Royal & Sun Alliance Insurance PLC & Ors v Textainer Group Holdings Limited & Ors

The principal issue on this appeal is whether insurers, who have paid out under excess of loss policies, are entitled to a proportionate share of relevant recoveries subsequently made by the insured, or whether those recoveries are to be applied first to uninsured losses, applying the “top down” approach adopted by the House of Lords in Lord Napier and Ettrick v Hunter [1993] AC 713.

The respondents (“Textainer”) are companies in the Textainer group, one of the largest intermodal container lessors in the world. For the year commencing 1 October 2015 Textainer benefited from a container lessee default insurance programme, consisting of a US$5m retention, a primary policy of US$5m (in excess the US$5m retention) and five excess of loss policies providing cover up to US$80m in excess of the US$5m retention (“the Policies”). The respondents are, or represent, most of the insurers subscribing to the primary policy and the first to third excess policies (“the Insurers”).

On 31 August 2016 one of Textainer’s lessees, Hanjin Shipping Co. Ltd (“Hanjin”), applied to the Seoul Central District Court for receivership, an event entitling Textainer to an indemnity under the Policies against the loss of containers on-hire to Hanjin and not recovered within 183 days, uncollected rental during that period, and the costs of retrieving and/or repairing recovered containers. At that date Hanjin was in possession of about 113,000 of Textainer’s containers, held on a mixture of operating leases and finance leases. Hanjin was adjudged to be bankrupt on 17 February 2017, at which date Textainer’s losses were assessed as being about US$117.7m. That figure changed over time as all but 8,820 containers were recovered, counterbalanced by increasing costs and lost rental. For the purposes of this appeal, Textainer’s total losses are agreed to have been US$101,856,624, of which the net sum of US$478,265 was received from Hanjin Bankruptcy Trustee in early 2019 in respect of container rentals, paid in order to release a vessel arrested by Textainer.

The primary policy and the first to fourth excess policies paid out in full and, on 31 December 2018, the claim under the fifth excess policy was settled for US$25.1m (the limit under that policy being US$30m) so that the total sum paid to Textainer was US$75.1m, leaving uninsured losses in the region of US$21m (in addition to the retention of US$5m). As part of the settlement, the rights of subrogation under the fifth excess policy were transferred to Textainer. The rights of subrogation of insurers under the fourth excess policy (providing cover for US$10m) were also transferred to Textainer pursuant to a settlement agreement dated 26 June 2019.

In May 2019, at the invitation of Hanjin’s Bankruptcy Trustee, Textainer made a claim against Hanjin. By a settlement agreement dated 28 October 2019 Hanjin agreed to pay US$25,886,647.60 to Textainer in respect of its claim relating to operating leases only (“the Hanjin Settlement”). Between November and December 2019 Hanjin’s Bankruptcy Trustee paid Textainer US$14,706,880.84 of that agreed sum.

In these proceedings the Insurers claimed to be entitled to 39.3% of the sums received or to be received by Textainer pursuant to the Hanjin Settlement (having insured US$40m of Textainer’s total loss of US$101,856,624). Textainer resisted the claim, maintaining that it was entitled to retain all recoveries up to a total of US$56,378,359 (being its net loss of US$101,378,359 less the US$40m covered by the Insurers and the US$5m retention).

The Insurers’ claim was dismissed on 5 October 2022 by David Railton KC, sitting as a deputy High Court judge (“the Judge”), following a trial. In his reserved judgment dated 27 July 2022, the Judge held, in so far as relevant to this appeal, that:

The Insurers were granted permission to appeal that decision, challenging each of the Judge’s findings set out above. Textainer opposed the appeal and further contended, by its Respondent’s Notice, that even if the principle of averaging applied between the insured tranches of loss and the uninsured, the recoveries should nevertheless be applied “top-down” as between the layers of insurance. As the rights of subrogation in respect of the top two layers had been transferred to Textainer, Textainer submitted that the Judge had in any event been right to dismiss the claim.

The Judge explained that the detailed structure of the insurance was as follows:

The Policies were governed by English law. Each of the excess policies incorporated the terms of the primary policy, save that the fifth excess policy excluded loss of earnings, a factor in the level of settlement of Textainer’s claim in respect of that layer. Each of the excess policies also included a “no drop down” clause, providing that the policy would only ever provide cover for the identified tranche of excess loss, and would not provide cover for the layers below in any circumstances.

The primary policy included the following provisions:

It was common ground that the doctrine of subrogation allows an insurer who has indemnified an insured to “take advantage of any means available to the insured to extinguish or diminish the loss for which the insurer has indemnified the insured”: MacGillivray on Insurance Law 15^th ed. 2022 para 22-001. By reason of that right, any recoveries received by the insured will inure to the benefit of the insurer with a view to diminishing the loss which the insurer has paid and indemnified: MacGillivray paras 22-005, 22-067.

In the present case, recoveries from the Hanjin Settlement had the effect of reducing Textainer’s total losses, but those losses nevertheless remained well above the upper limit of the cover provided by the Insurers. On the face of matters, those recoveries did not therefore engage the Insurers’ undoubted right to be subrogated to recoveries made in respect of insured losses. This mirrors the position as it would have been if the recoveries had been made before Textainer had claimed under the Policies: the recoveries would have reduced Textainer’s losses, but however those recoveries were allocated, losses would still have exceeded the cover under the Policies when a claim was made.

The above approach also reflects that of the House of Lords in Napier. In that case Lloyd’s names, including members of the Outhwaite syndicate, held stop loss insurance, covering losses made in a year of account above an excess up to a specified limit. The names sustained underwriting losses for 1982 and their claims were met by insurers. The names subsequently recovered damages from the managing agents of the syndicate in respect of the losses. The issues included whether the names were entitled to be fully indemnified against their losses (including in relation to losses below the excess) before the stop loss insurers were entitled to be reimbursed for the sums paid out to the names under the policy. The House of Lords held that the losses should be applied “top down”, illustrated by an example in which it was assumed that for the 1982 year of account a hypothetical name suffered a net underwriting loss of £160,000, that the excess was £25,000, and that the limit was £100,000. After the stop loss insurer had paid £100,000 to the name, damages of £130,000 recovered from the managing agents were attributable to the net loss of £160,000 suffered by the hypothetical name. Lord Templeton explained the approach to be taken as follows at p.730A:

Napier was followed and the top down approach was applied by Langley J in Kuwait Airways Corporation v Kuwait Insurance Co S.A.K [2000] 1 Lloyd’s Rep 252. The Judge helpfully summarised the facts of that case at [87] as follows:

Langley J, at p.261, recorded the arguments as follows:

Langley J accepted KAC’s submissions, stating at p.261 as follows:

The Insurers contended that the Policies in the present case were distinguishable from the stop loss policies considered in Napier. The stop loss policies, they argued, applied to a single (or “unitary”) financial loss for a specified period of underwriting accounting ascertainable on a single date, and covered the excess of that loss above a specified amount. In contrast, the Policies did not insure a unitary loss, but covered the physical loss or damage to individual containers and related costs/loss of earnings as and when those losses were incurred, eroding first the retention, then the layers of cover, one by one. Kuwait Airways, they argued, was distinguishable because all the losses occurred simultaneously. Alternatively, it was wrongly decided.

Therefore, the Insurers submitted, recoveries by Textainer under the Hanjin Settlement were not to be regarded simply as a reduction in the total loss it had sustained (as was necessarily the case in Napier), but should be applied to the specific losses suffered by Textainer where it was possible to do so. It was possible to do so in the present case, according to the Insurers, because the Hanjin Settlement was pro rata settlement of Textainer’s losses in respect of operating leases (being 40% of Textainer’s claim in respect of those leases), and it should be inferred that losses in respect of operating leases and finance leases were suffered evenly and regularly over time. The result, the Insurers contend, is that the Hanjin Settlement recoveries should be applied proportionately across the layers, so that 39.3% is payable to them.

In order to succeed on that line of argument, the Insurers needed to succeed on both its legal argument that the top down approach was not applicable and its factual argument that the Court should infer that losses under the operating leases and finance leases were suffered evenly and regularly.

In the alternative, the Insurers argued that Napier could be distinguished because the Policies are marine container insurance, thereby engaging section 81 of the 1906 Act, which provides:

The Insurers contended that Textainer had underinsured its containers, demonstrated by the fact that its losses exceeded the upper limit of the Policies, with the result that Textainer was deemed to be its own insurer in respect of the uninsured losses. It followed, they said, that the principle of averaging applied, and the losses fell to be shared proportionately between the Insurers and Textainer, as established in The Commonwealth [1907] P 216.

The Judge first expressed the view (at [85]) that the differences between the stop loss policies considered in Napier and the Policies were not as “sharp” as the Insurers suggested. Whilst it might well be that payments under the Policies could be connected to a particular loss and a particular recovery, that was also possible in the case of the stop loss policies in Napier.

But in any event the Judge did not regard such differences as there were in the formulation or ascertainment of the aggregate loss to be indemnified as undermining the rationale for applying the top down approach in Napier, pointing out at [86] that the approach:

At [87]-[94] the Judge considered Kuwait Airways and rejected the Insurers’ attempts to distinguish that case on the facts. At [94] the Judge concluded that in Kuwait Airways:

The Judge further recorded, at [96] and [97], that the effect of the Insurers’ primary case (that recoveries should be applied against losses with which they could be connected), was that insurers on higher layers (whether or not they had paid) would not get the benefit of the reduction in the insured’s aggregate loss effected by recoveries applied to the lower layers, or to the insured’s retention. The result would also be that the effective limit of aggregate cover provided to the insured would be reduced. The Judge went on to say at [98]:

The Judge supported his conclusion at [99] by referring to the fact that the Insurers’ approach would produce a different result depending on whether recoveries were made before or after payment by the insurers:

At [100] the Judge further pointed out that the Insurers’ approach would risk Textainer not receiving a full indemnity, contrary to a fundamental principle of insurance law articulated by Brett LJ in Castellain v. Preston (1883) 11 QBD 380 at p.386.

On this appeal, the Insurers argued that the Judge had failed to recognise the fundamental distinction between the case of a single or unitary loss (such as the net underwriting loss at 36 months in Napier) and the case of multiple losses, such as the present case. In the former case, it was plain that subsequent recoveries would reduce that single loss “top down”, but where there were multiple losses of different items of property at different times, recoveries in respect of those specific items not only could but must be allocated to the insurer who had indemnified against their loss.

The Insurers gave an example of where 1000 containers (“the A containers”) were lost and the insurers under the primary policy paid out in respect of that loss, and subsequently a further 1000 containers (“the B containers”) were lost and were indemnified under the first excess policy. If there was a later recovery in respect of the A containers, that recovery must be applied to the primary policy, not the first excess policy, because the insurers under primary policy had indemnified against the relevant loss and, applying the basic principle of subrogation, the recoveries in that regard enure to the benefit of the insurer. The insurers under the first excess policy had not indemnified the relevant losses: to pay the recoveries to them on a “top down” basis would be unjustifiable.

The Insurers further contended that the fact that the Policies contained provisions providing the period over which losses were recoverable (for example, rental for 183 days after a Date of Occurrence) and for aggregate limits says nothing about the date on which losses actually occur. Neither do they undermine the fact that losses, as they occur, erode the retention, then the cover under the primary policy and then each excess layer in turn, each loss above the retention being paid by a specific one of the Policies.

In my judgment, however, the Judge was right to identify that the true nature of the cover being provided by the Policies was against particular layers of loss, and that the manner in which losses were determined and aggregated in that context was not an important factor. The nature of the cover required that the Policies “pay up and recover down” (as the approach was summarised by Staughton LJ in the Court of Appeal in Napier, p. 24). Otherwise, if recoveries were not applied “top down” but proportionately to the insured layers as well as to the uninsured losses above the limit of cover:

Far from the stop loss policies in Napier being distinguishable from the Policies in the present case, providing excess of loss cover, Arnould: Law of Marine Insurance and Average 20^th ed. para 31-465 regards the application of the top-down principle to those policies as mirroring the position under excess of loss policies:

That approach was taken by Langley J in Kuwait Airways, a decision which was neither distinguishable (for the reasons given by the Judge) nor wrong. Indeed, in my judgment it was plainly right, giving KAC the benefit of the cover for which it had bargained and declining to produce a result which would have been different had the eight recovered aircraft not been taken in the first place.

In my view the Insurers’ argument is based on an overly formalistic and largely theoretical approach to the allocation of individual losses and the notional connection of recoveries to those losses, when no such process or requirement is apparent from the terms of the Policies themselves, nor from the manner in which Textainer and the Insurers dealt with the claim. As Textainer pointed out, cover is not really provided in relation to containers, most of which will eventually be recovered (as in the present case), even if they are initially treated as giving rise to a constructive total loss after 183 days. The real subject of the insurance is the multiple strands of lost rental, costs and expenses which will be ongoing and intertwined leading up to the date on which they will be ascertained and aggregated. The simplified example of two sets of 1000 containers given by the Insurers does not represent the reality of the cover or the ascertainment of losses.

In my judgment, rather than the rigid (and largely theoretical) analysis suggested by the Insurers, a broader approach is required to conform most closely with the underlying rationale of subrogation, as recognised by Rix J in a preliminary decision in Kuwait Airways, reported at [1996] 1 Lloyd’s Rep 664, 695. That approach, as recognised in the authorities and the leading textbook, is that in the case of excess of loss policies, recoveries should be applied top down.

It follows that, largely for the reasons given by the Judge, I would reject the challenge to the application of the top down principle to the recoveries from the Hanjin Settlement.

If I am right that recoveries from the Hanjin Settlement fell to be applied “top down” as a matter of principle, the further factual issue does not require determination. Nevertheless, I will briefly set out my reasons why I consider that the appeal would also fail on this issue.

The Judge accepted at [102] the Insurers’ contention that (if it were to be a relevant exercise to undertake), each individual loss would contribute to the erosion of the retention and then the limits of the primary and excess policies in the chronological order in which it was suffered. The Judge further accepted that it would be a matter for evidence as to when each particular loss was suffered, and that it might, in appropriate cases, be permissible to draw inferences from the available material as to this.

After further recognising at [103] that the recovery pursuant to the Hanjin Settlement was in respect of operating lease losses only, the Judge rejected the Insurers’ argument that it should be inferred that those losses, and those in respect of finance leases, were all suffered evenly and regularly throughout the relevant period:

The Insurers criticised the Judge for engaging in what they characterised as speculation, the Judge having postulated that there may have been a significant difference between the timing and pattern of losses under the two types of lease when there was no evidence to suggest that was the case. In order to reach the conclusion that it was possible that none of the recovery in respect of the operating leases should be applied against losses covered by the Insurers, the Judge was envisaging a scenario in which almost all the losses in respect of finance leases occurred before the losses in respect of the operating leases. That was unrealistic and should have been discounted as a practical possibility.

Whilst accepting that they bore the burden of proof, the Insurers maintained that the starting point should have been a presumption of equality and regularity, a presumption which was not rebutted. Indeed, the Insurers argued that the Schedule submitted to the claim against Hanjin (showing losses as at 16 February 2017) gave support to that assumption, demonstrating that rental income had been lost month by month from September 2016 to February 2017 in respect of both operating leases (totalling just over US$8m) and finance leases (totalling just over US$11m) and that recovery costs (which could not be broken out by month because most units had multiple expenses such as storage, transport or handling) were comparable over that period (about US$8.5m for operating leases and about US$9.6m for finance leases). The Insurers also asserted that an examination of the spreadsheets would not have assisted in the exercise because they were concerned with quantum of losses, grouped together, not the date losses occurred, but is unclear how that assertion can be accepted given that the spreadsheets were not in evidence below or on this appeal.

Contrary to the Insurers’ contention, however, there is no presumption that losses occurred evenly and regularly, it being necessary to infer any such pattern from the evidence. The correct approach was explained in Arnould para 33-25 as follows:

In Municipal Mutual there was “sparse” evidence available as to when thefts had occurred during the three year period, so the court accepted that, on the balance of probabilities, they had occurred “on a straight line basis”. In the present case, in contrast, there was a wealth of material available as to when losses had occurred in the form of the spreadsheets submitted by Textainer and the underlying data. To the extent that such material would have been insufficient for the Insurers to make their case, it was open to them to seek further disclosure on the issue of the timing of the losses suffered.

I consider that, where a party with the burden of proof on an issue elects not to adduce available evidence to support its case, the court is entitled to decline to draw inferences which might otherwise have been drawn from the limited material which is adduced in evidence. That may be analysed, as the Judge did in this case, as a failure of the party to establish the factual premise of its case, or it might be viewed as a strong countervailing factor which makes it inappropriate to draw the inference in question. In any event, allowing a party to prove a case by inference, when it has not adduced available evidence on the issue, would encourage parties to mount inferential cases that they know or suspect the evidence, if adduced, would not support, and would risk injustice. I do not say that a Judge could never be persuaded to draw an inference in the absence of evidence that could have been adduced, but I do not consider that the Judge in this case can be criticised for refusing to do so.

In any event, the Judge was also entitled to say that he could not discount the possibility that losses on finance leases largely occurred before those in respect of operating leases. Whether there was any relevant distinction between the two types of lease in this regard was a “known unknown”. It is plain that Hanjin’s Bankruptcy Trustee made some distinction for an unknown reason, as only the claim in relation to operating leases was settled. Textainer suggested, by way of example, that Hanjin may have abandoned containers on finance leases first due to the capital element in rental payments. Such matters are not simply speculation, but a demonstration that it is not safe to draw the inference proposed by the Insurers.

For the above reasons, the Judge was right to hold (subject to the question of averaging addressed below) that Textainer’s recoveries from the Hanjin Settlement fell to be applied top down.

The Judge did not doubt the potential distinction between non-marine insurance (as in Napier) and marine insurance (arguably effected by the Policies), but gave short shrift to the argument that the present case represented under-insurance for the purposes of section 81 of the 1906 Act. At [95] the Judge stated that the concept of under-insurance:

The Insurers argued that the Judge had failed to recognise that, whilst structured as layers, the Policies constituted property insurance, just like insurance of a vessel. If losses were suffered in relation to a vessel in excess of the amount for which it was insured, that would constitute under-insurance and the insurer would only to liable for a pro rata share of the loss. The same applies, the Insurers submitted, where losses in respect of the containers insured by the Policies exceeded the upper limit of cover. Staughton LJ’s comment cited by the Judge was an obiter aside, without any analysis, and should not be followed.

In my judgment, however, the concept of undervaluation is simply not engaged in relation to layers of insurance. In the case of the insurance of a ship, for example, where the whole vessel is insured and the insurer is liable for any damage it suffers, insurance at less than the value of the ship can readily be seen to be under-insurance by value, exposing the insurer to the same risk (up to the limit of cover) as if the insurer had insured the full value of the vessel. On the other hand, where cover is in relation to a defined portion of loss, the insurance cover by definition matches precisely the value of that which is insured and the insurer is not exposed to a risk greater than the cover provided. Staughton LJ’s dictum in the Court of Appeal in Napier seems to be a succinct recognition of the above analysis.

Further, and as Textainer pointed out, whereas the insurable value of a ship is readily ascertained by reference to section 16(1) of the 1906 Act, insurance which includes the costs of recovery and repair of containers and loss of rental falls within section 16(4) (insurance on any other subject-matter) and insurable value is “the amount at the risk of the assured”. In the present case, typical of excess of loss insurance, Textainer’s risk was undefined and indefinite, demonstrating why the principle of averaging is inapposite for excess of loss insurance.

As Textainer further pointed out, if averaging due to undervaluation did apply in relation to the Policies, the Insurers would only have been liable for a proportion of the amount of cover they had underwritten in the first place, but they in fact paid in full. The attempt belatedly to introduce the concept of averaging to obtain a proportionate share of recoveries (rather than to deny full liability) is misconceived and further demonstrates the fallacy of the approach in this context.

For those reasons I reject the challenge to the Judge’s finding in respect of section 81 of the 1906 Act. In view of that conclusion, the issue raised by the Respondent’s Notice does not arise for consideration.

I would dismiss the appeal.

I agree.

I also agree.