Various Eateries Trading Limited v Allianz Insurance Plc

The judge summarised the court’s approach to aggregating provisions in insurance policies in his judgment in Stonegate, and adopted that summary in his judgment in the present case. As there was no dispute about it, I reproduce it here as a valuable summary of the principles, although it will be necessary in due course to examine more closely the cases dealing with remoteness:

“78.

What determines this issue is the proper construction of the relevant clauses of the Policy itself. There is, however, a considerable amount of authority relating to aggregating provisions, and to how they should in general be approached, and it is helpful to begin by considering such guidance.

79.

The function of aggregation clauses, as was said by Rix LJ in Scott v Copenhagen Reinsurance Co (UK) Ltd [2003] EWCA Civ 688, [2003] Lloyd’s Rep IR 696 at [12] is 'to police the imposition of a limit by treating a plurality of linked losses as if they were one loss'.

80.

The application of an aggregation provision can, depending on the nature of the perils and losses which occur, benefit either the insured or the insurer. They are therefore to be construed in a balanced fashion without a predisposition towards a narrow or a broad interpretation: Lloyds TSB General Insurance Holdings Ltd v Lloyds Bank Group Insurance Co Ltd [2003] UKHL 48, [2003] 4 All ER 43 at [30] per Lord Hobhouse; AIG Europe Ltd v OC320201 LLP sub nom AIG Europe Ltd v Woodman [2017] UKSC 18, [2017] 1 WLR 1168 at [14] per Lord Toulson JSC; Spire Healthcare Ltd v Royal & Sun Alliance Insurance Ltd [2022] EWCA Civ 17, [2022] Bus LR at [19] per Andrews LJ.

81.

'[T]he choice of language by which the parties designate the unifying factor in an aggregation clause is … of critical importance': Lloyds TSB loc. cit. at [17] per Lord Hoffmann.

82.

There are a number of well-known and frequently-encountered types of unifying factors which are used in aggregation clauses. In particular, parties often choose either a 'cause' or 'originating cause' unifying factor, or an 'occurrence' or 'event' unifying factor. In Axa Reinsurance (UK) plc v Field [1996] 2 Lloyd's Rep 233, Lord Mustill said (at 239), in relation to a provision which aggregated matters 'arising out of one event':

'The contrast is between "originating" coupled with "cause" in Cox v Bankside Members Agency Ltd [1995] 2 Lloyd's Rep 437, and "event" in the present case. In my opinion, these expressions are not at all the same, for two reasons. In ordinary speech, an event is something which happens at a particular time, at a particular place, in a particular way. … A cause is to my mind something altogether less constricted. It can be a continuing state of affairs; it can be the absence of something happening. Equally, the word "originating" was in my view consciously chosen to open up the widest possible search for a unifying factor in the history of the losses which it is sought to aggregate. To my mind the one expression has a much wider connotation than the other.'

83.

The courts have considered that the term 'occurrence' is virtually or entirely synonymous with 'event'. In an aggregating provision where the unifying factor is an 'occurrence', the Court will need to determine whether there was a relevant 'occurrence' to which matters can relate. The meaning of the word 'occurrence' 'must take its meaning finally from the surrounding terms of the policy including the object sought to be achieved': Mann v Lexington Insurance Co [2001] 1 Lloyd’s Rep 1 at [36] per Waller LJ.

84.

In considering whether there has been a relevant 'occurrence' 'the matter is to be scrutinised from the perspective of an informed observer in the position of the insured'. In making that assessment, an important aspect will be 'the degree of unity in relation to cause, locality, time, and, if initiated by human action, the circumstances and purposes of the persons responsible' (as it was put by Michael Kerr QC in the Dawson's Field Award, which is quoted in Kuwait Airways Corp v Kuwait Insurance Co. SAK ('KAC v KIC') [1996] 1 Lloyd's Rep 664 at 685-6).

85.

In KAC v KIC Rix J further said at 686, correctly in my judgment:

'In assessing the degree of unity regard may be had to such factors as cause, locality and time and the intentions of the human agents. An occurrence is not the same thing as a peril, but in considering the viewpoint or focus of the scrutineer one may properly have regard to the context of the perils insured against.'

86.

The so-called 'unities' are not to be applied mechanistically: they are 'merely an aid in determining whether the circumstances of the losses involve such a degree of unity as to justify their being described as ‘arising out of one occurrence’, Simmonds v Gammell [2016] EWHC 2515 (Comm), [2016] 2 Lloyd's Rep 631 at [29] per Sir Jeremy Cooke.

87.

Whether any and if so what causal link is required between the unifying factor and the losses must depend on the linking words used. Typically aggregation clauses require a significant causal link. In Scott v Copenhagen Re the relevant clause was 'arising from one event'. It was accepted, in line with Caudle v Sharp [1995] 4 Lloyd’s Re LR 389, that 'arising from one event' did not necessarily import a requirement of proximate causation. It nevertheless required a significant causal connection. Rix LJ said this, at [68]:

'… Nevertheless, it seems to me ultimately to be inherent in the concept of aggregation ("arising out of one event") that a significant causal link is required. … A plurality of losses is to be regarded as a single aggregated loss if they can be sufficiently linked to a single unifying event by being causally connected with it. The aggregating function of such a clause is antagonistic to a weak or loose causal relationship between losses and the required unifying single event. This is the more easily seen by acknowledging that, once a merely weak causal connection is required, there is in principle no limit to the theoretical possibility of tracing back to the causes of causes. The question therefore in my judgment becomes: Is there one event which should be regarded as the cause of these losses so as to make it appropriate to regard these losses as constituting for the purposes of aggregation under this policy one loss?'

88.

There is also usually a distinct requirement of lack of remoteness between the aggregating event and the losses. Thus in Caudle v Sharp at 394, Evans LJ, with whom Rose and Nourse LJJ agreed, said:

'In my judgment, the three requirements of a relevant event are that there was a common factor which can be properly described as an event, which satisfied the test of causation and which was not too remote for the purposes of the clause.'

89.

In Scott v Copenhagen Re, Rix LJ referred to the requirement of a lack of remoteness recognised in Caudle v Sharp, and referred to it as a 'tool to limit the otherwise infinite reach of the workings of causation', and 'a legal tool to separate out relevant from irrelevant causes' (at 713). As a result of its use, the court will look for a 'nearer and more relevant cause than for a more distant one' (ibid).

90.

The issue of whether losses can properly be aggregated and if so around what event or cause is to be answered by an exercise of judgment based on all the relevant facts and the purpose of the clause. Rix LJ in Scott v Copenhagen Re said, at [81]:

'… Are the losses to be aggregated as all arising from one event? That question can only be answered by finding and considering all the relevant facts carefully, and then conducting an exercise of judgment. That exercise can be assisted by considering those facts not only globally and intuitively and by reference to the purpose of the clause, but also more analytically, or rather by reference to the various constituent elements of what makes up one single unifying event. It remains an exercise of judgment, not a reformulation of the clause to be construed and applied’.”

In the present case Allianz had pleaded as many as 18 different candidates as constituting an occurrence giving rise to an SBIL. The judge grouped these into five broad groups as follows: (i) what may be described as “virology arguments” including as to the initial mutation(s) which led to the progenitor virus and as to the MRCAs of the virus or identified lineages; (ii) arguments in relation to the first transmission or outbreak of the disease in Wuhan; (iii) the “first or first relevant transmission” in the Vicinity, “introduction to the Vicinity” or “outbreak” in the Vicinity cases; (iv) a case as to the continuation or spread of the pandemic within the Vicinity; and (v) a case as to government actions.

The judge considered each of these groups in turn, applying the test of what would have been known to an informed observer in the position of the insured in early 2020, and noting that the arguments overlapped significantly with those which he had considered in the Stonegate case. As to the first two groups, and as in Stonegate, he concluded (at [92]) that

“(i)

the initial human infection(s) could be described as a single occurrence; (ii) the ‘outbreak’ of the disease in Wuhan thereafter was not a single occurrence; (iii) none of the ‘virology’ options in that case or what I have identified as the virology arguments in this case would have been regarded as a single occurrence”.

It appears, therefore, that the judge was distinguishing between “the initial human infection(s)” (or what he described in Stonegate as the first zoonotic or animal to human infection(s)) and the “outbreak” of the disease in Wuhan “thereafter”, by which he meant the subsequent occurrence of human to human infections on a wider scale.

Thus the only single occurrence which the judge found to have occurred in Wuhan was the initial infection of at least two and possibly as many as 15 human beings in late November and December 2020 (see the Stonegate judgment at [134]). The judge concluded at [93] in the present case that this occurrence had a sufficient causal relationship with VE’s losses to satisfy the causal requirement implicit in the definition of an SBIL, which was that “only a relatively loose link was required” (see the Stonegate judgment at [113]). However, the judge concluded at [94] that these initial infections were too remote. His reasoning for this last conclusion is best seen from his judgment in Stonegate, to which he cross-referred:

“154.

Here, that which I have accepted was an occurrence was remote from the losses in a significant number of respects. It was geographically remote. While it can of course be said that, modern communications being what they are, there is the possibility of rapid spread of a disease internationally, nevertheless the occurrence was very distant from the Territorial Limits of the Policy. It was temporally remote. The relevant losses only started to be incurred some months after the occurrence. It was also causally remote. It depended on a very large number of intermediate events and occurrences, involving first the establishment of the disease in the human population in China, secondly the spread of the virus to the UK, and thirdly the governmental and public response to the virus, to have an effect on Stonegate's business. The 'nearer and more relevant causes', to use Rix LJ's language in Scott v Copenhagen Re were those which more directly occasioned the losses, in particular the governmental response to the arrival and threat of spread of the virus in the UK.”

As to the introduction of Covid-19 into England or the UK, the judge concluded that there was no relevant single occurrence. There was no single introduction of Covid-19 to the UK, but rather there were multiple introductions, which were genetically distinct and led to onward chains of transmission. The epidemic in the UK (or the losses arising from it) could not be said to be “connected with” any one such introduction. Nor was the “initial outbreak” (i.e. “the early cases of the disease in the Vicinity”) a single occurrence. As the judge put it at [98], “there were multiple cases, occurring at different times in different places, no single one of which can be said to have been causatively related to all the cases in the Vicinity which led to governmental action and public reaction”. If there was a single occurrence, it was too remote. For the same reasons, the further spread of the disease was not a single occurrence either.

However, the judge did accept that the government action taken in response to the spread of Covid-19 did amount to a number of single occurrences which (implicitly) were not too remote:

“101.

This case was put as a fallback by both parties. I nevertheless consider that there clearly were occurrences consisting of governmental action with which the alleged losses were connected. These were more extensively canvassed in the Stonegate v MS Amlin and Greggs v Zurich actions than in this. The facts of this case are simpler than those relevant to those cases because it is only the actions of the UK Government applicable to England which are relevant. For reasons more fully expressed in the Judgments in those cases, I reject the idea that there was a single government response, which amounted to one occurrence, covering everything done in response to SARS-CoV-2. I do, however, accept that there was a single occurrence consisting of the Government's decision made on 16 March 2020 to instruct people to avoid social venues and the instructions given on that date to that effect. I would also accept that there was an occurrence on 20 March 2020, when restaurants were instructed to close. Equally, I would accept there to be an occurrence constituted by the announcement and implementation from 24 September 2020 of early closing and other restrictions on restaurants. Further, should they be relevant, I would consider that the bringing into force of the three-tiered system on 14 October 2020, and the imposition of the second lockdown from 5 November 2020, were capable of being relevant single occurrences. I do not accept that there were separate occurrences when measures were renewed, immaterially changed, or relaxed.”

The judge rejected VE’s submission that aggregation should be on a per Insured Location basis, finding this inconsistent with the terms of the policy.

Mr Charles Dougherty KC for Allianz challenged the judge’s conclusions concerning aggregation. In brief outline, he submitted that the “initial outbreak” in Wuhan on which Allianz relied was the initial animal to human infection(s) which the judge had found to constitute a single occurrence, but that the judge had been wrong to decide that this was too remote. The aggregation clause permitted a causal link (“in connection with”) which was relatively weak or loose, as the judge had accepted, and it followed from this that the concept of remoteness likewise should be applied fairly loosely. The judge had failed to stand back and consider the position overall. If he had done so, he would have concluded that the initial outbreak of Covid-19 in Wuhan was the most relevant or meaningful explanation of VE’s losses, and accordingly was not too remote to qualify as an aggregating occurrence. Alternatively, if the judge was right to regard the “initial outbreak” in Wuhan as something later and more widespread than the initial animal to human infection(s), there were nevertheless events in Wuhan, such as the initial human to human infection(s) which qualified as a single occurrence for the purpose of the definition of an SBIL. While it might not be possible to identify the particular individuals concerned, or the precise time and place at which they had been infected, inevitably there had been such initial infection(s) and the time and place where these had occurred could be identified with reasonable precision, so as to satisfy the requirement of an event which happens “at a particular time, at a particular place, in a particular way”, to use Lord Mustill’s phrase in Axa Reinsurance v Field.

As an alternative to the Wuhan case, Mr Dougherty submitted that the introduction of Covid-19 into England or the UK qualified as a single occurrence. Again, there must have been such an initial introduction marking the start of the pandemic in the Vicinity, which was sufficient to constitute a single occurrence, even if it could not now be identified, and that introduction was “in connection with” VE’s loss. For that purpose, there is no need to show a chain of causation from the first introduction to the loss claimed: it is sufficient that there is a relatively loose connection. Moreover, that first introduction could not be regarded as too remote. It was by definition a “nearer and more relevant cause”, to use Lord Justice Rix’s phrase in Scott v Copenhagen Re, than events in Wuhan.

In response, Ms Leigh-Ann Mulcahy KC for VE took two preliminary points. The first was that the exercise which the judge had undertaken involved an evaluative judgment following a factually complex trial, and that this court should not interfere with his conclusion in the absence of any error of principle in his approach. The second was that, although the judge had found that the initial animal to human infection(s) in Wuhan constituted a single occurrence for the purpose of the definition of an SBIL, this was not a case which Allianz (unlike the insurers in the Stonegate and Greggs cases) had pleaded or advanced at trial. Rather, Allianz’s pleaded case on the “initial outbreak” had been concerned with the further spread of the disease after the early human to human transmissions. In any event, however, Ms Mulcahy submitted that the judge had been right to conclude that the initial animal to human infection(s) (and, if relevant, events in Wuhan generally) were too remote for the reasons which he gave. As to the introduction of Covid-19 into England or the UK, the judge had been right on the facts to find that this was not a single occurrence, but a series of multiple occurrences, and in any event that it was too remote.

The use of remoteness as a principle of the construction of an aggregation clause in an insurance policy appears to have first emerged in Caudle v Sharp [1995] 4 LRLR 433.

In that case the reinsured sought to recover in respect of claims which had been payable under the original insurances as a result of the negligent failure of the underwriter to conduct the necessary research and investigation into the underlying problem of asbestosis which had given rise to the claims. The reinsurance policy contained a limit of liability “each and every loss” and provided that this meant each and every loss or series of losses “arising out of one event”. Arbitrators held that the underwriters’ negligence amounted to an aggregating event and that conclusion was upheld by Mr Justice Clarke. However, the Court of Appeal disagreed.

The reasoning of Lord Justice Evans (with whom Lord Justice Rose agreed) was that the concept of remoteness was inherent in the idea of a loss “arising out of” an event:

“The losses or series of losses envisaged by the clause must have ‘arisen out of’ one event, which in this context straightaway implies some causative element and some degree of remoteness, or lack of remoteness, which must be established in the circumstances of the particular case. In my judgment, the three requirements of a relevant event are that there was a common factor which can properly be described as an event, which satisfied the test of causation and which was not too remote for the purposes of the clause.”

Lord Justice Evans held that the claim to aggregate failed at the first hurdle. The underwriter’s negligent state of mind was not an event and there was no single event, but rather a series of separate events on every occasion when he negligently entered into an insurance contract. Even if there had been a single event, it would have been “too remote even to be regarded as a single event out of which the losses arose, for the purposes of the clause”. However, perhaps because it was unnecessary, there was little consideration of why a single event which satisfied the causation requirement would nevertheless have been too remote.

The concept of remoteness was developed further in Scott v Copenhagen Reinsurance Co (UK) Ltd [2003] EWCA Civ 688, [2003] Lloyd’s Rep IR 696, which was concerned with Iraq’s invasion of Kuwait on 2^nd August 1990. On that day Iraqi forces took control of the airport and, over the course of the following week, removed 15 aircraft belonging to Kuwait Airways Corporation, together with spares, to Iraq. A British Airways aircraft also present at the airport remained there and was later damaged by coalition bombing between 13^th and 26^th February 1991. The issue was whether the loss of all 16 aircraft constituted a “loss or series of losses arising from one event”, i.e. the invasion of Iraq and seizure of the airport. The Court of Appeal held that the loss of the 15 KAC aircraft on 2^nd August 1990 arose out of one event and should therefore be aggregated, but that the loss of the BA aircraft (as a result of later destruction by bombing) did not. Although the invasion was an event without which the loss of the BA aircraft would not have occurred, it was a different event which did not arise from and was not the inevitable result of the invasion, which occurred much later in time and only after other events (“the tortuous events of international diplomacy and coalition building which led to war, and the loss of the BA aircraft in and as a result of war, some six months later”) had also occurred; moreover, unlike the seizure of the Kuwaiti fleet, it was not something which the Iraqi invaders had intended.

Lord Justice Rix (with whom Lord Justices Schiemann and Keene agreed) cited Caudle v Sharp, which he regarded as having identified four requirements for the operation of an “arising out of one event” clause, namely:

“61.

… (i) something that can be called an ‘event’; (ii) the function of that event as being prior to the aggregated losses; (iii) a causative link between losses and event, undefined other than being looser than proximate cause; and (iv) the absence of remoteness. To which of course can be added the underlying concept of aggregation itself, that of a single unifying event.”

He continued:

“63.

As for causation and remoteness, it seems plain that the latter concept was used by this court as a tool to limit the otherwise infinite reach of the workings of causation. That is the function of the concept of remoteness in the law generally. It is in other words a legal tool used to separate out relevant from irrelevant causes. As often happens, however, the use of this tool is somewhat opaque. What the decision in Caudle v Sharp does seem to me to suggest, however, is that, even though the causative link is looser than that of proximate cause, the courts will look for a nearer and more relevant cause than for a more distant one. Another way of saying this is that the causative link has to be a significant rather than a weak one.”

All this was said in relation to a clause whose language, “arising out of one event”, required what Lord Justice Rix described as “a significant causal link” between the event and the loss:

“68.

… Nevertheless, it seems to me ultimately to be inherent in the concept of aggregation (‘arising out of one event’) that a significant causal link is required. In this connection I would refer to Lord Hoffmann’s substantial contribution in recent years to an understanding of what lies behind the courts’ intuitive judgments on issues of causation: see, for instance, Empress Car Co (Abertillery) Ltd v National River Authority [1999] 2 AC 22 at 29/35. Lord Hoffmann emphasises that it is not possible to give an informed answer to a question of causation when attributing responsibility under some rule without knowing the purpose and scope of the rule. In the present context, the purpose and scope of the rule has to be found in the concept of aggregation inherent in wording such as ‘arising out of one event’. A plurality of losses is to be regarded as a single aggregated loss if they can be sufficiently linked to a single unifying event by being causally connected with it. The aggregating function of such a clause is antagonistic to a weak or loose causal relationship between losses and the required unifying single event. This is the more easily seen by acknowledging that, once a merely weak causal connection is required, there is in principle no limit to the theoretical possibility of tracing back to the causes of causes. The question therefore in my judgement becomes: Is there one event which should be regarded as the cause of these losses so as to make it appropriate to regard these losses as constituting for the purposes of aggregation under this policy one loss?”

Lord Justice Rix described the way in which the court should approach its task as follows:

“81.

… Are the losses to be aggregated as all arising from one event? That question can only be answered by finding and considering all the relevant facts carefully, and then conducting an exercise of judgement. That exercise can be assisted by considering those facts not only globally and intuitively and by reference to the purpose of the clause, but also more analytically, or rather by reference to the various constituent elements of what makes up one single unifying event. It remains an exercise of judgement, not a reformulation of the clause to be construed and applied.”

Lord Justice Rix dealt also with a submission that, in principle, the Court of Appeal should not interfere with the trial judge’s evaluative conclusion in the absence of some error of law or principle. He described this at [86] as “a powerful submission”, but one which it was unnecessary to determine.

Simmonds v Gammell [2016] EWHC 2515 (Comm) was an appeal from arbitrators concerned with claims made against the Port of New York following the attack on the World Trade Center on 9^th September 2001. The issue was whether multiple claims from employees consisting of (1) workers’ compensation claims by workers injured or trapped at the WTC site at the time of the attack, and (2) respiratory claims by about 10,000 firemen, policemen, cleanup and construction workers and volunteers engaged in the rescue and clear up operation, could be aggregated pursuant to a clause referring to “loss, damage, liability or expense or a series thereof arising from one event”. The majority arbitrators held that the claims could all be aggregated as arising from one event, namely the attack on the WTC, which was not too remote.

Sir Jeremy Cooke dismissed the appeal, holding that this was a conclusion which the arbitrators had been entitled to reach:

“36.

The majority arbitrators were therefore involved in an exercise of judgment as to whether or not there was a sufficiently significant causal connection and they found the causal link between the respiratory claims and the attacks to be clear and obvious. Whether that is seen as a finding of fact or a mixed conclusion of law and fact matters not since this is well within the ambit of an exercise of judgement with which this court will not interfere. …”

The aggregation clause in Spire Healthcare Ltd v Royal & Sun Alliance Insurance Ltd [2022] EWCA Civ 17, [2022] Bus LR 170 was a “cause” and not an “event/occurrence” clause. It provided for aggregation of losses “arising out of all claims … consequent on or attributable to one source or original cause”, the issue being whether claims for personal injury arising out of the serial misconduct of a consultant breast surgeon over many years arose out of the same source or original cause. This court held that they did.

After referring to the difference between a cause and an event/occurrence, Lady Justice Andrews (with whom Lord Justices Underhill and Bean agreed) referred to the concept of remoteness in these terms:

“24.

‘Original cause’ in this context does not mean ‘proximate cause’, but instead connotes what Christopher Clarke J described as a ‘considerably looser causal connection’: Beazley Underwriting Ltd v Travelers Companies Inc [2011] EWHC 1520 (Comm), [2012] 1 All ER (Comm) 1241 at [27]. It follows that the ‘original cause’ need not be the sole cause of the insured’s liability. However, as Moore-Bick J observed in American Centennial Insurance Co v INSCO Ltd [1996] LRLR 407, 414, it is still necessary for there to be some causative link between the originating cause and the loss, and there must also be some limit to the degree of remoteness that is acceptable. As Mr Shapiro QC, for Spire, put it, not every ‘but for’ cause is sufficient to amount to an ‘original cause’. In searching for the unifying factor, one must not go back so far in the causal chain that one enters the realm of remote or coincidental causes which provide no meaningful explanation for what has happened.”

She referred again to the idea that the relevant contrast is between a cause which is too remote and one which provides “a meaningful explanation” for the loss at [36] and [37].

From this review of the cases I derive the following points of relevance to the present case.

First, whether and to what extent the remoteness principle applies depends on the true construction of the aggregation clause in question. It is an aspect of causation which needs to be considered as a result of language such as “arising out of”, which necessarily involves a causal link between the loss and the event out of which the loss arises.

Second, the application of the remoteness principle therefore depends on the nature and strength (or weakness) of the causal link which the aggregation clause requires. This is implicit in Lord Justice Evans’ reference in Caudle v Sharp to an event which is “not too remote for the purposes of the clause”. In the present case the judge concluded that the language of the aggregation clause (“arise from, are attributable to or are in connection with a single occurrence”) does require such a causal link, but that the inclusion of the words “in connection with” means that only a weak or loose causal link is required. The first of those conclusions was in issue below but is not challenged on appeal, in my judgment rightly so: words such as “in connection with” may not necessarily imply the need for a causal link when taken on their own (e.g. Campbell v Conoco (UK) Ltd [2002] EWCA Civ 704, [2003] 1 All ER (Comm) 35 per Lord Justice Rix at [19]), but do so when read in conjunction with “arising from” and “attributable to”. The present clause may therefore be contrasted with language such as “arising out of one event”, as in Scott v Copenhagen Re, where it was held that “a significant causal link” was required in order for aggregation to operate. Thus the principle of remoteness in the present case should apply with less rigour, or greater flexibility, than the equivalent clause in cases such as Scott v Copenhagen Re where a stronger causal link is required.

Third, remoteness is ultimately a “legal tool” which may be employed in a variety of circumstances. It may need to be considered, not only when there is only one candidate unifying event, in order to determine whether that event is too remote, but also when there are several candidates. In either case the search is for the (or a) significant or relevant event, or for an event which provides a meaningful explanation for the loss; these are all synonyms which express the same concept. They are not, however, to be equated with the proximate or effective cause. Particularly in a case where only a relatively weak causal link is required, that would be to introduce a strict requirement of proximate cause by the back door.

Fourth, the analysis has to be carried out as described in Scott v Copenhagen Re. That is to say, it calls for an exercise of judgment which is to some extent intuitive, but which also requires analysis of all the relevant circumstances of the case, including the nature of the causal link required by the aggregation clause. The analysis also includes consideration of the unities, so that an event which is far separate from the loss in time or place is more likely to be too remote than one which is contemporaneous with or geographically closer to the loss in question. Similarly, where the occurrence of the loss depends on a series of contingencies which may or may not occur after the happening of the event, that may also suggest that the event in question is too remote, i.e. that in all the circumstances it does not provide a meaningful explanation for the loss. But these are guidelines, not inexorable rules.

Finally, the nature of the exercise required is such that an appellate court should not interfere with the trial judge’s evaluation of the circumstances unless the conclusion reached is plainly wrong in the sense that it was not reasonably open to the trial judge or that the judgment discloses some error of principle. That was a submission which the Court of Appeal did not need to decide in Scott v Copenhagen Re but, in my view, it is correct in principle. For example, in Assicurazioni Generali SpA v Arab Insurance Group [2002] EWCA Civ 1642, [2003] 1 WLR 577 Lord Justice Clarke said:

“16.

Some conclusions of fact are, however, not conclusions of primary fact of the kind to which I have just referred. They involve an assessment of a number of different factors which have to be weighed against each other. This is sometimes called an evaluation of the facts and is often a matter of degree upon which different judges can legitimately differ. Such cases may be closely analogous to the exercise of a discretion and, in my opinion, appellate courts should approach them in a similar way.”

This approach has often been followed. Again by way of example, in Walter Lilly & Co Ltd v Clin [2021] EWCA Civ 136, [2021] 1 WLR 2753, Lady Justice Carr said:

“86.

An evaluation of the facts is often a matter of degree upon which different judges can legitimately differ. Such cases may be closely analogous to the exercise of a discretion and appellate courts should approach them in a similar way. The appeal court does not carry out a balancing task afresh but must ask whether the decision of the judge was wrong by reason of some identifiable flaw in the trial judge's treatment of the question to be decided, such as a gap in logic, a lack of consistency, or a failure to take account of some material factor, which undermines the cogency of the conclusion.

87.

The degree to which appellate restraint should be exercised in an individual case may be influenced by the nature of the conclusion and the extent to which it depended upon an advantage possessed by the trial judge, whether from a thorough immersion in all angles of the case, or from first-hand experience of the testing of the evidence, or because of particular relevant specialist expertise.”

Some further support for this view in the present context can be derived from Simmonds v Gammell, where Sir Jeremy Cooke was right to consider that the relevant question was whether the arbitrators were “entitled” to decide the case as they did. Although an appeal from arbitrators involves the additional feature that the court’s jurisdiction is limited to determining a question of law arising out of the award, and that it has no jurisdiction to review the arbitrators’ factual findings, Sir Jeremy Cooke was right in my view to regard the issue as requiring the kind of exercise of judgment with which an appellate court will not interfere in the absence of an error of principle.

As I have explained, the judge found that the initial human infection(s) in Wuhan could be described as a single occurrence. That satisfied the first of the three requirements identified in Caudle v Sharp which I have cited at [39] above. As I understood her submissions, Ms Mulcahy did not challenge that finding in itself, but submitted that this was not an occurrence which Allianz had pleaded or advanced at trial: in effect, Allianz had peppered the target but still missed the bull’s eye. I am inclined to agree with that submission. Although Mr Dougherty submitted that VE’s pleaded reliance on the “initial outbreak” in Wuhan amounted to or included reliance on the first animal to human infection(s) as an occurrence for the purpose of the aggregation clause, it is clear that this is not how the judge understood that plea. Rather, he understood it to refer to the further spread of the disease in Wuhan “thereafter”. In my view the judge was correct in his understanding.

However, I would be reluctant to decide the case on this pleading point and do not do so. The events in Wuhan were fully considered, not only in the present case, but in the two related cases which were tried consecutively with the present case. Whether the initial human infection(s) in Wuhan amounted to a single occurrence was a live issue in the Stonegate and Greggs cases (where the insurers’ pleadings concerning the initial outbreak in Wuhan were similar to Allianz’s pleading in this case) and the parties in each case were given the opportunity to comment on the evidence and submissions in the other cases. Evidently the judge saw no injustice in concluding in the present case that the initial human infection(s) did amount to a single occurrence and in my view he was right in that conclusion.

The judge found also that this occurrence had a sufficient causal relationship with VE’s losses to satisfy the relatively loose causal requirement implicit in the definition of an SBIL. That satisfied the second of the three requirements. There is no challenge on appeal to this conclusion of the judge.

However, Allianz’s primary case on aggregation failed because, in the judge’s view, the initial human infection(s) at Wuhan was too remote from VE’s loss for the reasons which I have set out at [30] above. Thus the third requirement, the absence of remoteness, was not satisfied. In my judgment this was a conclusion which the judge was entitled to reach. His decision involved both an intuitive exercise of judgment and the analysis and weighing of all the relevant circumstances, in which he, as the trial judge after a trial lasting several days which included expert evidence, was thoroughly immersed. There was in my judgment no error of principle or other comparable error in his approach which would entitle this court to interfere. Mr Dougherty submitted that the judge had adopted a “mechanistic” approach which failed to stand back and view the position as a whole through the eyes of an informed observer at the time when the question of aggregation had to be considered, but I can see no valid basis for that criticism.

Moreover, although it is not necessary to go this far, I agree with the judge’s conclusion. The loss suffered by VE as a result of the closure of its restaurants was remote in time and place from the initial animal to human infection(s) in Wuhan. There were a number of intermediate steps, as the judge explained, none of which were inevitable or would have been perceived as inevitable by an informed observer in the position of the insured in early 2020. Thus it was necessary for the disease to take hold in the human population in Wuhan, to prove serious and even fatal in many cases in the way that it did, to spread to the UK either directly or via one or more third countries, to spread within the UK, and to do so to such an extent that the Government decided to instruct the public to avoid social venues and to reinforce that instruction in law. In these circumstances, and even applying a relatively generous or flexible approach to the remoteness requirement in order to reflect the weak causal link required by the clause, the meaningful explanation for VE’s losses was not the initial human infection(s) in Wuhan but the later Government response to what happened thereafter. If the informed observer had asked the question why VE was suffering losses at any material time after 16^th March 2020, the answer would have been that the Government had required all its restaurants to close and remain closed.

Broadly the same analysis applies to Allianz’s secondary case that the introduction of Covid-19 into England or the UK amounted to a relevant single occurrence, although here the judge’s findings are different. The judge did not accept that this was a single occurrence at all. He considered that there were multiple introductions. Nevertheless, on the basis of the judge’s findings on the evidence, which I have set out at [8] above, there must have been an initial introduction of Covid-19 into England and, even if it was not possible to give a precise date when this happened, it was highly likely that this occurred between 1^st January and about 23^rd January 2020. I do not see why this should not be capable of qualifying as a single occurrence for the purpose of the aggregation clause, even if a considerable number of infected individuals arrived here at much the same time. The arrival of Covid-19 in the UK was something which undoubtedly happened at a particular time and in a particular way, even if it is not possible to give a precise date or place within the UK where it first arrived. But the same applies to the first human infection in Wuhan, which the judge was unable to date more precisely than late November and December 2020, and of which he was able to say only that at least two and possibly as many as 15 human beings had been infected at this time, but which nevertheless qualified as a single occurrence for the purpose of the clause.

The judge found also that the introduction of Covid-19 into the UK did not have even a weak causal link to all the cases in the UK which led, via the Government’s response and public reaction, to VE’s losses. That was because Covid-19 cases in this country were not linked by a continuous chain of infection to the initial introduction of the disease into the UK. Rather, very large numbers of introductions (at least 1179) were identified, each of which seeded multiple independent lines of transmission. In my view, however, that is too stringent an approach. I doubt whether the informed observer would regard the existence of a continuous chain of infection as necessary in order to satisfy the requirement of only a weak causal link between the arrival of Covid-19 in the UK and the losses suffered by VE. There is in my view a sufficient causal connection to satisfy the requirements of the clause.

I would therefore be inclined to hold that the first two requirements of the clause were satisfied in relation to Allianz’s secondary case. However, I consider that the judge was entitled (and if necessary, right) to conclude that the introduction of Covid-19 into the UK was too remote. As he said, although geographically more proximate, the first introduction of the disease into the UK was still temporally remote from the losses. The losses depended also on the spread of the disease within the UK to the extent that the Government intervened in the way it did, which (even looking at the position as it stood in late March) was by no means certain to have happened.

For these reasons I would dismiss Allianz’s appeal on the issue of aggregation.

I should add, however, that in considering the issue of aggregation neither the parties nor the judge drew any distinction between losses suffered by VE before and after the Government intervention in March 2020. Nor was any distinction drawn in the submissions in this court. I have therefore followed the same approach. However, it should be noted that VE’s pleaded case is that it began to suffer losses from no later than 27^th February 2020. Obviously, to the extent that VE did suffer losses before the Government’s instruction to the public on 16^th March 2020, those losses can only be recoverable under the “Notifiable Diseases” clause as there was no “Prevention of Access” or “enforced closure”. Further, the Government instructions or subsequent legislation cannot amount to an aggregating occurrence in respect of losses suffered before 16^th March 2020. The parties advanced no argument as to whether the initial infection(s) in Wuhan or the introduction of Covid-19 into the UK would have been too remote from any losses suffered before 16^th March 2020. That may have been because, as Ms Mulcahy suggested, any losses suffered in this early period did not exceed the £2.5 million limit, so that aggregation would be irrelevant. The point may matter in another case, but nothing I say now is intended to determine how it should be decided if it does arise.

VE submitted that, if losses fall to be aggregated, the aggregation should be on a “per Insured Location” basis, that is to say that the limit of £2.5 million should be applied separately to each of its restaurants.

The judge rejected this submission in a single paragraph:

“103.

VE made the case that, if there was any aggregation, it should be on a per Insured Location basis. I do not accept that that is the case. There is no justification for it in the wording of the definition of SBIL. Instead, the Limit of Liability per SBIL for Business Interruption – Property Damage, and for Insuring Clauses 2.3 (i), (ii), (iii), (iv) and (vi) are consistent only with the possibility that there can be a SBIL in respect of losses deriving from different Insured Locations; and that is also implicit in the terms of the second paragraph (beginning 'Where the Insured…') in Item 5 – Retention of the Schedule.”

On behalf of VE, Ms Mulcahy submitted that the triggers for cover under the Disease Clause, including “enforced closure”, are expressed by reference to matters occurring in relation to an Insured Location. Aggregation, when applicable, should therefore be calculated separately in relation to each Insured Location affected. She recognised that the Prevention of Access clause refers to “Insured Locations” in the plural, but submitted that this made no difference: although prevention of access may affect multiple premises simultaneously, it is a premises-specific peril. She submitted also that elements of the factual matrix support this construction. Thus the Insured Locations were divided into two groups in the policy, with different limits and Maximum Indemnity Periods applicable to each, while VE’s business is described in the policy schedule as “Operating a chain of Italian restaurants …” which included different kinds of venues, at different locations, and different brands, which would therefore be expected to be affected in different ways. She submitted that it would be surprising if a total limit of £2.5 million applied to an SBIL affecting all the Insured Locations when the declared profit of the business as a whole was some £36.5 million and the limit for some elements of Business Interruption was £48.6 million. She relied also on Corbin & King Ltd v Axa Insurance UK Plc [2022] EWHC 409 (Comm), [2022] Lloyd’s IR Rep 299, where Mrs Justice Cockerill had held that the limits in that case applied on a “per premises” basis.

On behalf of Allianz, Mr Dougherty submitted that there is nothing in the definition of an SBIL referring to losses at an individual Insured Location, but only a reference to loss arising from, attributable to or in connection with a single occurrence. Some occurrences might affect only a single location, while others might affect multiple locations. In the latter case, there is nothing in the clause to suggest that aggregation should operate on a “per premises” basis and the language of the clause should be given its natural meaning. Other clauses confirm that Business Interruption Loss is to be viewed by reference to the effect on the business as a whole and not only by reference to individual premises, while the Retention provision expressly envisages the possibility of a single SBIL applying to losses suffered at multiple locations.

In my judgment the judge was right to reject VE’s submission that aggregation operates on a “per Insured Location” basis, for the reasons which he gave, and on which Mr Dougherty expanded.

I accept that the language of the definition of an SBIL contains nothing to suggest that aggregation is intended to operate on a “per Insured Location” basis and that, on the contrary, it is perfectly capable of applying where a single occurrence affects multiple locations. There is no reason to think that, when this happens, the definition is intended to operate separately in relation to each such location.

This view is confirmed when Insuring Clause 2.3 is examined in the light of the definitions which it contains. Thus it refers to “interruption or interference to the Insured’s Business” as a result of specified causes, in which case Allianz agrees to pay “the resulting Business Interruption Loss”. The “Insured’s Business” is defined as “Operating a chain of Italian restaurants …”, that is to say by reference to the business as a whole. The “Business Interruption Loss” which is to be paid is defined to mean the “Reduction in Turnover”, while “Turnover” is also defined by reference to the business as a whole, as carried on at multiple locations:

“Turnover means:

i.

The money paid or payable to the Insured for goods sold and/or services rendered in the course of the Insured’s Business at the Insured Locations; and

ii.

Rent Receivable; and

iii.

interest income on the Insured’s capital deposits and monetary balances.”

The position is put beyond doubt, in my view, by the Retention provision. Item 5 of the policy schedule provided for a Retention of £5,000 to apply to each Single Property Loss and to subsidence, with a retention of £500 for computer or laptop loss. It continued:

“Where the Insured has made a claim for a Single Property Loss and/or a Single Business Interruption Loss affecting one or more Insured Locations that arise from, are attributable to or are in connection with the same single occurrence, only one Retention being the largest applicable will apply to all Single Property Losses and Single Business Interruption Losses combined. …”

This tracks the language of the definition of an SBIL and makes clear that an SBIL may affect multiple Insured Locations, in which case a single Retention will still apply.

Ms Mulcahy identified other provisions of the policy in which Insured Locations are expressly referred to in the plural, submitting that when this is intended, the drafter had made a positive decision, but these were provisions dealing with other matters which, in my view, have no real bearing on the current issue. In my judgment the provisions to which the ordinary and conscientious policyholder would turn, if necessary assisted by his broker, in order to discover how the aggregation clauses are intended to operate, are those to which I have referred. Having obtained a clear answer that there is nothing to suggest that aggregation is intended to operate on a “per Insured Location” basis, that policyholder would not scour the policy for terms or definitions which might call that answer into question.

In the light of this clear language, the factual matrix points on which Ms Mulcahy relied carry little or no weight. In particular, the limits and declared values in the policy simply reflect the extent of cover which Allianz was prepared to offer and for which VE was prepared to pay, which were agreed at a time when an imminent global pandemic was not even a cloud on the horizon as small as a man’s hand.

Nor does Corbin & King assist. The policy there was a composite policy insuring the business of multiple insureds, where each restaurant was separately owned by a separate company, which is a material distinction from the present case. That was one of two reasons given by Mrs Justice Cockerill for her conclusion. The more detailed points of construction which she made, on which Ms Mulcahy relied, were in relation to a different wording from what we have here.

As already explained, the judge concluded that there were three SBILs during the Period of Insurance which were not too remote, i.e. (1) the Government’s instruction on 16^th March 2020 that social venues should be avoided, (2) the Regulations of 20^th March 2020, and (3) the announcement and implementation of early closing and other restrictions on restaurants from 24^th September 2020. VE submitted that there were separate occurrences when Government measures were renewed, changed or relaxed. The judge rejected the submission at [104]. His reasoning is to be found in his judgment in the Greggs case:

“86.

Fourthly, I do not consider that an informed observer would have regarded announcements or measures which simply continued existing restrictions or made trivial changes as being separate ‘single occurrences’ for the purposes of the SBIL definition. I do not believe that it conforms to the parties’ intentions to have aggregation by reference to such matters, which effectively continued a status quo rather than marking any significant change to it. Nor would I consider that an informed observer would have regarded changes which simply reduced restrictions as being separate ‘single occurrences’ for the purposes of the definition. They were such as would of their nature be expected to reduce losses not to lead to them and thus would not constitute the type of matter which would sensibly be regarded as a factor unifying different losses.”

VE seeks to challenge this conclusion, contending that a decision to renew, change or relax a Government measure should also count as a single occurrence. For example, in July 2020 the restrictions which formed part of the first lockdown were relaxed, but VE says that losses continued to be suffered throughout the summer. It submits that if those losses are to be aggregated, they should be aggregated by reference to the relaxation of the March 2020 measures in July and not the March 2020 measures themselves. The effect, as I understand it, would be that VE would then contend that losses suffered between March and July would be aggregated by reference to the March occurrence with a limit of £2.5 million, while a further limit of £2.5 million would apply to losses between July and the introduction of new restrictions in September.

In order to challenge the judge’s decision, VE needs permission to appeal, which the judge declined to grant. I would refuse permission to appeal. I consider that the judge’s reasoning is clearly right.

It is relevant to note, however, that the way in which the March 2020 regulations worked was that they were to apply for a six month period, unless revoked, with provision for periodic reviews. Thus, for so long as they remained in force, the review decision was simply to maintain or relax the status quo. The analysis might have been different if the regulations had applied only for a specified period, after which they had to be renewed by a positive decision to do so – as would have been the case in September 2020 if they had remained in force until then.

The judge decided that if there is a Covered Event giving rise to Business Interruption Loss, there will always be at least one SBIL. He explained his reasoning between [74] and [81] of his judgment, while observing at [75] that it was unclear whether the issue had any real significance. VE seeks to challenge that conclusion, but here too needs permission to appeal in order to do so as this was refused by the judge.

In circumstances where, on both parties’ cases in this court, there is at least one SBIL in the present case, I cannot see that this issue has any practical significance and neither counsel was able to explain why it has. For that reason I would refuse permission to appeal and leave the point to be decided, if it ever has to be, in a case where it matters.

Allianz challenges the judge’s decision as to the scope of cover under the Prevention of Access and enforced closure clauses.

It will be recalled that Insuring Clause 2.3 provides for payment of Business Interruption Loss caused by “Notifiable Diseases [the definition of which includes “enforced closure”] … during the Period of Insurance” and by “Prevention of Access – Non-Damage during the Period of Insurance”.

On behalf of Allianz Mr Dougherty submitted that the effect of the words “during the Period of Insurance” is that only losses suffered during the Period of Insurance, that is to say up to and including 28^th September 2020, can be recovered. Thus if access was prevented on 1^st September 2020 and the restaurants were required to remain closed until 31^st December 2020, only losses incurred in the first 28 days could be recovered.

The judge recognised, as was common ground, that Business Interruption Loss incurred after the expiry of the Period of Insurance can only be recovered if proximately caused by a Covered Event occurring during the Period of Insurance. However, in the example just given of a four-month lockdown beginning on 1^st September 2020, there would be no difficulty about this. Otherwise, the judge rejected Allianz’s submission:

“67.

I do not consider that Allianz’s argument is correct. In my view the correct construction of the Policy is that there is a Covered Event when, in the case of Enforced Closure, there is an enforced closure of an Insured Location within the Period of Insurance, ie, if the closure takes place within the Period of Insurance. There could then be recovery for the resulting interruption and interference with the business, and the extent of that interruption or interference would depend on how long the closure lasted, irrespective of whether the whole period of such closure was within or after the Period of Insurance. Similarly in relation to Prevention of Access, if there are actions or advice which have, within the Period of Insurance, the effect of preventing or hindering the use of or access to Insured Locations, then there is cover for any resulting interruption or interference, and the extent of that interruption or interference would depend on how long the prevention or hindrance lasted, and the Clause does not require any period of such prevention or hindrance after the Period of Insurance to be disregarded.”

In the judge’s view, this was the meaning which would be given to the policy terms by a reasonable policyholder. It did not involve any manipulation of the words used, but simply read the provision that the enforced closure or Prevention of Access should occur ‘during the Period of Insurance’ as referring to the initiation of the closure or prevention. As the judge also pointed out, Allianz’s proposed construction would lead to some odd and paradoxical results:

“69.

I consider that the construction for which Allianz contends would produce uncommercial and unintended consequences. It would mean, for example, that if an Insured Location were the subject of enforced closure on the last day of the Period of Insurance, and remained closed for a week, the only cover under the Policy would be for the consequence of the first day of closure. While on Allianz’s contention, the remainder of the period of closure would fall within the next policy year, it would be quite possible, indeed probable, that insurers for the next year would exclude cover for an already subsisting closure / prevention or hindrance. It is also difficult to reconcile with the terms of Insuring Clause 2.3(xii), given that that clause requires, for there to be a covered Prevention of Access – Non Damage, an interruption or interference for more than 8 consecutive hours. If the whole of that period had to be within the Period of Insurance, and any period after the Period of Insurance did not count as relevant interruption or interference, then if an insured experienced a prevention or hindrance of use/access within the 8 hours before the expiry of the Period of Insurance, there could be no cover under the Policy, and almost certainly there would be no possibility of a recovery under an insurance for the next year. More generally, Allianz’s construction would mean that these two Insuring Clauses provided cover in a markedly different manner from how other Insuring Clauses would cover similar situations. For example, if there were a fire at an Insured Location during the Period of Insurance, and it led to the closure of an Insured Location for a significant period beyond the end of the Period of Insurance, then the entirety of that closure, up to the end of the MIP, would be relevant interruption or interference. But on Allianz’s case, if there was an enforced closure for health reasons before the end of the Period of Insurance, no part of the closure after the end of the Period of Insurance would be relevant interruption or interference. I consider that to be paradoxical, and reinforces me in my view as to how the two Insuring Clauses would reasonably be understood.”

This is, as Mr Dougherty recognised, a short point of construction of the policy. I respectfully agree with the judge. The function of the Insuring Clauses is to identify the Covered Events under the policy. The relevant Covered Event is a Prevention of Access or an enforced closure occurring during the Period of Insurance, that is to say between 29^th September 2019 and 28^th September 2020. A prevention or enforced closure occurring on 1^st September 2020 is such a Covered Event because it occurs during the Period of Insurance. Accordingly VE is entitled to recover the Business Interruption Loss proximately caused by that Covered Event, even if that loss extends beyond the Period of Insurance, subject only to the longstop that the Maximum Indemnity Period in the policy schedule is 12 (or in the case of some restaurants, 24) months.

Indeed, the definitions of “Indemnity Period” and “Reduction in Turnover” confirm the judge’s construction. These provide:

“Indemnity Period means the period of time during which interruption or interference to the Insured’s Business occurs as a consequence of the Covered Event beginning with the occurrence of the Covered Event and ending not later than the end of the Maximum Indemnity Period thereafter.”

“Reduction in Turnover means:

i.

the amount by which the Turnover during the Indemnity Period falls short of the Standard Turnover;

LESS

ii.

any costs normally payable out of Turnover (excluding depreciation) as may cease or be reduced during the Indemnity Period as a consequence of the Covered Event. …”

Reading these definitions together with the Insuring Clauses, Allianz agrees to pay the Business Interruption Loss proximately caused by a Covered Event which occurs during the Period of Insurance. The Business Insurance Loss which it agrees to pay is the Reduction in Turnover caused by the Covered Event, beginning on the date of the Covered Event and continuing for a maximum of 12 (or 24) months. Necessarily, therefore, the losses which VE is entitled to recover may continue beyond the end of the Period of Insurance.

I would therefore dismiss the appeal on this issue.

I would dismiss the appeal by Allianz and the cross appeal by VE. I would refuse VE permission to appeal on the two grounds for which permission is needed.