decision of the House of Lords in Pan Atlantic.

61.

Evans LJ put the position thus at p 127:

‘The existence of such a presumption is recognised in the authorities; see Halsbury's Laws vol 31 par 1067 where the law is stated as follows:

Inducement cannot be inferred in law from proved materiality, although there may be cases where the materiality is so obvious as to justify an inference of fact that the representee was actually induced, but, even in such exceptional cases, the inference is only a prima facie one and may be rebutted by counter evidence.’

…

It appears to me that a presumption of this kind really amounts to no more than this. It simply operates where the evidence before the Court is enough to lead to the inference that the insurer or reinsurer was, as a matter of fact, induced to enter into the contract.

62.

In all the circumstances I would summarise the relevant principles of inducement in this context in this way:

1.

In order to be entitled to avoid a contract of insurance or reinsurance, an insurer or reinsurer must prove on the balance of probabilities that he was induced to enter into the contract by a material non-disclosure or by a material misrepresentation.

2.

There is no presumption of law that an insurer or reinsurer is induced to enter in the contract by a material non-disclosure or misrepresentation.

3.

The facts may, however, be such that it is to be inferred that the particular insurer or reinsurer was so induced even in the absence from evidence from him.

4.

In order to prove inducement the insurer or reinsurer must show that the non-disclosure or misrepresentation was an effective cause of his entering into the contract on the terms on which he did. He must therefore show at least that, but for the relevant non-disclosure or misrepresentation, he would not have entered into the contract on those terms. On the other hand, he does not have to show that it was the sole effective cause of his doing so.”

It is sometimes said that the presumption of inducement will only come into play in cases in which for good reason the insurer is unable to call the underwriter who wrote the risk in question. I doubt, however, whether this is a rule of law, although no doubt it will generally be true as a matter of fact.

Colman J sounded an important note of caution in North Star Shipping Ltd v Sphere Drake Insurance Plc [2005] EWHC 665 (Comm), [2005] 2 Lloyd's Rep 76 at [254] and [255] as to how a court should approach an underwriter’s evidence that, if he had known the true facts, he would not have written the policy, either at all or on the terms on which he did:

“254.

In evaluating the underwriters' evidence it is important to keep firmly in mind that all their evidence is necessarily hypothetical and that hypothetical evidence by its very nature lends itself to exaggeration and embellishment in the interests of the party on whose behalf it is given. It is very easy for an underwriter to convince himself that he would have declined a risk or imposed special terms if given certain information. For this reason, such evidence has to be rigorously tested by reference to logical self-consistency, and to such independent evidence as may be available.

255.

Further, in a case such as this, where diffuse matters can be said to be material facts, it is unrealistic to evaluate the issue of inducement except on the hypothesis of the disclosure of all the facts found to be material facts as distinct from isolated material matters.”

For the reasons given at [13] above, this caution is particularly necessary in a case such as the present where the underwriter called by the reinsurer has no recollection of the transaction in question. In contrast with a typical fact finding exercise where the court’s task is to determine on the balance of probability whether something did happen, taking into account the contemporary documents, witness recollection and credibility, as well as the inherent probabilities of the situation where those can be ascertained, the issue here is whether something which ex hypothesi did not happen would have happened if the facts had been different. In such a case the usual tools for evaluating a witness’s evidence (such as the extent to which his evidence either generally or on the particular point at issue is supported by or consistent with the contemporary documents) are not available, at any rate in the same way. Moreover, as the witness will generally have convinced himself that he would not have written the risk and will generally be (as Mr Holzapfel was) an honest man, his assertion that he would not have done so if a fair presentation had been made will often be made with conviction and will appear plausible. A healthy scepticism is therefore appropriate in evaluating such evidence, together with a proper regard for the burden of proof and an appreciation of where on the wide spectrum of materiality the particular non-disclosure which is relied upon falls. As Staughton LJ put it in Kausar v. Eagle Star Insurance Co Ltd [2000] Lloyd’s Rep. IR 154:

“Avoidance for non-disclosure is a drastic remedy. It enables the insurer to disclaim liability after, and not before, he has discovered that the risk turns out to be a bad one; it leaves the insured without the protection which he thought he had contracted and paid for. Of course there are occasions where a dishonest insured meets his just deserts if his insurance is avoided; and the insurer is justly relieved of liability. I do not say that non-disclosure operates only in cases of dishonesty. But I do consider that there should be some restraint in the operation of the doctrine. Avoidance for honest non-disclosure should be confined to plain cases.”

Axa has not suggested that any of the non-disclosures or misrepresentations on which it relies were deliberate or even negligent. Its case is that it is entitled to avoid for innocent non-disclosure or misrepresentation. In principle that is so, but Axa bears the burden of making that case good.

The main issue in the case is whether Axa is entitled to avoid the treaties for non-disclosure of or misrepresentation as to the existence of loss statistics relating to Arig’s existing book of inwards marine energy construction risks written during the period 1989 to 1995.

It is convenient to begin with the question of misrepresentation. NMB’s fax dated 17 July 1996 in which the 1996 treaty was first proposed to Albingia contained the statement that:

“This is a new Treaty for the Reassured and as such does not have a corresponding loss record.”

Axa’s case is that this statement constituted a representation to the effect that Arig had no loss statistics for energy construction risks of the kind which would be declared to the 1996 treaty, which was untrue as Arig did have such statistics. I accept that, if that representation was made, it was untrue. Arig challenged this, on the ground that “energy construction risks of the kind which would be declared to the 1996 treaty” were risks of the kind which would be written by Mr Hylander, which were different because of his change of underwriting strategy from the risks which had resulted in the statistics set out in the table at [32] above, and that Arig had no such statistics. I would not have accepted this. That would not have been a fair understanding of such a representation and in any event Arig did have statistics for the 1992 and 1993 underwriting years for risks which had in fact been written by Mr Hylander, albeit that the risks written in 1994 and 1995 were too few in number to provide useful information.

Arig’s case, however, is that there was no such representation, the statement made meaning no more than that because the treaty was a new treaty (in contrast with a renewal of an existing treaty) Arig had no records of losses incurred under the treaty in previous years. In this connection Arig emphasises the words “as such”, reading the statement as meaning “This is a new Treaty for the Reassured and as a new treaty does not have a corresponding loss record”.

This is a short point on which I prefer Arig’s case. Axa’s case ignores the effect of the words “as such”. Moreover the subject of the statement is not Arig but the new treaty: thus it is the new treaty which does not have a corresponding loss record, not Arig as the reassured. Albingia has to read the statement as if it said that “This is a new Treaty for the Reassured and the Reassured does not have a record of losses for risks corresponding to those of the kind to be declared under the treaty”. But that is not what it says. It is true that if the statement is understood as Arig contends, the second half of the sentence is largely pointless or obvious: if the treaty was a new treaty, there could not possibly be records of losses incurred under it in previous years. However, people do sometimes say things which are unnecessary or obvious. The argument that, understood as contended by Arig, the statement is no more than a statement of the obvious does not justify giving it a meaning which it does not properly bear. In order to found a right to rescission a representation must be clear.

The position might have been different if Arig or NMB on its behalf had made a true statement that there were no loss records for the treaty because it was a new treaty, with the deliberate intention of causing Albingia to believe that Arig had no past loss records for energy construction risks of any kind. However, that was not suggested and need not be considered further.

Nevertheless I can understand that Albingia may not have focused on the precise meaning of the statement in the same way as has been necessary in this litigation. It would not be surprising if Mr Holzapfel or Ms Jerabek had understood it as saying that Arig had no past loss statistics for energy construction risks and, if they did, that may explain why neither of them asked whether Arig did have any such statistics. By the time that he came to give evidence Mr Holzapfel was clearly convinced that he would have understood the statement in this way but, as he had no recollection of seeing it, it is impossible to say whether he did so at the time. The same applies to Ms Jerabek from whom there was no evidence to say one way or the other what impact if any this statement had upon her.

Accordingly Axa’s case based on misrepresentation fails although ultimately that case would have added little or nothing to its case based on non-disclosure. That is because, if the statement in question had not been made so that nothing at all had been said about past records, the question would still arise whether Arig was under a duty to disclose its past loss statistics. I turn now to that question.

As noted at the outset of this judgment, past loss statistics relating to insurance written by a proposed reinsured will generally be material as affecting the judgment of a prudent reinsurer deciding whether to accept the offered risks or upon what terms to do so and must therefore be disclosed. Although statements to that effect are included in legal textbooks such as Arnould and Carter, the proposition is one of fact rather than law, materiality being a question of fact. It is nevertheless obvious common sense, at any rate as a general proposition. The prudent prospective reinsurer will want to know what the reinsured’s experience is of losses of the kind which it is being asked to reinsure and at least to take that experience into account in deciding whether to accept the offered risks or upon what terms to do so. If the experience is particularly bad, he will expect to be told. As Longmore J said in Marc Rich & Co AG v Portman [1996] 1 Lloyd’s Rep 430 at page 443:

“If an assured has a substantial loss experience and makes no mention of this fact to an insurer who must be taken to know that there is or is likely to be a loss experience, has there been a fair presentation of the risk? In my judgment the answer is "No" for the simple reason that, even if the insurer must be taken to be aware of the existence of a loss experience, he does not know how substantial that loss experience is. He is entitled to assume that there has been a fair presentation of the risk. A presentation which makes no reference to an existing loss experience can only be fair if the losses are modest or insignificant; a prudent underwriter will be entitled to assume that if losses exist, they are not such as to be worth mentioning. If the fact is that there is a history of substantial losses, there has not been a fair presentation. The fact that the insurer knows or is presumed to know that a loss experience exists says nothing about its size and, as I say, if nothing is disclosed about it to the insurer he is entitled to assume it is insignificant.”

Although this was a case of insurance against liability for demurrage, the point is of general application.

It may be that when a reinsured has past loss records extending back over a number of years, there is room for debate as to how far back any presentation of such records must extend. Mr Outhwaite’s evidence was that such disclosure would not generally go further back than four or five years. I would accept, based on Mr Outhwaite’s experience, that this represents the usual practice of the market which is therefore a relevant consideration, but the principle must be that the disclosure must be sufficient to constitute a fair presentation of the risk. What that will require will depend on all the circumstances of the case, including the terms of the proposed reinsurance, the extent to which more recent records give a fair and representative picture of the reinsured’s loss experience, and the extent to which records more remote in time are relevant to the risk which is under consideration. In general, the longer the “tail” applicable to any class of business, the further back it is likely that past records will need to be disclosed in order to provide a fair presentation, while bearing in mind that a reinsurer who wants to ask for more can always do so.

In the present case there is an issue whether, if disclosure of past loss records was material at all, it was necessary for disclosure to extend back to the 1989 and 1990 underwriting years which represented the worst losses incurred by Arig. However, although that issue is relevant at a later stage of the case, Arig did not disclose any past loss records and it is therefore necessary to consider whether the presentation of the risk without any such disclosure was unfair.

Arig’s case is that in the particular circumstances of this case disclosure of its historic loss statistics would not have influenced the judgment of a prudent underwriter considering whether to write the 1996 or 1997 treaties. For this purpose it relies on the evidence of Mr Outhwaite, who gave three reasons for reaching this conclusion:

“In my view it was not material to the risk that Mr Holzapfel was presented with for ARIG to have disclosed energy construction loss statistics, for three reasons:

(a)

Energy construction risks are unique risks and therefore loss statistics in relation to other risks written historically are simply not material from an actuarial point of view – every risk is, in a sense, ‘one-off’ and is written individually;

(b)

Like is not being compared with like as there was a change in underwriting strategy and so the nature of risks written changed, coupled with the fact that Arig’s underwriter, Lars Hylander (“LH”), had implemented a conservative underwriting policy which would have diminished the risk underwritten by Thomas Holzapfel; and

(c)

Since LH implemented the conservative underwriting policy, LH has underwritten an insignificant number of risks, such that no meaningful conclusions could be drawn from them.”

In cross examination, however, Mr Outhwaite made clear that his first reason, the uniqueness of energy construction risks, was not an independent reason for saying that past loss records were immaterial in the sense that a prudent underwriter would not wish to take them into account. On the contrary he accepted that they might be “relevant” or “of some significance” (i.e. material), and explained that he was saying no more than that they were not actuarially relevant as providing a reliable prediction of losses which would be incurred under the proposed reinsurance. While that latter point may be true, it does not address the test for materiality.

Mr Outhwaite’s third reason, the absence of sufficient statistics for meaningful conclusions to be drawn after Mr Hylander had implemented his change of strategy, is valid so far as it goes but is dependent on his second reason, namely that losses incurred on risks written before the change of strategy were immaterial. That, therefore, is the critical issue so far as materiality is concerned.

Despite Mr Outhwaite’s many years experience of insurance and reinsurance, I have no doubt that past loss records of a prospective reinsured would influence the judgment of a prudent reinsurer in fixing the premium, or determining whether he will take the risk, and that this would be so (a) in the case of reinsurance of energy construction risks and (b) notwithstanding a change in underwriter and/or underwriting strategy on the part of the reinsured. The prudent underwriter considering whether or on what terms to reinsure such risks would wish to be told about such losses. It could then be explained to him, if appropriate, that such losses were not properly comparable with the risks that were likely to be ceded to him under the proposed reinsurance because the identity of the underwriter had changed or because a more conservative strategy had been or was going to be adopted and he could assess the weight to be given to that explanation for himself and decide what to do. But it would not be consistent with the duty of utmost good faith for the reinsured to conceal from the reinsurer the existence of poor or disastrous loss records on the grounds that the underwriter or the strategy had changed, even if that is true. To conclude otherwise would go a long way to deprive the reinsurer of the protection of a fair presentation to which it is entitled and which, in the case of reinsurance where the reinsurer makes no underwriting judgment other than as to the skill and judgment of his reinsured, is the only protection which he has. As Waller LJ said in Bonner v Cox [2005] EWCA Civ1512, [2006] 1 CLC 126:

“106.

But the reinsurer is not defenceless or at the mercy of his reinsured. As a market professional he is able to protect himself by way of pre-contract disclosure. Before he writes the reinsurance or on renewal he is entitled to a fair presentation of the risk which includes the type of business his reinsured proposes to write or has written and its history. The wordings should clearly define the nature of the risks reinsured and could permit the reinsurer to monitor the progress of the business if necessary. If he fails to protect himself in this way he should not be able to blame his reinsured.

107.

The fact that the reinsurer does not exercise his own underwriting judgement about the individual risks to be reinsured is a feature, as we have said, of almost all reinsurance. The reinsurer exercises his underwriting judgement as to which underwriter he will reinsure and upon what terms.”

I reach this conclusion as to materiality for four reasons, any one of which would be sufficient. Together they are overwhelming.

First, although Mr Outhwaite expressed his opinion, and despite his general experience of the market, his evidence on this issue was only an expression of opinion as distinct from being based upon any actual experience of any comparable situation or market practice in circumstances where there had been a change of underwriter or change of strategy. Although that opinion was of some weight, it did not depend on any factors peculiar to the reinsurance of energy construction risks. Mr Hartigan’s view was the opposite. Although he has less experience than Mr Outhwaite of the reinsurance of energy risks, that does not in my judgment detract from his evidence on this issue. On this issue I prefer Mr Hartigan’s evidence as according more with commercial sense.

Second, I consider that the actions of those involved at the time were important. Although the question whether the past loss records would influence the judgment of a prudent reinsurer is objective, those involved were reasonable professionals in the market and their actions are therefore capable of casting light on this question. In this regard it is in my view significant that when first loss protection was sought by Arig through Swire Fraser it was treated as axiomatic that Arig’s past loss records should be disclosed to reinsurers, together with the best gloss upon those records that Mr Bader could think of to persuade reinsurers that the historic results did not reflect the outcome which might be expected on the new risks for which Arig was now seeking reinsurance: see [38] to [42] above. There was no question of withholding those records on the ground that they were immaterial because circumstances had changed. Mr Stow’s evidence of occasions when he had undertaken a reconstruction exercise similar to that undertaken by Arig’s lawyers for the purpose of this case was to the same effect: see [108]. Again there was no question of a unilateral decision by the reinsured or the broker that if the reconstruction exercise produced sufficiently good figures, nothing at all need be said to the reinsurer about past losses.

Third, the evidence of Mr Stow and Mr Bader that past loss records would inevitably be part of a conversation about placing reinsurance (“a part of underwriting, any underwriting discussion, you look at the information, loss records is part of the information”: see [56] above) supports Mr Hartigan’s view. Although it is the case that not even the existence of Arig’s past loss records was disclosed to Albingia, there was no satisfactory explanation for this.

Fourth and importantly, Mr Hylander’s memorandum of 7 July 1996 puts the materiality of Arig’s experience of past losses beyond doubt: see [59] to [65] above.

I conclude, therefore, that Arig’s past loss records were material and that there was therefore a failure by Arig to disclose a material circumstance known to the assured. The presentation made was not fair. That is so regardless of whether a fair presentation would have required disclosure of records going back to 1989 or only to 1991.

Arig contended (a) that Albingia waived disclosure of past loss records because it failed to ask for them and (b) that disclosure was unnecessary because the change in underwriter and underwriting strategy diminished the risk of loss which Albingia was being asked to assume.

As to waiver, Mr Bryan for Arig submitted (perhaps ambitiously) that the correct approach was the minority approach of Rix LJ in WISE (Underwriting Agency) Ltd v. Grupo Nacional Provincial SA [2004] EWCA Civ 962, [2004] 2 Lloyd’s Rep 483 at [64] which was to be preferred to that of Longmore and Peter Gibson LJJ in the same case; that on the issue of waiver Marc Rich & Co AG v Portman [1996] 1 Lloyd’s Rep 430 was wrongly decided; and that Flaux J was wrong to say in Synergy Health (UK) Ltd v CGU Insurance Plc [2010] EWHC 2583 (Comm), at [172] that this court is bound by the majority approach of Longmore and Peter Gibson LJJ in WISE. As I understand it the issue here is whether as a matter of law there can only be a waiver if there has first been a fair presentation of the risk or whether (as Arig submitted) there may be some circumstances where fairness to both parties may lead to the conclusion that the insurer (or here, the reinsurer) has waived disclosure because even in the absence of a fair presentation he is put on inquiry as to the circumstances in question.

Arig’s case on this issue, however, is premised upon a finding that Albingia knew or ought to have known that past loss records existed, and therefore ought to have asked for them if it wanted to see them before deciding whether to write the treaty. I would accept that, if Albingia had known that Arig had been writing energy construction risks on any scale, it ought to have known that there would be losses, given the hazardous nature of such construction projects.

However, I do not accept that Albingia knew or had reason to know that Arig had written such risks. Arig submitted that this ought to have been apparent from the reference to construction risks in the presentation for the 1996 quota share treaty and from the note that “concrete construction risks” were no longer written (see [44] above) but this presentation had been some eight months earlier and referred to only two construction risks which on any view did not represent a significant volume of existing business and did not necessarily mean that there had been losses. Moreover the fact that concrete construction risks were no longer written told Albingia nothing of any relevance. In any event, as the authorities make clear, a reinsurer is not required to act as a detective. Albingia was not required to search through other placing files to see whether they might contain information relevant to but not included in the presentation with which it was faced.

Accordingly Arig’s case of waiver fails on the facts whichever of the two approaches in law referred to above is correct.

The further argument that disclosure of past loss records was unnecessary because the change in underwriter and underwriting strategy diminished the risk fails, in my judgment, to engage with Axa’s case. That case was not that the change in underwriter or underwriting strategy ought to have been disclosed, but that Arig’s past loss records ought to have been. It was open to Arig to seek to persuade Albingia that those losses ought to be viewed in the context of the change in underwriter and strategy which had occurred, and therefore that they should not affect Albingia’s decision, but to say that the change was a factor which diminished the risk misses the point.

The final question in relation to avoidance of the 1996 treaty is whether Albingia has discharged the burden of proving that if there had been a fair presentation of the risk, it would not have written the treaty or would have done so on different terms.

One issue here is whose decision would have been relevant, Mr Holzapfel’s or Ms Jerabek’s. There are potentially three different issues in which their respective roles might need to be considered. The first is the purely factual question of whose decision it was to conclude the treaty. I have set out my factual findings on that issue in the narrative section of this judgment. The second issue, potentially at any rate, is the issue of materiality, but that is concerned objectively with the judgment of a hypothetical prudent underwriter. Except to the extent that the evidence of Mr Holzapfel bears on that question (which in the event I do not think it does to any material extent) it is not relevant to that issue. The third is the issue of inducement which, as already noted, is hypothetical – what would have happened if a fair presentation had been made. A fair presentation would have involved at least the presentation of some past loss records, together with at least some “as if” statistics and some explanation of the change in underwriter and change in strategy. In that hypothetical scenario I am confident that the decision would have been referred to Mr Holzapfel and that he would have looked at the position with care. That would be so even if, as Arig contended, there was in fact minimal or even no involvement by Mr Holzapfel in the actual writing of the risk in the circumstances in which it was actually presented, and all relevant decisions were in fact taken by Ms Jerabek. To that extent, the factual issue as to who in fact made the decision to accept the risk as it was actually presented is largely irrelevant.

I conclude, therefore, that the relevant question for the purpose of the issue of inducement is whether Mr Holzapfel would have written the risk if a fair presentation had been made and that Axa’s failure to call Ms Jerabek is ultimately of no real importance. (If it is relevant, I find that no good reason was given for this failure, although it would not have been surprising if Ms Jerabek also had no recollection of this transaction and therefore little or no relevant evidence to give).

The next question is what a fair presentation would have consisted of and, in particular, whether it would have required Arig to disclose the far worse 1989 and 1990 figures as well as those going back to 1991. On this issue Mr Outhwaite’s evidence represents a useful starting point. He said that the figures which would normally be presented to reinsurance underwriters in 1996 would be those from 1991 onwards (i.e. going back five years), that this represented standard practice, and that the general view was that it was not desirable to go back too far as old records would relate to different market conditions and different risks which would not be so relevant to current conditions. As Mr Outhwaite pointed out, his view was supported by the fact that records going back to 1991 but no earlier had been presented in the case of other first loss reinsurances broked by NMB (see [47] to [51] above). While it appears that the Atkin syndicate had only been formed in 1992, there is no reason to suppose that AIG and the Copping syndicate had not been active in the market before 1991, but in both cases they limited the disclosure to 1991 onwards and Albingia did not ask to see anything going further back.

Moreover it was well known (and Mr Holzapfel either did know or should have known) that 1989 and 1990 had been disastrous years in the energy market generally and that market conditions during those years had been very different from those prevailing in 1996. There had been an influx of capital into the insurance market during the 1980s, which resulted in risks being placed with very low deductibles, low premiums, and much wider cover than was wise. As Mr Outhwaite put it, if an underwriter offered a risk in 1996 had declined it because of losses incurred by the reinsured on business written in 1989 or 1990, he would have ended up writing no new business at all. His view, therefore, was that statistics going back to the 1989 and 1990 underwriting years did not need to be disclosed and ought not to be.

One further factor to which Mr Outhwaite drew attention was that the risks which could be declared under the proposed 1996 treaty were limited to a maximum original policy period of “36 months, plus discovery/maintenance period”. It was therefore likely, even in the case of risks attaching towards the end of the 18 month period for declarations, that at any rate most claims would be made during a five year period from inception. This, he suggested, was a further reason why disclosure of five years worth of claims experience should be sufficient to constitute a fair presentation.

I consider that there is force in the evidence of Mr Outhwaite summarised above and that in general disclosure of figures going back five years should be sufficient to constitute a fair presentation of the risk. Nevertheless I conclude on balance that the 1989 and 1990 losses ought to have been disclosed if the presentation was to be fair. That was the course adopted by Arig when seeking reinsurance through Swire Fraser. The contrast as at the end of 1995 between the loss ratios for 1989 and 1990 on the one hand (424% and 225% respectively) and 1991 and 1992 on the other (109% and 65%) was so great that fairness required this to be drawn to the reinsurer’s attention. As Longmore J explained in the passage from Marc Rich & Co AG v Portman [1996] 1 Lloyd’s Rep 430 cited at [131] above, a prudent underwriter is entitled to assume that if losses have not been disclosed, they are not significant. Presentation of loss ratios which were positive in 1992 and only marginally negative in 1991 would not put the reinsurer on notice of the massive losses incurred in the 1989 and 1990 years. If those losses could be explained as being attributable to particularly bad market conditions in those years or to a particularly disastrous underwriting strategy, the fair thing to do was to disclose them together with whatever was the explanation. The points made by Mr Outhwaite would then be powerful points for Albingia to consider in deciding whether to write the treaty. While Arig makes the telling point that AIG’s and Copping’s disclosure goes back only to 1991, there is no reason to suppose that their 1989 or 1990 statistics were comparable to Arig’s.

Accordingly, the relevant question on the issue of inducement is whether Albingia has discharged the burden of proving that if the 1995 loss statistics going back to 1989 as set out in the table at [32] above had been disclosed, together with whatever fair explanation would also have been given of those figures, Mr Holzapfel would not have written the 1996 treaty or would have done so on different terms.

Mr Kimmins submitted on behalf of Axa that, if the 1989 and 1990 figures had to be disclosed, Albingia would certainly not have written the 1996 treaty. It would be, in his words, “game over”. I do not accept, however, that this is a fair reflection of the evidence. I begin by examining what Mr Holzapfel said about this and then consider some of the other factors which would or might have had a bearing on this question.

In his first witness statement Mr Holzapfel said that if Arig’s historic loss record as at 31 December 1995 had been included in the NMB offer fax of 17 July 1996, he would probably have instructed Ms Jerabek to decline the offer. That evidence is not difficult to accept in isolation, but it assumes that the figures were presented without explanation, which is not the relevant scenario. However, his witness statement went on to say that if Arig had told him that it had changed underwriter in 1991 and that the new underwriter wrote a different kind of book, this would not have made any difference: he would still have declined the treaty irrespective of this explanation. In his second statement Mr Holzapfel insisted that such an explanation would not have affected his usual approach to assessing risk, which “always involved a careful assessment, based on the documents presented to me by the broker, of the profitability of the account over time”. This, however, was an over statement. While I accept that examination of loss statistics was Mr Holzapfel’s usual practice, it was not invariably so. There is at least one case in evidence where he did not do so, which was a case where Albingia already had an existing relationship with the reinsured (AIG: see [51] above).

In cross examination Mr Holzapfel was first asked some general questions about how he would have reacted if he had been told by Mr Stephen Card, a broker at NMB whom he knew well, that Mr Hylander had implemented a conservative underwriting approach. His response was that he would have wanted to know more about this new approach, how it differed from what had gone before, and what it was intended to achieve, but that he would have accepted what the broker had said. When he was asked to assume that he had been told that before Mr Hylander joined Arig its underwriting had been somewhat directionless but that Mr Hylander had taken over in mid 1991, Mr Holzapfel said that he would still have wanted to see the company’s performance on its prior underwriting, but that such disclosure would have been the basis for a discussion (“put the facts on the table first, and then we can discuss how things will develop into a wonderful world going forward”). If he had been told that Mr Hylander was “focusing on” (not necessarily writing exclusively) shallow water risks, relatively benign weather areas, tested technology and narrow bore pipelines, that was something on which he would have placed reliance.

He was next asked for his reaction to the Arig reconstruction, which he had not seen before, and which went back only to 1991. I have already recorded his initial reaction at [106] above. He remained sceptical about this kind of hindsight exercise. Subsequently, however, when asked to assume that as well as the actual figures going back to 1991 the NMB presentation had included information that Mr Hylander had only taken over in 1991 after which he had implemented a new strategy, which if applied to the 1991 and 1992 years would have resulted in loss ratios below 30% (as in the Arig reconstruction), Mr Holzapfel said that Albingia would probably have written the risk on the same terms which it did.

Finally, the scenario was put to Mr Holzapfel that he had been told that Arig had written a number of energy construction risks in the past between 1989 and 1993, but had written hardly any in 1994 and 1995; he had been shown the loss records which had been provided to Swire Fraser (see [38] to [40] above); and he had been told that although the loss figures for some of the underwriting years shown in those records were very bad, in 1991 there had been a change of underwriter at Arig and the new underwriter was adopting a more rigorous and selective approach to the risks which he was prepared to accept, so that quite a number of the risks which the previous underwriter had been prepared to accept would probably not have been accepted under the new approach. That or something like it would in my view have represented a fair presentation of the risk. Mr Holzapfel’s answer was that he did not think that he would have entered into a relationship with Arig on the first loss treaty, because Arig had demonstrated that it was not able to manage the business properly over time. Although that represented Mr Holzapfel’s view when the scenario was put to him, and was to some extent only a tentative view, I find his answer difficult to understand. The scenario was that there had been a period of poor underwriting decisions with an underwriter following what turned out to be a bad policy, but that the situation had been managed by the introduction of a new underwriter who was successfully implementing a more rigorous approach.

The evidence summarised above represented Mr Holzapfel’s opinion as to what he would have done when the various scenarios were put to him. The weight to be given to his answers must take into account that, in some cases, he was reacting on the spur of the moment to hypothetical scenarios or documents not previously considered. I must also bear in mind that these answers required Mr Holzapfel to focus exclusively on the likely impact of past loss statistics upon him without also taking account of other factors which would or might have been relevant. I now turn to consider some of these.

In my judgment a relevant and important factor, which would have been an obvious element of any broke intended to present the Arig loss figures in a less damning light, was that Albingia had an existing relationship with Arig through the 1996 quota share treaty on which it had offered to write a 35% line. Even though this had been signed down to 19.57%, this remained a substantial commitment by Albingia to Arig’s energy underwriting. Indeed, the fact that the treaty had been oversubscribed in this way itself represented a vote of confidence in Arig by the market of which NMB would have been able to make some play.

Mr Holzapfel agreed that for the purpose of the Arig quota share treaty, he would have needed to know quite a lot about Arig as Albingia would be committing itself to follow Arig’s fortunes in the writing of energy business, although he also said that he had no discussions about Arig’s approach to the writing of energy risks before agreeing to take a line of 35%. Although the quota share presentation showed that Arig’s existing book consisted almost entirely of operating risks, the context was that even though construction risks were known to be higher risk, energy underwriters were moving away from operating risks in favour of construction risks as better terms and more attractive premiums were available for such risks. Mr Holzapfel knew this. In accepting a large line on Arig’s quota share treaty, Mr Holzapfel was (as he accepted) reposing a significant degree of trust in Arig’s ability to write appropriate construction risks. He described Arig in his evidence as a company which was established in the market, particularly in the Middle East, and a welcome addition to Albingia’s portfolio about which he had no misgivings. He must therefore either have made whatever inquiries he regarded as appropriate to satisfy himself that it was right to repose this trust in Arig and its underwriter or alternatively must have considered such inquiries to be unnecessary.

Although the existing relationship with Arig was not as important to Albingia as its relationship with AIG, it would nevertheless have been an important factor, in my judgment, if a fair presentation of the risk had been made.

As a general proposition, Mr Holzapfel’s evidence was that the level of deductibles in the original insurance policies written by the reinsured would be of obvious importance in determining whether to accept a first loss risk. That had also been his evidence in relation to avoidance of the Copping treaty where he described a deductible of £500,000 as “an absolute essential feature of writing the business and distancing ourselves from attritional claims” (see [49] above). In the case of the Arig treaty, however, nothing at all was ever said about the deductibles applied by Arig and Albingia never asked for this information. When asked about this apparently surprising omission, Mr Holzapfel said that he had to leave it to the ceding company to decide whether a risk was adequate in terms of pricing, conditions and deductibles. He added that with hindsight “it would have been nice” to have information about Arig’s deductibles, but that “it was sort of a carte blanche to say, Yes, we want to support them” and was “a shot in the dark, insofar we had no, say, information about deductibles or whatever”.

Mr Holzapfel was, I accept, doing his best to recreate his state of mind at the time, but his evidence before me seems hard to reconcile with what he said in the Copping case about the absolutely essential nature of information about deductibles. To my mind this illustrates the danger of focusing on an isolated factor (in the Copping case deductibles and in this case loss records) when attempting to answer many years after the event a hypothetical question about what an underwriter would have done if circumstances had been different. It illustrates also that apparently categorical evidence given in good faith may be, if not downright wrong, at least not the whole story. If that is true of deductibles, as in the present case it appears to be, it may equally be true of loss records. While there will of course be plain cases in which what the underwriter would have done is obvious, the decision will often be one which would have depended on a multitude of factors not all of which can be known.

Cash flow provides one illustration of the way in which recollection of the factors bearing on an underwriting decision can be faulty. Mr Holzapfel insisted that cash flow had not been an important factor in deciding which risks to accept. No doubt that represented his genuine recollection. But it was apparent from the contemporary documents that this was indeed a factor of some importance, albeit secondary to his assessment of the likely profitability or otherwise of the risk. It was referred to in numerous documents passing between Albingia and NMB. In the case of the 1996 first loss treaty with Arig, NMB would have been able (as it did in some other cases) to point to the cash flow advantage of the fact that Arig had already written a number of risks which were to be ceded to the treaty, thus providing immediate premium. This would not have been a major factor, but it would have been of some benefit.

From all this material, I draw the following conclusions.

First, despite his honesty as a witness, it is not safe to rely on Mr Holzapfel’s assertions as to whether or not he would have written the risk in the various scenarios about which he was asked. It is easy for an underwriter, focusing twenty years after the event on a single issue such as poor loss records and inevitably affected to some degree by hindsight, to express a genuine and firm opinion that if he had known about such records, he would never have written the business. However, the environment of a trial cannot faithfully recreate all the circumstances as they would actually have existed, at any rate when the trial takes place so long after the event that the underwriter in question has no recollection at all of the actual transaction.

Second, Albingia did in fact write the treaty despite the absence of any information at all about one factor, namely Arig’s approach to the deductibles in its original policies, which both in this action and elsewhere Mr Holzapfel has described as an important consideration. It follows that even matters which in general he would regard as essential may in some circumstances not be so.

Third, Mr Holzapfel regarded Arig as a high quality reinsured which in general he was keen to support. Whether or not he knew Mr Hylander (he said that he did not know of him, although that seems surprising) he must have satisfied himself that it was a well-managed company with competent underwriters when agreeing to write a substantial line of the quota share treaty.

Fourth, Mr Holzapfel was a fair minded man who would have been willing to listen to whatever explanation of poor loss records the broker (with whom he had a good working relationship) provided and to consider it on its merits. Indeed, he had done precisely this in the case of the Copping first loss treaty where there were poor loss records but where Albingia nevertheless agreed to take a line of 50%: see [49] above.

Fifth, NMB could (and would) fairly have explained that the poor results achieved by Arig were on risks written for the most part under a previous underwriter and that Mr Hylander, following his arrival in 1991, had adopted a more rigorous and selective approach and, moreover, an approach to energy risks in general which had satisfied Mr Holzapfel for the purpose of writing the quota share treaty.

Sixth, in addition, NMB could have made the point made in evidence by Mr Outhwaite that the particularly bad results achieved in the 1989 and 1990 years reflected not only the premium driven strategy of the previous underwriter but the adverse market conditions and disastrous losses sustained on risks written during those years by the market in general, which were not comparable with the situation as it was believed to be in 1996.

Seventh, although first loss reinsurance on energy construction risks was a particularly hazardous kind of reinsurance to write, and many underwriters were not willing to do so, Mr Holzapfel was not of their number. On the contrary, he wrote a number of such treaties and was not averse, therefore, to running such risks.

In all those circumstances there is in my judgment real doubt as to what Mr Holzapfel would have done if a fair presentation of Arig’s past loss records going back to 1989 had been made to him, together with the explanatory points which Arig or NMB on its behalf could fairly have made. It is not a plain case in which the answer is obvious. In the end I am not persuaded that it is more likely than not that he would have refused to write the treaty, or would only have done so on different terms. It follows that Axa’s case for avoidance of the 1996 treaty must fail.

If I had found that Axa was entitled to avoid the 1996 treaty, so that if a fair presentation had been made the 1996 treaty would not have been written or would have been written on different terms, it is obvious that Axa would also be entitled to avoid the 1997 treaty. If the 1996 treaty would not have been written at all, there would have been nothing to renew. If it would have been written on different terms, it might have been renewed in 1997, but would certainly not have been renewed on the terms on which it was in fact renewed.

However, on the basis that (as I have found) Axa is not entitled to avoid the 1996 treaty by reason of Arig’s failure to disclose its past loss statistics, it is not entitled to avoid the 1997 treaty on this ground either. The experts agreed that, once a treaty was in place, the reinsurer would not expect to be provided with pre-treaty loss statistics in subsequent years when it came up for renewal. What would matter at that stage would be the claims experience under the treaty. It is to that topic that I now turn.

Axa’s alternative case is that it was entitled to avoid the 1997 treaty for non-disclosure and misrepresentation in relation to the three incidents referred to at [81] to [89] above. By the end of the trial its case was that non-disclosure and misrepresentation in relation to the Clyde Petroleum incident was itself a ground of avoidance, and that (whether or not that was so) the three incidents in combination entitled it to avoid the 1997 treaty. It did not contend that either of the NPCC or Ras Laffan incidents was itself sufficient to found renewal.

The typical process which occurs when there is a casualty is that it will first be reported to the insurers under the original policy, who will decide whether to appoint loss adjusters; those adjusters will then investigate and report on such matters as the causes of the incident, the extent of any damage, whether there is coverage under the policy, and the likely quantum of any claim. The adjusters’ report will typically include a recommendation as to whether the insurers should place a reserve in their books. Generally the insurers will follow the loss adjusters’ recommendation.

Mr Outhwaite’s evidence was that these matters are left to the original insurers without involvement by reinsurers; that the reinsured will generally disclose to reinsurers on renewal those claims for which a reserve has been made; but that there would not normally be disclosure of incidents for which no reserve had been made or recommended but which might result in a future claim, as these would not influence the judgment of a prudent reinsurer in deciding whether to renew the policy; a fortiori there would be no need for disclosure of incidents where loss adjusters had positively advised against a reserve, at any rate for the time being. Mr Hartigan agreed that it was for the reinsured’s claims department to decide whether a reserve was needed and at what level.

The loss adjusters’ report would typically be addressed to the lead underwriter and would deal with the extent of liability under the original cover as a whole. If Arig had only a small line on the original policy it might be that any potential claim would be so minor as not to require disclosure to reinsurers.

As Axa submitted, there would be occasions when an incident not recorded as a reserve would be material for disclosure, and it is necessary to consider substance rather than form. Thus the position would or might be different, as Mr Outhwaite accepted, if a casualty was particularly significant even if it had not yet progressed to the stage of a reserve being recommended or entered in the reinsured’s books; or if it was known that there was a particularly large number of incidents being investigated which might in due course mature into claims even if there was as yet no recommendation for a reserve. No doubt there are other circumstances in which disclosure might be necessary in order for a presentation on renewal to be fair, bearing in mind that materiality is a question of fact and that the factors which may influence the judgment of a prudent underwriter will depend on all the circumstances. However, I would accept the evidence summarised above as representing a reasonable starting point so far as the materiality of claims experience on renewal is concerned.

I apply these principles in considering Axa’s case in relation to non-disclosure of the three incidents in issue.

In the case of the Clyde Petroleum incident the loss adjusters had recommended a reserve, albeit described as “precautionary”, but Arig’s claims department had failed to act on that recommendation by the time of the renewal. I would accept in those circumstances that the representation that “there is only one claim at this stage”, namely the other Clyde Petroleum incident referred to in NMB’s fax of 27 June 1997, was a misrepresentation. There were in fact two claims by this stage. I would accept also that the misrepresentation was material, as a prudent underwriter would be influenced by the claims experience under the first year of the treaty and the amount of the recommended reserve was not insignificant in the context of the treaty. For the same reasons, even without the representation, the incident was material and ought to have been disclosed.

However, if it had been disclosed, that would not necessarily have made any difference to Mr Holzapfel’s decision (assuming that it was his decision) to renew. Ultimately his evidence as to this was that this incident was “not a catastrophe and we might possibly have renewed, yes”. Although Mr Kimmins sought in re-examination to rescue this answer by suggesting (in, it has to be said, a somewhat leading way) that renewal might have been on different terms, I did not find this attempt impressive. Mr Holzapfel was well aware of the issue and, if he had meant to say that renewal would have been on different terms, he would have said so.

I find, therefore, that in relation to this incident Axa has not proved its case on inducement.

In the case of both the NPCC and Ras Laffan incidents the loss adjusters stated in terms that no reserve was recommended. Axa did not in the end contend that either of these incidents was material. I find that they were not. Mr Holzapfel said that the NPCC incident in isolation would not have stopped him from renewing and that the Ras Laffan incident in isolation “would not have triggered concern … would not have caused us to cancel or not to renew or whatever”.

That leaves Axa’s case that even if none of the incidents was itself sufficient to found avoidance, the combination of the three incidents was material and would have caused Mr Holzapfel not to renew or only to do so on different terms. I would accept in principle that a combination of such incidents none of which in itself was sufficient to found avoidance could require to be disclosed, but in my judgment this is not such a case. It was not at all clear that the NPCC incident would give rise to any liability, or in what amount, while the potential Ras Laffan liability was very minor. I do not accept that disclosure of the information which Arig had about these claims would have made any difference to Albingia’s decision to renew. As Mr Hartigan acknowledged, any prudent underwriter would expect there to be some loss reserves on an energy construction book after 18 months.

Arig contends that even if there was a time when Axa would have been entitled to avoid one or both of the treaties, it is now too late to do so. It accepts that if (contrary to its case) Axa did have a right to avoid, Axa only learned of the facts giving rise to that right in the course of the inspection carried out in October 2012 and that avoidance followed reasonably promptly thereafter. It does not contend, therefore, that Axa affirmed the treaties with knowledge of its right to avoid, but relies instead on equitable principles.

There are two different issues arising on this aspect of Arig’s case. The first is whether Axa lost the right to avoid the treaties. As to this, Arig’s case in summary is that the right to avoid is an equitable right; it is therefore subject by analogy to the limitation periods contained in the Limitation Act 1980 in accordance with section 36 of the Act, the applicable analogous provision being the six year time limit for contract claims set out in section 5; alternatively the right to avoid may be lost as a result of equitable principles such as laches, estoppel and change of position; and that the right to avoid has been so lost in this case as Arig has conducted its business since 1996 on the understanding (encouraged by Axa) that the treaties were valid, has used the monies paid by Axa in respect of claims in the course of its business, and has taken no steps to protect its position by making a claim against NMB.

In response, again in summary, Axa contends that avoidance is a self help remedy which is not to be equated with or analogous to a cause of action; that accordingly the Limitation Act does not apply, even by analogy; that its right to avoid is not subject to the equitable principles invoked by Arig; that in any event the application of the equitable principles on which Arig relies would depend on knowledge on the part of Axa of its right to avoid; and that Arig has failed to establish any prejudice as a result of the delay as it has had the benefit of the use of the claim payments, would still be able to restore the monies paid, and has failed to explain what claim it would have had against NMB.

The second issue is whether, on the assumption that Axa is still entitled to avoid the policies, Axa is also entitled to recover the claim payments expended by it (as distinct from having a defence to any action for outstanding claims). Here Arig contends that a claim to recover the payments made is time barred as those payments were made more than six years before this action was commenced. In response, Axa contends that the repayment of claims paid is not to be equated with a cause of action to which the Limitation Act would apply, but rather is to be regarded as a working out of the principle of restitutio in integrum following a valid avoidance; that in any event the running of time was suspended pursuant to section 32(1)(c) of the Limitation Act 1980 as the action to recover the monies paid out is for relief from the consequences of a mistake; and that until the date of the inspection Axa did not know of and could not with reasonable diligence have discovered the mistake in question, namely that it had a right to avoid for non-disclosure or misrepresentation.

It was common ground that on some at any rate of the legal submissions summarised above there is no direct authority. However, as I have found on the facts that Axa is not entitled to avoid either of the treaties on any of the grounds replied upon, the issues concerning loss of the right to avoid or to achieve restitution in integrum do not arise and anything I might say about them would necessarily be obiter. That being so, I do not propose to lengthen this judgment or to delay its delivery by purporting to decide those legal issues. I propose instead briefly to state my conclusions on the factual issues which would or might arise in relation to them, in case they should ever become relevant.

These conclusions are as follows. First, Axa did not know and had no reason to know, believe or suspect that it had or might have a right to avoid either of the treaties at any time prior to the inspection of October 2012. Second, it did not at any stage lead Arig to believe, nor is there any evidence that Arig did believe, that Axa would not exercise any right to avoid for misrepresentation or non-disclosure which it might subsequently discover that it had. Third, although Arig has used the claims monies paid by Axa in its business, that was a benefit to Arig rather than the reverse. There is no evidence which I accept that Arig would have acted differently in any material way if it had been aware of the possibility that Axa might seek to avoid the treaties in the event that it ever found out that it had a right to do so. Nor is there any reason to suppose that Arig would as a result of any such action (or at all) be unable to meet Axa’s claim for repayment. Fourth, Arig has not explained what claim it would have had against NMB. Fifth, Axa’s payment of claims was made in the reasonable belief that the treaties were valid and binding, with no right of avoidance. Finally, although Axa could at any time have exercised its right to call for an inspection, there is no reason to conclude that the exercise of reasonable diligence would have required it to do so at any time earlier than April 2007, that is six years before the commencement of the action in April 2013. In particular, it would be wrong to make any finding that reasonable diligence did require an earlier inspection in circumstances where this was not put to any Axa witness and Mr Bryan confirmed that Arig’s case was that Axa could have conducted an inspection at any time and not that it should have done so.

It seems probable that these findings of fact, all of which in my judgment are clear on the evidence, would render Arig’s legal submissions relating to delay academic even if there was at one time a right to avoid. That is an additional reason why it would not be useful to explore those submissions further in this case.

On the basis that the 1996 treaty is valid and binding an issue arises in relation to one risk on which Axa has paid a claim which, it contends, it was not obliged to pay. It says that it only discovered the relevant facts in the course of the October 2012 inspection and is entitled to recover the net amount of the payment, US $215,155.41, as having been made under a mistake. The issue arises out of the provision in the slip that “ARIG will not cede any layered contracts which represent less than 20% of the full contract value”.

The relevant provision in full read as follows:

“INFORMATION: Schedule of current risks as attached. (N.L.O.W.) Rating scale graph seen and noted.

Average line US$4,500,000 - US$5,000,000 based on maximum exposure as entered in the Reassured’s books.

Estimated Nett Premium Income:

Construction Account: US$330,000 - US$370,000

Operating-Account: US$ 40,000 - US$ 80,000

Anticipated Account will consist of 30 - 40 policies split:

USA based accounts - 20%

Middle East/Africa/Turkey - 35%

Asia including rest of the World - 45%

ARIG will not cede any layered contracts which represent less than 20% of the full contract value.

Details of 15 policies already written seen and noted.”

One risk written by Arig and ceded to the treaty was referred to as the Norsk Hydro risk. This covered the construction of a floating production platform and associated pipelines, subsea sections and related equipment, with a full contract value of NOK 4.5 billion which was insured in a number of layers. Accordingly, if this provision was effective to limit the risks which could validly be ceded to the treaty, Arig could not declare a line on a layer of the risk with a value of less than NOK 900 million (20% of NOK 4.5 billion).

The layer offered to Arig was a primary layer covering both physical damage (in section 1 of the policy) and third party liabilities (in section 2), each with a limit of NOK 750 million. Arig wrote a 5% line on both sections.

Axa contends that what I will call the layered contracts restriction had the effect of restricting the risks which Arig could cede to the treaty. Arig contends that the whole of the clause set out above, including the layered contracts restriction, is subject to the introductory words “Information (N.L.O.W.)” meaning that what follows was given by Arig for information only with no limitation or warranty.

The point is a short one. In my judgment Arig’s submission is clearly correct as a matter of construction. All other matters in the clause in question are evidently given for information only and are expressly stated not to constitute limitations on what could be ceded or to be the subject of warranties. I see no reason why the layered contracts restriction alone should be construed as having a radically different contractual effect. In common with the other matters dealt with in this clause, it was at most a statement of Arig’s intention as to the risks which it would cede. It was not suggested that Arig did not have this intention.

In the alternative, Axa invokes principles of estoppel, contending that there were occasions when Arig sought a special acceptance of a risk which did not comply with the 20% provision on the basis that it would otherwise be “in breach” of the relevant provision and that when Arig proposed to Albingia in June 1998 that the provision be deleted, NMB referred to it as “the self-imposed restriction”, which “currently prevents [Arig] from considering” otherwise attractive risks. Axa contends that these facts established a conventional basis for the parties’ understanding of the effect of the provision and that it would be inequitable to allow Arig now to resile from this shared conventional understanding.

I reject this alternative case. There is no evidence that Albingia’s conduct was affected in any way by any shared understanding of the effect of this provision. I can see no reason why it would be inequitable to allow Arig to reject Axa’s claim to recover monies which it paid on a claim many years ago which, on the true construction of the policy, it was obliged to pay.

Accordingly Axa’s claim to recover the Norsk Hydro payment fails. That makes it unnecessary to deal with two further issues relating to this claim. I will mention them only briefly.

The first such issue is whether, if the clause did amount to a contractual restriction, Arig was in fact in breach of it. That depends on whether the figure of 20% of the full contract value refers to the physical damage cover written by Arig (in which case the layer on which Arig wrote a line was NOK 750 million which was less than 20% of the full contract value of NOK 4.5 billion) or whether (as Arig contends) it is permissible to add together the physical damage and liability coverage in sections 1 and 2 of the original policy (in which case the overall limit of cover accepted by Arig was over 20% of the full contract vaue). I would if necessary accept Axa’s case on this for the sake of consistency. The full contract value refers to physical value and does not include any element of liability cover. The figure of 20% must likewise refer to the layer of physical damage cover.

The second issue with which it is strictly unnecessary to deal is the issue of limitation. Although Axa contends that it is entitled to rely on section 32(1)(c) of the Limitation Act 1980 in relation to the Norsk Hydro claim, there was very little attention paid to this particular issue at the trial and none at all in the parties’ closing submissions either written or oral. Accordingly I find that Axa has not discharged the burden of proving that it could not with reasonable diligence have discovered its mistake, if mistake there was, in paying the claim.

Arig counterclaims the sum of US $397,456.46 in respect of four claims made under the 1996 and 1997 treaties. On the basis that, as I have concluded, Axa is not entitled to avoid the treaties, the only defence is one of limitation. It is common ground that Arig’s cause of action under the treaties accrued when its liability to its own insured was ascertained by agreement, award or judgment: see Teal Assurance Co Ltd v W.R. Berkley Insurance (Europe) Ltd [2011] EWHC 91 (Comm), [2011] 1 Lloyd’s Rep IR 285 at [30] for a recent statement of this principle.

There are four claims. It is common ground as I understand it that two of them, by BHP and Nabors, are time barred and that a third claim, the smaller of two claims by Agip, is not.

There may still be an issue as to the fourth claim, also by Agip. Time did not allow consideration of this claim at the trial. If the parties are unable to agree whether this claim is time-barred, this issue will have to be determined.

For the reasons given above: