United Kingdom Supreme Court

128.         These are cogent arguments, but we are not persuaded that the court erred. The words “restrictions imposed” are general and unqualified. As the court recognised, in most cases the relevant restrictions would be directed at the insured premises or the use of the premises by the policyholder, but they are not required to be so. The court's police cordon example is relevant as an illustration of an inability to use the premises resulting not from a restriction directed at the premises or their use by the policyholder, but from a restriction which keeps the public out. Having full regard to context, we do not consider that the general words used should be read down in the way that Hiscox contends.

129.         The public authority clauses in Hiscox 1-4 (set out at para 111 above) do not cover all business interruption due to “restrictions imposed” by a public authority following an occurrence of a notifiable disease. They apply only where the interruption is caused by the policyholder's “inability to use” the business premises due to such restrictions.

130.         The court below held that the words “inability to use” in these clauses mean a complete inability to use the premises, save for use that is de minimis. The court's reasoning is set out in para 268 of the judgment:

“In our view this [phrase ‘inability to use'] plainly does not embrace any and every impairment of normal use. ‘Unable to use' means something significantly different from ‘hindered in using' or similar. Furthermore, the phrase is used in a context which includes the various sub-clauses (a) to (e) (in Hiscox 1), in each of which situations restrictions amounting to a complete inability to use the premises for the purposes of the business (albeit typically for a limited time) are readily foreseeable. We agree with Hiscox that there will not be an ‘inability to use' premises merely because the insured cannot use all of them; and equally there will not be an ‘inability to use' premises by reason of any and every departure from their normal use. Hiscox accepted, however, in our view correctly, that partial use might be sufficiently nugatory or vestigial as to amount to an ‘inability to use' the premises. Whether that was so would depend on the facts of a particular case.”

131.         Both the FCA and the Hiscox Interveners appeal against this conclusion.

132.         The FCA contends that the court should have concluded that the requirement for an “inability to use” the business premises will be satisfied where the policyholder is able to demonstrate that it has suffered an inability to use the premises for the ordinary purposes of its business. In particular, there will be an inability to use premises for the ordinary purposes of the business where part or all of the premises have had to be closed in response to restrictions imposed.

133.         The FCA gives an example of a bookshop which is required to close with the loss of all its walk-in customer business, representing some 80% of its income. The fact that it can continue to use the premises for telephone orders, representing 20% of its income, does not alter the fact that there is an inability to use its premises for a discrete part of its business activities. To hold that there is no cover available in those circumstances is unrealistic and uncommercial.

134.         The Hiscox Interveners submit that “inability to use” should “be read as referring to the material inability to use the premises for the insured's normal business activities, which could be partial or total.” They submit that the word “inability” does not denote the specific extent to which the policyholder lacks the ability to use the premises or entail that the inability to use the premises must be complete. To the contrary, it is a perfectly acceptable and correct use of the term to state that a policyholder is unable to do something, to the extent of that inability.

135.         Hiscox contends that the court's conclusion was correct for the reasons it gave. The question is a binary one. Can the insured use the premises for its business activities or not? Inability is not a matter of the extent to which someone is able or unable to do something. It means they cannot do it at all.

136.         We agree with the court and Hiscox that an inability of use has to be established; not an impairment or hindrance in use. On the other hand, we do not accept that the inability has to be an inability to use any part of the premises for any business purpose. The reference to “the business premises” is naturally read as including a discrete part of those premises which is capable of being used separately from other parts. Such an interpretation also makes commercial sense, as there may be little difference from a business point of view between the ability to use a small part of the premises for a limited purpose and closure of the whole premises. Furthermore, the language of the clause does not require there to be a complete inability to use the premises for all purposes. The court below appears to have accepted this, as it refers in the passage quoted above to “a complete inability to use the premises for the purposes of the business” (our emphasis). We agree that a qualification of this kind is implicit in the clause but again see no reason why different business purposes should not be distinguished if the relevant activities are capable of being conducted separately.

137.         We consider that the requirement is satisfied either if the policyholder is unable to use the premises for a discrete part of its business activities or if it is unable to use a discrete part of its premises for its business activities. In both those situations there is a complete inability of use. In the first situation, there is a complete inability to carry on a discrete business activity. In the second situation, there is a complete inability to use a discrete part of the business premises. To that extent the question is indeed binary.

138.         Whilst all cases will be fact dependent, the FCA's bookshop example would potentially be a case of inability to use the premises for the discrete business activity of selling books to walk-in customers. A department store which had to close all parts of the store except its pharmacy would potentially be a case of inability to use a discrete part of its business premises.

139.         The department store example also shows that the court was not correct to state that the other fortuities covered in sub-clauses (a) to (e) necessarily involve complete inability of use if by that it is meant inability to use any part of the premises for any purpose. A flood or a drains problem in a department store may well only affect a discrete part of that store.

140.         An example which potentially covers both cases would be a golf course which is allowed to remain open but with its clubhouse closed so that there is an inability to use a discrete part of the golf club for a discrete but important part of its business, namely the provision of food and drink and the hosting of functions.

141.         We should add that the FCA accepts that there is only cover for that part of the business for which the premises cannot be used. If, for example, a restaurant which also offers a takeaway service decides to close down the whole business it could only claim in relation to the restaurant part of the business. Equally, if there was a travel agent whose business was 50% walk-in customers, 25% internet sales and 25% telephone sales, it could only claim in relation to the loss of walk-in business, even though all parts of the business may have been depressed by the effects of COVID-19 and the governmental measures taken.

142.         The FCA and the Hiscox Interveners further criticise the court's conclusion that regulation 6 of the 26 March Regulations was most unlikely to lead to any inability of use. Having accepted that there could be “restrictions imposed” which were not directed specifically at the policyholder or its use of the premises, the court went on to say (at para 270 of the judgment) that the restrictions on leaving home imposed by regulation 6 could not be said to have led to an “inability to use” the premises of all policyholders where the business had relied on the physical presence of customers. As the court explained:

143.         It is suggested by the FCA and the Hiscox Interveners that the court here wrongly confused the questions of how loss might be mitigated by, for example, working from home or contacting customers at home with the question of whether there was an inability to use the premises. They contend that an inability to use business premises can arise because of restrictions imposed on others, such as employees or customers, and that there is no basis for a conclusion that an inability to use by virtue of such restrictions would be rare.

144.         We do not accept that there was any such confusion. Nor do we consider, even taking into account the wider interpretation of the requirement which we consider to be appropriate, that the court was wrong to say that the cases in which regulation 6 would cause an “inability to use” premises are likely to be rare. As the court points out, it must be an inability of use rather than hindrance or disruption. It is likely that it will be difficult for Category 3 and Category 5 businesses which were allowed to remain open to demonstrate the requisite inability.

145.         In summary, we would allow the appeal of the FCA and the Hiscox Interveners on this issue on the ground that “inability to use” the business premises may include a policyholder's inability to use either the whole or a discrete part of its premises for either the whole or a discrete part of its business activities.

146.         Similar issues arise in relation to whether only the total (as opposed to partial) closure of premises for the purposes of the existing business could qualify as “prevention” or “denial” of access to the premises under the prevention of access clauses in the Arch wording (see judgment at para 330), Hiscox 1, 2 and 4 (para 407), MSA 1 and 2 (paras 431-432) and Zurich 1 and 2 (para 495).

147.         The prevention of access clause in the Arch wording provides as follows:

“We will also indemnify You in respect of reduction in Turnover and increase in cost of working as insured under this section resulting from …

Government or Local Authority Action

Prevention of access to The Premises due to the actions or advice of a government or local authority due to an emergency which is likely to endanger life or property …”

148.         The court held that anything short of complete closure would not constitute “prevention of access” to the premises. The court considered (at paras 326-327 of the judgment) an example of a restaurant which in addition to in-restaurant dining offers a takeaway collection or delivery service. Regulation 4 of the 26 March Regulations required the closure during the emergency period of any premises, or part of the premises, in which food or drink were sold for consumption on those premises; but it did not require the premises to close to the extent that they were used for the purposes of providing a takeaway service. The court considered that, while it could be said that the restaurant owner policyholder and its employees were impeded or hindered in their use of the premises because they could not operate the restaurant for in-house dining, the government action did not cause prevention of access, as they were not prevented from accessing the premises for the purposes of carrying on that part of the existing business which involved providing the takeaway service. By contrast, the court accepted that if the restaurant did not previously offer a takeaway service but started one during lockdown, the position would be different. In such circumstances there would be prevention of access for the purposes of the business as it had existed when the insurance policy was issued.

149.         It will be apparent that there is considerable overlap between both the arguments and the reasons for the court's conclusions in relation to the “inability to use” and the “prevention of access” wordings. Mr John Lockey QC for Arch suggests that there is a significant difference between the clause in the Arch wording and the public authority clause in Hiscox 1-4 because the Arch wording focuses on access to the premises rather than use of the premises. Access refers to the means of entry to the premises. If the premises can be entered for the purpose of carrying on the business there, then there has been no prevention of access.

150.         In the present context we do not, however, consider that this provides a material distinction between the two wordings. As Mr Lockey accepts, the prevention of access does not have to be physical so that if, for example, the policyholder was able to and did enter the premises to carry out essential maintenance, that would not mean that the clause does not apply if access was prevented by law for the purposes of carrying on the business. Once, however, it is conceded - as is inevitable - that continued access to the premises for some purposes is compatible with there being cover, the question becomes: for what purposes? Furthermore, there is again no good reason to construe “the premises” as referring only to the entire premises rather than as encompassing part of the premises.

151.         In our view, for essentially the same reasons as given in relation to Hiscox 1-4, the Arch wording may, depending on the facts, cover prevention of access to a discrete part of the premises and/or for the purpose of carrying on a discrete part of the policyholder's business activities. We agree with Arch that prevention means stopping something from happening or making an intended act impossible and is different from mere hindrance. In both the situations contemplated, however, access to a discrete part of the premises or access to the premises for a discrete purpose will have been completely stopped from happening.

152.         The example of the restaurant which offers a takeaway service illustrates the commercial sense of this interpretation. The distinction drawn by Arch, and accepted by the court below, between continuing to operate such a service (where it is said that there would be no prevention of access or inability to use the premises) and starting a new takeaway service after closing the restaurant for dining is an unsatisfactory and arbitrary distinction. It is also illogical. If the premises can be put to such use, then it can be said that there is an ability to use them and that access to the premises for the purposes of carrying on the policyholder's business is not prevented. A more realistic view is that there is prevention of access to (and inability to use) a discrete part of the premises, namely the dining area of the restaurant, and prevention of access to (and inability to use) the premises for the discrete business activity of providing a dining in service.

153.         The FCA further criticises the court's conclusions (at paras 328-329 of the judgment) as to the limited relevance of regulation 6 to prevention of access. In essence, the court held that the restrictions on free movement imposed by regulation 6 did not in themselves prevent access to premises which remained open; and to the extent that, in consequence of the regulation, fewer people went to the relevant shop or office or only did so for the purposes of buying essential supplies or transacting business which could not be carried out remotely from home, this amounted to hindrance in the use of the premises but not to “prevention of access to The Premises”.

154.         Whilst we accept that it is possible for regulation 6 to result in a prevention of access, we consider that such cases are likely to be rare for the reasons set out when addressing para 270 of the judgment in respect of the related issue of inability of use. As the court below stressed, a prevention needs to be established; hindrance does not suffice. We agree with the court below that the Prime Minister's statement of 16 March 2020 did not cause prevention of access to the relevant insured business premises for the reasons given at para 328 of the judgment.

155.         We would therefore allow the FCA's appeal on this issue in relation to the Arch wording on the ground that “prevention of access” may include prevention of access to a discrete part of the premises or to the whole or part of the premises for the purpose of carrying on a discrete part of the policyholder's business activities.

156.         We consider that in principle the same analysis applies to the other prevention or denial of access wordings in relation to which the FCA has raised a similar issue, but that - as with the force of law point - there is no sufficient reason to address this issue separately in relation to clauses in respect of which it is academic.

The meaning of “interruption”

157.         Finally, Hiscox has raised a point about the meaning of the term “interruption” in the Hiscox wordings quoted at para 111 above, which refer to losses resulting from “an interruption to your activities …” The court below held that “interruption” in this context meant “business interruption generally” and included interference or disruption, not just a complete cessation of the policyholder's business or activities (see paras 274 and 409-414 of the judgment). Hiscox disputes this conclusion, renewing its argument made below that the term “interruption” naturally means a stop or break and is different from “interference”, which refers to circumstances where something continues but cannot be carried on properly. Alternatively, it is said that the term “interruption” must involve a more demanding test than “interference” and cannot extend to any kind of disruption, however slight.

158.         We reject these arguments. The ordinary meaning of “interruption” is quite capable of encompassing interference or disruption which does not bring about a complete cessation of business or activities, and which may even be slight (although it will only be relevant if it has a material effect on the financial performance of the business). Furthermore, the possibility that interruption may be partial is inherent in the policy provisions which deal with the calculation of loss and which envisage that the business may continue operating during a period of interruption but with reduced income or increased costs of working. In addition, as the court below pointed out (at paras 409-414 of the judgment), the policies contain a number of heads of cover for perils causing “interruption to your activities” which are plainly intended to apply in circumstances where there is only limited interruption and not a complete cessation of activities. Examples given included clauses covering interruption caused by loss of attraction by reason of damage in the vicinity of the premises and interruption caused by damage at the premises of a particular customer or supplier.

159.         We accordingly see nothing wrong with the court's reasoning and conclusion on this point.

VII      Causation

160.         We noted at para 59 above that, on the interpretation of the disease clause in RSA 3 (and most of the other sample wordings) accepted by the court below, questions of causation largely answered themselves. That is because, if the insured peril is COVID-19 (from the date when a case of the disease occurs within the specified distance of the insured premises), it follows that, from the date when such a case occurs, the policy covers all effects of COVID-19 on the policyholder's business. It is not in dispute that the measures taken by the Government in response to the disease and the business interruption consequent on those measures were caused by COVID-19 whatever the precise nature of the required causal link. It makes no difference for these purposes whether the occurrence of the disease within the specified area is seen as part of an indivisible cause, constituted by COVID-19 (the analysis preferred by the court below), or whether each of the individual cases of the disease is treated as a separate but equally effective cause of the actions taken by the Government and ensuing business interruption (the court's alternative analysis).

161.         On what we consider to be the correct interpretation of the disease clauses, however, questions of causation are of crucial importance. We have concluded that the clauses cover only the effects of cases of COVID-19 occurring within the specified radius of the insured premises. On this basis, the question of what connection must be shown between any such cases of disease and the business interruption loss for which an insurance claim is made becomes critical.

162.         Many different formulations may be found in insurance policy wordings of the required connection between the occurrence of an insured peril and the loss against which the insurer agrees to indemnify the policyholder. This may be illustrated by the variety of phrases used in the sample wordings in the present case. We noted earlier that RSA 3 uses the word “following” to describe the required connection between occurrence of a notifiable disease and interruption of the business. So do MSA 1 and MSA 2. In the Argenta wording the phrase used is “as a result of”. In QBE 1, it is “arising from”; and in QBE 2 and QBE 3, it is “in consequence of”. We do not think it profitable to search for shades of semantic difference between these phrases. Sometimes the policy language may indicate that a looser form of causal connection will suffice than would normally be required, such as use of the words “directly or indirectly caused by …”: see eg Coxe v Employers' Liability Assurance Corpn Ltd [1916] 2 KB 629. The same may arguably be said in the present case of the word “following”. But it is rare for the test of causation to turn on such nuances. Although the question whether loss has been caused by an insured peril is a question of interpretation of the policy, it is not (unlike the questions of interpretation of the disease, hybrid and prevention of access clauses considered above) a question which depends to any great extent on matters of linguistic meaning and how the words used would be understood by an ordinary member of the public. What is at issue is the legal effect of the insurance contract, as applied to a particular factual situation.

163.         As a general approach to the question of causation in marine insurance cases, the common law developed the test of “proximate” cause. This is codified in section 55(1) of the Marine Insurance Act 1906, which states that:

Like other provisions of the 1906 Act, this is treated by the courts as also stating the law applicable to non-marine insurance. As is clear from the words “unless the policy otherwise provides,” however, the rule is not inflexible. The requirement of “proximate” causation is based on the presumed intention of the contracting parties: see Reischer v Borwick [1894] 2 QB 548, 550; Leyland Shipping Co Ltd v Norwich Union Fire Insurance Society Ltd [1918] AC 350, 365 (Lord Atkinson); Becker, Gray & Co v London Assurance Corpn [1918] AC 101, 113-114 (Lord Sumner). But it is a presumption capable of being displaced if, on its proper interpretation, the policy provides for some other connection between loss and the occurrence of an insured peril.

164.         The expression “proximate” cause is somewhat misleading, as it is no longer used in its original sense. As Lord Sumner observed in Becker, Gray & Co at p 114: “It would be the better for a little plain English.” The term originates from Sir Francis Bacon's Maxims of the Law (1596). His first maxim (Regula I) was “In jure non remota causa sed proxima spectatur”, which may be translated as “In law, it is not the remote cause but the near cause that is looked to”. Bacon's explanation was that:

165.         During the 19th century, however, a different concept derived from Aristotle's notion of an “efficient” cause, meaning something that is the agency of change, became increasingly influential. Ultimately, this concept supplanted the idea that the law is concerned with the immediate cause of loss, although in insurance law the expression “proximate cause” was retained. An important case in cementing this development was the decision of the Court of Appeal in Reischer v Borwick [1894] 2 QB 548. This concerned a claim under a marine insurance policy which covered loss or damage from collision with any object, but not loss from perils of the sea. The ship collided with an object floating in a river, which caused a leak. The ship was anchored and the leak temporarily repaired. A tug was sent to tow the ship to the nearest dock but, while the ship was being towed, the effect of the motion through the water was that the leak was re-opened and the ship began to sink. To save the lives of the crew, the ship was then run aground and abandoned. The Court of Appeal held that, notwithstanding the intervening events, the loss of the ship was proximately caused by the collision and was therefore covered by the policy.

166.         This decision was approved by the House of Lords in the leading case of Leyland Shipping Ltd v Norwich Union Fire Insurance Society Ltd [1918] AC 350. The facts were materially similar to those of Reischer v Borwick. A ship torpedoed by a German submarine was towed to the nearest port but had to anchor in the outer harbour exposed to the wind and waves. After three days the ship sank. The ship was insured against perils of the sea but there was an exception in the policy for “all consequences of hostilities or warlike operations”. The House of Lords affirmed the decision of the lower courts that the loss was proximately caused by the torpedo, which was a consequence of hostilities, and was therefore not covered by the insurance. By far the fullest discussion of the concept of proximate cause is contained in the speech of Lord Shaw of Dunfermline. He made it clear, first of all, that the test of causation is a matter of interpretation of the policy and that “[t]he true and the overruling principle is to look at a contract as a whole and to ascertain what the parties to it really meant” (see p 369). He went on to say:

“What does ‘proximate' here mean? To treat proximate cause as if it was the cause which is proximate in time is … out of the question. The cause which is truly proximate is that which is proximate in efficiency. That efficiency may have been preserved although other causes may meantime have sprung up which have yet not destroyed it, or truly impaired it, and it may culminate in a result of which it still remains the real efficient cause to which the event can be ascribed.”

167.         There are passages in the authorities which characterise the question whether a cause was, in Lord Shaw's words, “proximate in efficiency” as simply a matter of applying common sense. This was the view expressed in Leyland Shipping by Lord Dunedin, who said (at p 362):

“… I think the case turns on a pure question of fact to be determined by common-sense principles. What was the cause of the loss of the ship? I do not think the ordinary man would have any difficulty in answering she was lost because she was torpedoed.”

The high water mark of this appeal to common sense is a passage that has often been quoted from the speech of Lord Wright in Yorkshire Dale Steamship Co Ltd v Minister of War Transport [1942] AC 691, 706:

“This choice of the real or efficient cause from out of the whole complex of the facts must be made by applying common sense standards. Causation is to be understood as the man in the street, and not as either the scientist or the metaphysician, would understand it. Cause here means what a business or seafaring man would take to be the cause without too microscopic analysis but on a broad view.”

168.         The common-sense principles or standards to be applied in selecting the efficient cause of the loss are, however, capable of some analysis. It is not a matter of choosing a cause as proximate on the basis of an unguided gut feeling. The starting point for the inquiry is to identify, by interpreting the policy and considering the evidence, whether a peril covered by the policy had any causal involvement in the loss and, if so, whether a peril excluded or excepted from the scope of the cover also had any such involvement. The question whether the occurrence of such a peril was in either case the proximate (or “efficient”) cause of the loss involves making a judgment as to whether it made the loss inevitable - if not, which could seldom if ever be said, in all conceivable circumstances - then in the ordinary course of events. For this purpose, human actions are not generally regarded as negativing causal connection, provided at least that the actions taken were not wholly unreasonable or erratic.

169.         Thus, on the facts of Reischer v Borwick it is apparent that the leak caused by collision with an object (an insured peril) would inevitably have led rapidly to the sinking of the ship if efforts had not been made to plug the leak. Those efforts merely delayed the occurrence of the loss. The leak was re-opened not as a result of any unusual weather conditions but simply through the ordinary motion of the water while the ship was under tow.

170.         On the facts of Leyland Shipping, the first causally relevant event which was an insured or excluded peril was the torpedoing of the ship by a German submarine. This was found without difficulty to fall within what amounted to an exception for “all consequences of hostilities”. The question whether this was the proximate cause of the loss then essentially involved a judgment as to whether the torpedo damage led inexorably to the loss of the ship or whether anything that occurred between the time when the ship was struck by the torpedo and her subsequent loss was sufficiently abnormal to justify treating it as negativing the causal connection. The House of Lords considered that the answer to this was clear and that the subsequent events did not displace the damage inflicted by the torpedo as the proximate cause of the casualty.

171.         Although in Leyland Shipping Lord Shaw referred to “the” proximate cause or “the” real efficient cause of loss, and other speeches also used the definite article, in Reischer v Borwick [1894] QB 548, 551, Lindley LJ had contemplated the possibility that the ingress of water when the vessel was under tow was a concurrent proximate cause but that this would not prevent the loss from being covered, as the policy did not require the loss to be exclusively caused by the collision. It has since become well established that, as Lord Buckmaster expressed the principle in Board of Trade v Hain Steamship Co Ltd [1929] AC 534, 539:

“it is no answer to a claim under a policy that covers one cause of a loss that the loss was also due to another cause that was not so covered.”

172.         In Heskell v Continental Express Ltd [1950] 1 All ER 1033, 1048, Reischer v Borwick was treated by Devlin J as authority for a more general principle, extending beyond the field of insurance, that “if a breach of contract is one of two causes, both co-operating and both of equal efficacy, … it is sufficient to carry judgment for damages”. (Like Allsop J in McCarthy v St Paul International Insurance Co Ltd [2007] FCAFC 28; (2007) 156 FCR 402, para 91, we do not read Devlin J's later comments in West Wake Price & Co v Ching [1957] 1 WLR 45, 49-50; [1956] 2 Lloyd's Rep 618, 624 as retracting his recognition of the possibility of two co-operating and equally effective causes).

173.         The leading modern authority which illustrates this possibility in the insurance field is JJ Lloyd Instruments Ltd v Northern Star Insurance Co Ltd (The Miss Jay Jay) [1987] 1 Lloyd's Rep 32. This case concerned a yacht insured against loss caused by “external accidental means”. The yacht sank as a result of what was held to be a combination of causes which were “equal, or at least nearly equal, in their efficiency” (per Slade LJ at p 40). They were adverse sea conditions and design defects which rendered the yacht unseaworthy. The first of these causes fell within the scope of the insurance; the other, unseaworthiness, did not but nor was it an excluded peril. The Court of Appeal held that in these circumstances the loss was proximately caused by a peril insured against and was therefore covered. See also ENE Kos 1 Ltd v Petroleo Brasileiro SA (No 2) [2012] UKSC 17; [2012] 2 AC 164, paras 12 and 74. As Lord Clarke of Stone-cum-Ebony stated in para 74: “where there are two effective causes, neither of which is excluded but only one of which is insured, the insurers are liable”.

174.         This situation is to be contrasted with one where there are two proximate causes of loss, of which one is an insured peril but the other is expressly excluded from cover under the policy. Here, although it is always a question of interpretation, the exclusion will generally prevail: see Wayne Tank and Pump Co Ltd v Employers Liability Assurance Corpn Ltd [1974] QB 57; Midland Mainline Ltd v Eagle Star Insurance Co Ltd [2004] EWCA Civ 1042; [2004] 2 Lloyd's Rep 604; Atlasnavios-Navegação, LDA (formerly Bnavios-Navegação, LDA) v Navigators Insurance Co Ltd (The B Atlantic) [2018] UKSC 26; [2019] AC 136, para 49.

175.         In none of the cases in either of these categories that were cited in argument could it be said that either cause which was characterised as a proximate cause on its own rendered the loss inevitable in the ordinary course of events. In each case it was the combination of the two causes which together made the loss inevitable. Neither would have caused the loss without the other.

176.         There is, in our view, no reason in principle why such an analysis cannot be applied to multiple causes which act in combination to bring about a loss. Thus, in the present case it obviously could not be said that any individual case of illness resulting from COVID-19, on its own, caused the UK Government to introduce restrictions which led directly to business interruption. However, as the court below found, the Government measures were taken in response to information about all the cases of COVID-19 in the country as a whole. We agree with the court below that it is realistic to analyse this situation as one in which “all the cases were equal causes of the imposition of national measures” (para 112).

The “but for” test

177.         The principal ground on which the insurers resist this analysis is that it cannot be said that, but for any individual case of illness resulting from COVID-19, the Government measures would not have been taken. The insurers argue, as a central plank of their case on these appeals, that whatever the exact nature of the causal link required by each sample policy wording it is a minimum requirement of any causation test that the occurrence of the insured peril made a difference to the occurrence of loss: in other words, it is necessary to show, at a minimum, that the loss would not have been sustained but for the occurrence of the insured peril. Thus, MS Amlin, whose submissions on this issue were adopted by the other insurers, said in their written case:

178.         MS Amlin sought to support this assertion by citing various judicial dicta that the “but for” test is an essential test which must be satisfied before a circumstance can be regarded as a cause in law or before, in the insurance context, the occurrence of an insured peril can be regarded as a proximate cause. In oral argument leading counsel for MS Amlin, Mr Gavin Kealey QC, sought to buttress this submission by the rhetorical use of emphasis and by pointing to the alleged failure of counsel for the FCA to find any insurance case which has held that a loss was caused by an insured peril that did not satisfy the “but for” test. Along with all the insurers' counsel, he also relied on the Orient Express case which we will consider later in this judgment.

179.         As the FCA has pointed out, the area described by the disease clauses which refer to a radius of 25 miles of the business premises is an area of a little under 2,000 square miles. To put this in perspective, this is bigger than any city in the UK, more than three times the size of Surrey, roughly the combined size of Oxfordshire, Berkshire and Buckinghamshire, and around a quarter of the area of Wales. The FCA produced a map to show that the whole of England can be covered, more or less, by just 20 circles each with a 25 mile radius. Nevertheless, if - as the insurers submit - the relevant test in considering the Government measures taken in March 2020 is to ask whether the Government would have acted in the same way on the counterfactual assumption that there were no cases of COVID-19 within 25 miles of the policyholder's premises but all the other cases elsewhere in the country had occurred as they in fact did, the answer must, in relation to any particular policy, be that it probably would have acted in the same way. As already mentioned, the court below found as a fact (at para 112 of the judgment) that the Government response was a reaction to information about all the cases of COVID-19 in the country and that the response was decided to be national because the outbreak was so widespread. In these circumstances it is unlikely that the existence of an enclave with a radius of 25 miles in any one particular area of the country which was so far free of COVID-19 would have led to that area being excepted from the national measures or otherwise have altered the Government's response to the epidemic. That in turn means that in the vast majority of cases it would be difficult if not impossible for a policyholder to prove that, but for cases of COVID-19 within a radius of 25 miles of the insured premises, the interruption to its business would have been less.

180.         The facts of the present case are distinguishable in this respect from the facts of cases referred to above such as Wayne Tank and The “Miss Jay Jay” in which it has been held or recognised that an insured peril which acts in combination with other causes of equal efficacy to bring about loss is capable of being regarded as a proximate cause. On the facts of those cases, although the insured peril was not sufficient on its own to cause loss, it was necessary in the sense that, but for the occurrence of the peril, the loss would not have been sustained. For example, in Wayne Tank the conduct of an employee who negligently left equipment switched on and unattended overnight before it had been tested, which was a risk insured against, would not have led to a fire which burnt down the factory if the equipment had not been defective. Equally, the fire would not have occurred if the equipment had been switched off. Each cause therefore satisfied the “but for” test.

181.         We agree with counsel for the insurers that in the vast majority of insurance cases, indeed in the vast majority of cases in any field of law or ordinary life, if event Y would still have occurred anyway irrespective of the occurrence of a prior event X, then X cannot be said to have caused Y. The most conspicuous weakness of the “but for” test is not that it wrongly excludes cases in which there is a causal link, but that it fails to exclude a great many cases in which X would not be regarded as an effective or proximate cause of Y. If, for example, a cargo is lost when a ship sinks, an unlimited number of circumstances could be identified but for which the loss would not have occurred. These will include some which may be plausible candidates for selection as a proximate cause - for example, the unseaworthy state of the vessel or exceptionally severe weather conditions. But they will also include an endless number of other circumstances. For example, it might equally be said that the loss would not have occurred but for the decision to manufacture the vessel, the decision of the owner or charterer to deploy the vessel on this particular route, the buyer's decision to purchase the cargo and the seller's decision to ship the cargo on that particular vessel, and so on. The main inadequacy, in other words, of the “but for” test is not that it returns false negatives but that it returns a countless number of false positives. That explains why it is often - and for most purposes correctly - described as a minimum threshold test of causation.

182.         It has, however, long been recognised that in law as indeed in other areas of life the “but for” test is inadequate, not only because it is over-inclusive, but also because it excludes some cases where one event could or would be regarded as a cause of another event. An example given by Hart and Honoré in their seminal treatise on Causation in the Law, 2nd ed (1985), p 206 is a case of two fires, started independently of each other, which combine to burn down a property: see Minneapolis, St P & S S M Ry Co, 146 Minn 430, 179 NW 45 (1920); Kingston v Chicago & NW Ry Co 191 Wis 610, 211 NW 913 (1927). It is natural to regard each fire as a cause of the loss even if either fire would by itself have destroyed the property so that it cannot be said of either fire that, but for that peril, the loss would not have occurred. Another example, adapted from the facts of the decision of the Supreme Court of Canada in Cook v Lewis [1951] SCR 830, is a case where two hunters simultaneously shoot a hiker who is behind some bushes and medical evidence shows that either bullet would have killed the hiker instantly even if the other bullet had not been fired. Applying the “but for” test would produce the result that neither hunter's shot caused the hiker's death - a result which is manifestly not consistent with common-sense principles.

183.         In these examples each putative cause, although not necessary, was on the assumed facts sufficient to bring about the relevant harm. Such cases are thus often described as cases in which the result is causally “over-determined” or “over-subscribed”. There is, however, a further class of cases in which a series of events combine to produce a particular result but where none of the individual events was either necessary or sufficient to bring about the result by itself. A number of examples are given by Professor Jane Stapleton in her scholarly work on causation in law: see most recently J Stapleton, “Unnecessary causes” (2013) 129 LQR 39; and “An ‘extended but for' test for the causal relation in the law of obligations” (2015) 35 OJLS 697.

184.         A hypothetical case adapted from an example given by Professor Stapleton, which was discussed in oral argument on these appeals, postulates 20 individuals who all combine to push a bus over a cliff. Assume it is shown that only, say, 13 or 14 people would have been needed to bring about that result. It could not then be said that the participation of any given individual was either necessary or sufficient to cause the destruction of the bus. Yet it seems appropriate to describe each person's involvement as a cause of the loss. Treating the “but for” test as a minimum threshold which must always be crossed if X is to be regarded as a cause of Y would again lead to the absurd conclusion that no one's actions caused the bus to be destroyed.

185.         Other examples of a similar nature given by Professor Stapleton include a case where the directors of a company unanimously vote to put on the market a dangerous product which causes injuries, although the decision only required the approval of a majority. Again, it cannot be said that any individual director's vote was either necessary or sufficient to cause the product to be marketed and yet it is reasonable to regard each vote as causative rather than to say that none of the votes caused the decision to be made. Another example is where multiple polluters discharge hazardous waste into a river. In all these cases each individual contribution is reasonably capable of being regarded as a cause of the harm that occurs, even though it was neither necessary nor sufficient to cause the harm by itself.

The defence costs cases

186.         A leading modern case in the law of tort in which it was held that causation (in a case of successive conversions) did not require the “but for” test to be applied, at least in the ordinary way by asking whether the claimant would have suffered loss but for the defendant's wrongful act, is Kuwait Airways Corpn v Iraqi Airways Co (Nos 4 and 5) [2002] UKHL 19; [2002] 2 AC 883 (see the opinion of Lord Nicholls of Birkenhead at paras 72-83). Although such cases in any field of law are rare, counsel for the FCA were in fact able to point to a line of cases in the insurance field in which it has been held or accepted that policyholders are entitled to an indemnity even though the “but for” test of causation is not satisfied. These are cases concerning the recovery of defence costs.

187.         In McCarthy v St Paul International Insurance Co Ltd [2007] FCAFC 28; (2007) 157 FCR 402, a firm of solicitors engaged in mortgage-lending was sued by 39 claimants who alleged that they were induced to lend by misrepresentations in an investment document. 36 of the misrepresentation claims against the firm succeeded and three failed. The firm claimed an indemnity under an insurance policy which provided cover for loss (including the costs of investigating and defending claims by third parties) “arising from any claim … in respect of any description of civil liability” but excluding liability brought about by fraud. It was held by the Federal Court of Australia that the firm was not entitled to an indemnity in relation to the 36 claims as its liability had been brought about by fraud. However, the firm was entitled to be indemnified for the costs of defending the other three claims. Furthermore, as held by Allsop J (with whom Kiefel and Stone JJ agreed) at para 119:

“… even if some investigation and defence costs can be seen to be referable to both a claim in respect of which there is indemnity and a claim in respect of which there is not, the insureds are entitled to such costs because they fall into an indemnity, otherwise untouched in its operation by any exclusion.”

In other words, costs that would have been incurred but for the insured claims because of the uninsured claims to which they were also referable were nevertheless held to constitute loss arising from an insured peril and were therefore recoverable.

188.         A similar approach was taken by the Privy Council in New Zealand Forest Products Ltd v New Zealand Insurance Co Ltd [1997] 1 WLR 1237. In that case costs were incurred in defending a claim brought against a company director who was insured against “all loss … which such officer has become legally obligated to pay on account of any claims made against him/her … for a wrongful act.” The claim was also brought against another person who was not insured. Both defendants were represented by the same lawyers. The issue was whether the insured director was entitled to an indemnity for defence costs which related at one and the same time to the defence both of the claim against him and of the claim against the uninsured third person. The Privy Council held that on the correct interpretation of the policy he was. Although the words “on account of” clearly required there to be a causal connection between the loss (comprising defence costs) and the claim against the director for a wrongful act, the fact that the disputed costs would still have been incurred even if the director had not been legally obligated to pay them did not prevent their recovery. The wording did not confine the recoverable costs to those which related “solely and exclusively to the officer” (at p 1242). This decision was followed and its reasoning applied by the Supreme Court in International Energy Group Ltd v Zurich Insurance plc UK Branch (Association of British Insurers and another intervening) [2015] UKSC 33; [2016] AC 509, where Lord Mance (at para 37) described the costs covered by the indemnity in New Zealand Forest Products as arising “on a conventional causative basis” because of a claim made against the director for a wrongful act. As Lord Mance stated at para 38:

“Once it is shown that an insured has on a conventional basis incurred defence costs which are covered on the face of the policy wording, there is, as the New Zealand Forest case [1997] 1 WLR 1237 shows, no reason to construe the wording as requiring some diminution in the insured's recovery, merely because the defence costs so incurred also benefited some other uninsured defendant.”

Multiple concurrent causes

189.         The question of causation becomes more difficult when the number of separate events that combine to bring about loss is multiplied many times over, so that, instead of there being two or 20 such events, there are, say, 200,000. Some scholars have contended that it is not appropriate to recognise trivial contributions as causes - for example, a teaspoon of water added to a flooding river or a match added to a raging forest fire. Others dispute this. Professor Richard Wright, a nother leading writer on causation in the law, has argued:

“Yet the teaspoon of water and the match contributed to and are part of the flood and forest fire, respectively. What if the same flood or fire were caused by a million (or many more) different people all contributing a teaspoonful of water or a single match? Denying that any of the teaspoonfuls or matches contributed to the destruction of the property that was destroyed by the flood or fire would leave its destruction as an unexplained, non-caused miracle. As a pure matter of causation, it cannot possibly matter whose hands supplied the different bits of water, flame or fuel. What is driving the intuition of no causation is the judgment regarding attributable responsibility, which is especially brought to mind if the question is posed as ‘Did the teaspoon of water or match destroy the property?' rather than ‘Did the teaspoon of water or match contribute, even if only extremely minimally, to the flood or fire that destroyed the property?' What is generally agreed upon is that the trivial contributor should not be held liable when her contribution was trivial in comparison to the other contributing conditions and was neither strongly necessary nor independently strongly sufficient for the injury at issue, but this is a normative issue of attributable responsibility rather than causal contribution.”

See RW Wright, “The NESS Account of Natural Causation: A Response to Criticisms” in R Goldberg (ed), Perspectives on Causation (Hart 2011) 285 at pp 304-305.

190.         We do not agree with Professor Wright in so far as he is suggesting in this passage that causation is a pure question of fact, albeit one that may be overlaid in areas such as tort and criminal law by questions of responsibility. Whether an event which is one of very many that combine to cause loss should be regarded as a cause of the loss is not a question to which any general answer can be given. It must always depend on the context in which the question is asked. Where the context is a claim under an insurance policy, judgements of fault or responsibility are not relevant. All that matters is what risks the insurers have agreed to cover. We have already indicated that this is a question of contractual interpretation which must accordingly be answered by identifying (objectively) the intended effect of the policy as applied to the relevant factual situation.

191.         For these reasons there is nothing in principle or in the concept of causation which precludes an insured peril that in combination with many other similar uninsured events brings about a loss with a sufficient degree of inevitability from being regarded as a cause - indeed as a proximate cause - of the loss, even if the occurrence of the insured peril is neither necessary nor sufficient to bring about the loss by itself. It seems incontrovertible that in the examples we have given there is a causal connection between the event and the loss. Whether that causal connection is sufficient to trigger the insurer's obligation to indemnify the policyholder depends on what has been agreed between them.

192.         We return to the disease clauses in the present case. We agree with Mr Kealey's submission on behalf of MS Amlin that the right question to ask is: did the insured peril cause the business interruption losses sustained by the policyholder within the meaning of the causal requirements specified in the policy? Taking MSA 1 as an example, the question is whether the interruption of the business carried on by the policyholder at the insured premises occurred “following” illness sustained by any person resulting from COVID-19 within a radius of 25 miles of the premises. In particular, it is necessary to ask: would the causal requirement imposed by the word “following” be satisfied by showing that one or more cases of illness from COVID-19 had occurred within the specified radius before national restrictions which caused interruption of the insured business were imposed on the basis of those and all other cases of COVID-19 that had occurred by that date?

193.         The FCA submits that the causal requirement would be met in such circumstances, applying the alternative analysis of the court below that each individual case of illness from COVID-19 was an equally effective cause of the government measures and consequent business interruption. The insurers contend that a “but for” test should be applied or, alternatively, if that contention is rejected, that a single case of disease or a relatively small number of cases of disease occurring within the specified radius is not sufficient to satisfy the causal connection required by the policy.

194.         In deciding between these competing interpretations, we consider that the matters of background knowledge to which the court below attached weight in interpreting the policy wordings are important. The parties to the insurance contracts may be presumed to have known that some infectious diseases - including, potentially, a new disease (like SARS) - can spread rapidly, widely and unpredictably. It is obvious that an outbreak of an infectious disease may not be confined to a specific locality or to a circular area delineated by a radius of 25 miles around a policyholder's premises. Hence no reasonable person would suppose that, if an outbreak of an infectious disease occurred which included cases within such a radius and was sufficiently serious to interrupt the policyholder's business, all the cases of disease would necessarily occur within the radius. It is highly likely that such an outbreak would comprise cases both inside and outside the radius and that measures taken by a public authority which affected the business would be taken in response to the outbreak as a whole and not just to those cases of disease which happened to fall within the circumference of the circle described by the radius provision.

195.         We do not consider it reasonable to attribute to the parties an intention that in such circumstances the question whether business interruption losses were caused by cases of a notifiable disease occurring within the radius is to be answered by asking whether or to what extent, but for those cases of disease, business interruption loss would have been suffered as a result of cases of disease occurring outside the radius. Not only would this potentially give rise to intractable counterfactual questions but, more fundamentally, it seems to us contrary to the commercial intent of the clause to treat uninsured cases of a notifiable disease occurring outside the territorial scope of the cover as depriving the policyholder of an indemnity in respect of interruption also caused by cases of disease which the policy is expressed to cover. We agree with the FCA's central argument in relation to the radius provisions that the parties could not reasonably be supposed to have intended that cases of disease outside the radius could be set up as a countervailing cause which displaces the causal impact of the disease inside the radius.

196.         This conclusion is reinforced by the other matter to which the court below attached particular importance in interpreting the disease clauses. This is the fact that the relevant wordings do not confine cover to a situation where the interruption of the business has resulted only from cases of a notifiable disease within the radius, as opposed to other cases elsewhere. As leading counsel for the FCA, Mr Edelman, pointed out, to apply a “but for” test in a situation where cases of disease inside and outside the radius are concurrent causes of business interruption loss would give the insurer similar protection to that which it would have had if loss caused by any occurrence of a notifiable disease outside the specified radius had been expressly excluded from cover. If the insurers had wished to impose such an exclusion, it was incumbent on them to include it in the terms of the policy.

197.         We accordingly reject the insurers' contention that the occurrence of one or more cases of COVID-19 within the specified radius cannot be a cause of business interruption loss if the loss would not have been suffered but for those cases because the same interruption of the business would have occurred anyway as a result of other cases of COVID-19 elsewhere in the country.

198.         The alternative argument cogently made by Mr Michael Crane QC, counsel for QBE, is that even if - as we have concluded - the policies on their proper interpretation do not require the “but for” test to be satisfied, it is still wrong to regard each of the individual cases of disease as a cause of the imposition of national measures and consequent business interruption losses. Rather than view each case individually, Mr Crane submits that the correct approach is to aggregate all the cases of disease which fall within the scope of the policy and to ask whether those cases, taken together, had an equal or similar causal impact when compared with the aggregate impact of all the cases of disease not covered by the policy. Wherever in the country a policyholder's business is located, the answer to that question will almost certainly be no. That is because, by the time of the actions taken by the UK Government in March 2020, COVID-19 had spread across most of the country; and, wherever on a map of the UK a circle with a 25 mile radius was drawn, the number of cases which had occurred within that radius would have been relatively small compared with the total number of cases elsewhere. On this analysis, therefore, the overwhelmingly dominant cause of business interruption loss was occurrences of disease not covered by the insurance. Hence it cannot be said that the loss was proximately caused - or on a realistic view caused at all - by an insured peril.

199.         Once again, the question is one of interpretation. We have concluded earlier that the word “occurrence”, as it is used in the disease clauses, bears its ordinary meaning of something which happens at a particular time, at a particular place and in a particular way; that each individual case of disease is properly regarded as a separate insured occurrence; and that those policies which do not use that term should nevertheless be interpreted similarly. The starting point for the analysis of causation is therefore that the occurrence of each case of illness sustained as a result of COVID-19 is a separate peril and thus potentially a separate cause of loss.

200.         This does not mean that cases of disease cannot combine to cause loss that would not have resulted from any individual case of disease had it occurred alone. That is what would normally be expected to happen and precisely what has happened in the present case, albeit on a far greater scale than might have been anticipated. We would accept that the language of the policies is not inconsistent with an interpretation whereby, when cases of disease combine to cause loss, an insured occurrence of disease is not to be regarded as a proximate cause unless the other insured cases of disease with which it has combined, taken together, are of similar causal potency as any uninsured cases of disease (also viewed together). We do not consider, however, that this interpretation is one that makes commercial sense of the disease clauses.

201.         An approach which involves weighing the relative potency of insured and uninsured causes in such a way might be appropriate if it were feasible to apportion the financial loss sustained by a policyholder's business between different cases or groups of cases of disease. However, that is not a realistic possibility. Where interruption of a business is caused by an outbreak of an infectious disease, the situation is not one of discrete concurrent causes each of which, acting on its own, would have caused part of the loss but not the whole of it. Although we do not think that it was strictly accurate for the court below to describe all the cases of COVID-19 in the country as indivisible, what plainly is indivisible is the effect of such cases, via the measures taken by the UK Government, on any insured business. As the loss is indivisible, the question whether it was caused by an insured peril is an all or nothing one.

202.         To attempt to answer that question by weighing the relative potency of insured and uninsured cases of disease would in many situations be unworkable. Suppose, for example, that some local restrictions were imposed in response to an outbreak of disease occurring partly inside and partly outside the specified radius of the premises and the restrictions had a damaging effect on the insured business. It is difficult to see how the weighing approach could provide a rational basis for determining whether the cases within the radius should be regarded as causes of the business interruption within the meaning of the policy. In the first place, it would potentially involve a complex and costly factual inquiry to try to find out exactly where all the cases of disease had occurred and how many occurred within and without the radius. Assuming that this information could be obtained, the next question would be what number or proportion of the cases is required to fall inside the radius in order for those cases to be regarded as causes of the business interruption for the purpose of the policy. Is it necessary, for example, to show that at least half of the cases to which the authorities were responding occurred inside the radius? Presumably not, since causes, in order to be regarded as proximate, do not have to be of precisely equal efficacy. So what lesser proportion would suffice? Is the threshold 40%, or 30%, or some other proportion, of the total number of cases? Or would even a small proportion count if the number of cases involved was in absolute terms sufficiently large? The policies provide no basis for answering these questions or for preferring any particular criterion to another.

203.         The still more fundamental objection to this approach is that again, as with the application of a “but for” test, it sets up cases of disease occurring outside the territorial scope of the cover in competition with the occurrences of disease within its scope in determining whether the policy will respond. Staying with the example of an outbreak of disease occurring partly inside and partly outside the specified radius which results in the imposition of restrictions over an area which includes the insured premises, suppose that the majority of cases (or whatever is the number or proportion which means there is no cover) have occurred outside the radius. Suppose also that the same restrictions would still have been imposed if the cases outside the radius had been somewhat fewer such that the majority of cases were inside the radius. On the weighing approach the slightly different incidence of cases outside the radius would nevertheless have the result that the insurers were not liable to pay. The policies could not rationally have been intended to operate in such a whimsical way.

204.         It is important to note that the irrational consequences of the weighing approach that we are considering are not simply the effects of applying a hard-edged rule. A speed limit of 30 mph is in one sense arbitrary in that someone driving at a speed of 31 mph commits an offence, whereas someone driving at a speed of 29 mph does not, even though there may be no difference between them in terms of the danger actually posed to road users. Nevertheless, the benefits of having a clear rule which is relatively straightforward to apply far outweigh any objection based on its arbitrary effects. In a similar way, it is entirely reasonable for insurers to set a territorial limit to the scope of business interruption cover which is arbitrary in the sense that there is no particular logic for selecting a radius of 25 miles, rather than 24 or 26, to define it and though the consequence of selecting a specific distance will inevitably be that there will be no cover for business interruption in some cases when on very slightly different facts there would have been cover.

205.         What is different about the weighing approach is, first of all, that it in fact defeats the intention apparent from the choice of a 25-mile radius that there should be a hard-edged rule. As we have noted, on the weighing approach the policy provides no identifiable criterion, let alone a hard-edged one, for deciding what number or proportion of disease cases must fall inside the specified radius in comparison with the number or proportion occurring outside it in order to trigger cover. Secondly, making the availability of cover depend on the number or relative number of cases of disease occurring outside the territorial limit of the policy as well as on the number within it seems to us to introduce a different and further form of arbitrariness into the operation of the policy that - unlike the application of a clear rule - lacks a rational basis.

206.         By contrast, an interpretation that recognises the causal requirements of the policy wordings as being satisfied in circumstances where each case of disease informs a decision to impose restrictions and treats each such case as a separate and equally effective cause of the restrictions irrespective of its geographical location and the locations of other such cases avoids such irrational effects and the need for arbitrary judgments and is also clear and simple to apply. This accords with the presumed intention of the parties to an insurance product sold principally to SMEs and often with relatively low financial limits. (For example, under MSA 1 the maximum payable for any one loss is £100,000.) It also accords with the desire for certainty manifest in the definition of cover by reference to a specific radius of 25 miles (or one mile) of the insured premises.

207.         This interpretation of the policies arrives at a broadly similar result to that reached by the court below, but by a different route. The court below interpreted the disease clauses as covering the effects of each case of disease wherever in the country it occurs, provided that at least one case occurs within the radius specified in the clause. On the interpretation that we think makes best sense, only the effects of any case occurring within the radius are covered but those effects include the effects on the business of restrictions imposed in response to multiple cases of disease any one or more of which occurs within the radius.

208.         Counsel for the insurers advanced an argument against the interpretation adopted by the court below which, if valid, would also apply to our preferred interpretation. The argument is that it leads to absurd results. Counsel for MS Amlin invited the court to take as an example the Scilly Isles, which they say did not have a single case of COVID-19 until September 2020. They submit that, if the interpretation were correct, it would mean that a policyholder located on the Scilly Isles would be entitled to an indemnity if it could be shown that a trawler sailed within 25 miles of the insured premises with a single person on board who had contracted COVID-19 (and even if the disease had not yet been diagnosed). Another similar example postulated by Mr Kealey in oral argument is one where someone with COVID-19 passes within 25 miles of the insured premises while travelling on a train through the area.

209.         We do not accept the underlying premise of this argument that the radius provisions should be read in such a narrow and literal way. On a realistic view of what is meant by an occurrence of a notifiable disease within a 25 mile radius of the insured premises, we do not think that the insured peril would reasonably have been intended to include a person who is not even visiting let alone resident in the specified area and who merely passes through it on a journey which involves no contact with anyone living in the area and therefore no risk of transmitting the disease to any such person. Accordingly, we do not consider that these examples shed light on the correct interpretation of the policies.

210.         Although the court below concluded that each of the individual cases of COVID-19 which had occurred by any given date is properly regarded as a separate but equally effective cause of Government action taken at that date, it does not appear to have followed through the logic of its analysis. When considering the disease clause in QBE 2, which the court interpreted - correctly in our view - as covering only cases of a notifiable disease occurring within the specified radius of the premises, the court stated (at para 235 of the judgment):

“… it does not appear to us that the causation requirement could be satisfied on the basis that the cases within the area were to be regarded as … one of many independent causes each of which was an effective cause, because this clause, in our view, limits cover only to the consequences of specific events.”

Declaration 12.1 of the court's order, which relates to QBE 2, declares that there is cover “only if any interruption or interference was caused by such occurrence(s) [ie any occurrence(s) within the radius], as distinct from COVID-19 outside that area”.

211.         This conclusion, in our respectful view, overlooks the point previously described by the court (at para 102 of the judgment) as fundamental that the radius provisions do not limit cover to a situation where the interruption of the business was caused only by cases of disease occurring within the area, as distinct from other cases outside the area.

212.         We conclude that, on the proper interpretation of the disease clauses, in order to show that loss from interruption of the insured business was proximately caused by one or more occurrences of illness resulting from COVID-19, it is sufficient to prove that the interruption was a result of Government action taken in response to cases of disease which included at least one case of COVID-19 within the geographical area covered by the clause. The basis for this conclusion is the analysis of the court below, which in our opinion is correct, that each of the individual cases of illness resulting from COVID-19 which had occurred by the date of any Government action was a separate and equally effective cause of that action (and of the response of the public to it). Our conclusion does not depend on the particular terminology used in the clause to describe the required causal connection between the loss and the insured peril and applies equally whether the term used is “following” or some other formula such as “arising from” or “as a result of”. It is a conclusion about the legal effect of the insurance contracts as they apply to the facts of this case.

Prevention of access and hybrid clauses

213.         The above analysis is also applicable to those hybrid clauses which contain, as one element, an occurrence of an infectious disease within a specified distance of the insured premises. For example, the relevant clause in RSA 1 covers “loss as a result of … closure or restrictions placed on the Premises as a result of a notifiable human disease manifesting itself at the Premises or within a radius of 25 miles of the Premises”. In order to show that business interruption loss is covered by this clause, it will be sufficient to prove that the interruption was a result of closure or restrictions placed on the premises in response to cases of COVID-19 which included at least one case manifesting itself within a radius of 25 miles of the premises.

214.         As noted earlier, however, unlike the disease clauses, the hybrid and prevention of access clauses specify more than one condition which must be satisfied in order to establish that business interruption loss has been caused by an insured peril. Furthermore, the structure of these clauses is that the elements of the clause are required to operate in a causal sequence. A good example is the public authority clause in Hiscox 1-3 (set out more fully at para 111 above), which covers financial losses “resulting solely and directly from an interruption to your activities caused by … your inability to use the insured premises due to restrictions imposed by a public authority during the period of insurance following … an occurrence of any human infectious or human contagious disease, an outbreak of which must be notified to the local authority” (our emphasis).

215.         The first of these causal links - between financial losses and an interruption to the policyholder's activities - is of less significance than the others. That is because, although the FCA has suggested otherwise, we think it clear that the interruption is not part of the description of the insured peril. The concept of business interruption in insurance of this kind was in our view correctly analysed by Mr Simon Salzedo QC in his submissions on behalf of Argenta. It is a description of the type of loss or damage covered by the policy, in the same way as the type of loss or damage covered by, for example, a buildings insurance policy is physical destruction or damage. Thus, in a buildings insurance policy, unless the policy otherwise provides, the insurer is liable for the contractual measure of (i) destruction of or physical damage to the insured buildings, which is (ii) proximately caused by (iii) a peril insured against under the policy (such as fire, storm etc). In business interruption insurance an interruption to the policyholder's business or activities is the counterpart of the first of these elements. It describes the nature of the harm to the policyholder's interest in the subject matter of the insurance for which an indemnity is given if it is proximately caused by an insured peril.

216.         In the Hiscox clause quoted above the first causal link is therefore concerned with the pecuniary measure of the interruption caused by an insured peril. Nevertheless, the peril covered by the clause is itself a composite one comprising elements that are required to occur in a causal sequence in order to give rise to a right of indemnity. Setting out the elements of the insured peril in their correct causal sequence, they are: (A) an occurrence of a notifiable disease, which causes (B) restrictions imposed by a public authority, which cause (C) an inability to use the insured premises, which causes (D) an interruption to the policyholder's activities that is the sole and direct cause of financial loss. Counsel for Hiscox in their submissions on this issue usefully represented the structure of the clause in a symbolic form as A→B→C→D, where each arrow represents a causal connection.

217.         An important question which divides the parties is how in these circumstances the elements of the clause interact with each other in determining whether or to what extent loss has been proximately caused by an insured peril.

218.         The court below approached this question on the basis that it requires an application of the “but for” test by ascertaining what the financial position of the policyholder's business would have been but for the occurrence of the insured peril. This must then be compared with the actual financial position of the business to establish the extent of the indemnity. The court considered that where, as in the case of the Hiscox clause, the peril is a composite one requiring a series of elements to be established, this “but for” test should be applied by asking what the position of the insured business would have been if none of the elements had occurred. The court explained this approach “as a matter of application of the insuring clause” in the following passage (at para 278 of the judgment):

“… we consider that the exercise must give effect to the insurance effected. This means assuming that the insured peril did not occur. The insured peril is a composite one, involving three interconnected elements: (i) inability to use the insured premises (ii) due to restrictions imposed by a public authority (iii) ‘following' one of (a) to (e), relevantly (b) an occurrence of an infectious or contagious disease. What the insured is covering itself against is, we consider, the fortuity of being in a situation in which all those elements are present. In answering the counterfactual question as to what would have been the position of the insured's business but for the occurrence of the insured peril, it is accordingly necessary to strip out all three interconnected elements, including in this instance the national outbreak of COVID-19.”

219.         Hiscox criticises this reasoning, pointing out that, if correct, it would mean that, where all the elements of the insured peril are present, Hiscox would be liable to indemnify the policyholder against all the effects on its business of the national outbreak of COVID-19. This, however, is manifestly not what Hiscox has agreed to cover. More generally, the court's approach would have the result that, however many the elements of the causal combination in a composite peril and however narrow in consequence the insured peril, once all elements are acting in combination such that the insured peril has occurred, the insurer becomes liable for all the consequences of the first element of the causal chain. Using the Hiscox symbols, on the court's approach, once the A→B→C→D causal chain is established, all the consequences of A are recoverable under the insurance. Logically, Hiscox submits, the position should be the reverse: the narrower the insured peril, the narrower the consequences for which the policyholder is entitled to an indemnity .

220.         These points are well made. It seems to us that, having correctly identified the fortuity covered by the insurance as a situation in which all three interconnected elements are present, the court erred by adopting a test which does not reflect that fortuity. The effect of “stripping out” all three elements in considering what the position of the business would have been but for the occurrence of the insured peril is to ask what its position would have been if none of those elements had occurred - including, as the court said, the national outbreak of COVID-19. That, however, is to treat the insured peril as being, not the risk of all three elements occurring (in causal sequence), but the risk of any one or more of the elements occurring. That would include a situation where there was an outbreak of a notifiable disease which caused interruption and loss to the business but which did not lead to any restriction being imposed that resulted in inability to use the premises. That is not the indemnity which the insurer agreed to give.

221.         Although, for these reasons, we consider that the court's analysis was mistaken, so too in our opinion is the alternative approach for which Hiscox (and the other insurers for whom this question is relevant) contend. Counsel for Hiscox assert that the “core” or “essence” of the peril insured against by the public authority clause in the Hiscox policies is “restrictions imposed by a public authority”. From this, they invite the inference that the relevant counterfactual question to ask is what would the financial position of the business have been if the public authority had not imposed the restriction(s) which resulted in inability to use the insured premises. This should then be compared with the actual financial position of the business to determine the scope of the indemnity.

222.         The first difficulty with this approach is the internally inconsistent way in which Hiscox has sought to define the relevant counterfactual scenario. In oral argument leading counsel for Hiscox, Mr Jonathan Gaisman QC, took as an example a nail salon which was forced to close by regulation 4 of the 26 March Regulations, as nail salons were one of the businesses listed in Part 2 of Schedule 2. He submitted that the relevant counterfactual is one in which nail salons were allowed to stay open but everything else which happened as a result of the COVID-19 pandemic happened as it actually did. The indemnity is therefore limited to such additional loss of business as was suffered as a result of nail salons being forced to close compared with the loss that would have been suffered anyway in a world in which nail salons had been permitted to stay open but all other restrictions such as the other closures and restrictions imposed by regulations 4 and 5 and the prohibition in regulation 6 against leaving home without reasonable excuse, along with all other consequences of the pandemic, were the same as in the real world.

223.         There is no logical reason, however, why in this example the “restriction imposed” should be characterised as the inclusion of nail salons in Part 2 of Schedule 2, rather than, say, the whole of regulation 4 or the whole of the 26 March Regulations or, more narrowly, regulation 4 in so far as it resulted in inability to use the insured premises. Indeed, the logic of Hiscox's approach indicates that the last of these scenarios is the correct counterfactual. That is because - as Hiscox submits, in our view correctly - the protection which it agreed to provide is against all the elements specified in the clause acting in causal combination to cause interruption of the business. One of those elements is inability to use the insured premises . Each additional element in the causal chain narrows the consequences for which the policyholder is entitled to an indemnity . Thus, the final link narrows the consequences covered by the policy from all the consequences of the “restriction imposed” to only those consequences which result from the policyholder's inability to use the insured premises. If, therefore, the correct approach is - as Hiscox submits - to determine what the position of the business would have been but for the occurrence of (all the elements of) the insured peril, the appropriate assumption to make is that everything would have happened as it did except for the last link in the causal chain. In other words, the actual position of the business should be compared with what its position would have been if there had been a national outbreak of COVID-19 which led to the imposition of all the restrictions which were in fact imposed by the Government, including the restriction which required nail salons to close, but this restriction did not apply (uniquely) to the policyholder's business.

224.         This conclusion may not be an attractive one for Hiscox as in this counterfactual scenario - in which the insured business is assumed to have been the only nail salon in the country permitted to stay open - it seems plausible that, in the absence of any competition, the turnover of the business during the indemnity period might well have been higher than it in fact was.

225.         Hiscox seeks to avoid this conclusion by its argument that “restrictions imposed by a public authority” is the “core” or “essence” of the peril insured under the clause. This is then relied on as a justification for ignoring the final element in the causal chain (inability to use the insured premises) in framing the relevant counterfactual question. However, we can see no warrant for such an approach. The last element in the causal chain is just as much part of the insured peril as the other elements. Using the Hiscox symbols, leaving it out of account in framing the counterfactual question involves treating the insured peril as if it were not A→B→C→D, but A→B→C. That, as counsel for Hiscox said of the court's approach, involves an impermissible recasting of the parties' bargain.

226.         Hiscox did not agree to indemnify the policyholder against all business interruption resulting from restrictions imposed by a public authority but only against interruption resulting from the policyholder' s inability to use the insured premises due to such restrictions. At various points in their submissions counsel for Hiscox recognised the logic of their analysis - for example, when stating in their written case (at para 42) that “what one strips out of the counterfactual is COVID-19 insofar as it leads to restrictions imposed causing an inability to use causing an interruption … Otherwise, COVID-19 and its effects and consequences remain in the counterfactual.”

227.         Once it is recognised that the approach for which Hiscox contends logically requires assuming in the counterfactual scenario the imposition of all the restrictions which were in fact imposed by the Government but that those restrictions did not require the insured premises to close, it can readily be seen that this approach is just as open to criticism as that of the court below. That is because it treats the insured peril as the risk that, if there was an outbreak of a notifiable disease sufficiently serious to lead a public authority to impose restrictions, the only effect of those restrictions (and of the outbreak of disease) would be to cause business interruption through inability to use the insured premises. On Hiscox's interpretation of its policy wording, to the extent that the imposition of the restrictions and/or the outbreak of disease would have caused business interruption anyway even if the policyholder had remained able to use the premises, the interruption is not covered by the policy . In the present case the indemnity is thus confined on this interpretation to loss that would have been avoided if COVID-19 and its consequences, including the imposition of the Government restrictions, had all occurred as they actually did save that the policyholder had (uniquely) been allowed to keep its premises open . No reasonable policyholder would have understood the insurance cover which it was getting to be insurance against such a narrow and fanciful risk. If Hiscox's interpretation were correct, it would make the public authority clause in the Hiscox policies a wholly uncommercial form of insurance which we cannot imagine that any insurer would see any sense in offering or that anyone running a business would see any sense in buying.

The “but for” test again

228.         In our view, where both the analyses that we have so far been considering go wrong is in their initial assumption that the correct counterfactual question to ask is: what would the financial position of the business have been but for the occurrence of the insured peril? The effect of applying that test is to limit the indemnity to business interruption loss which is solely and exclusively caused by the insured peril and has no other concurrent proximate cause. However, whilst the wording does include the words “solely and directly ”, those words (as noted earlier) form part of the description of the loss; they do not restrict the scope of the indemnity to interruption of the business which is proximately caused by an insured peril and has no other concurrent proximate cause. In so far, therefore, as Hiscox seeks to rely on the words “solely and directly ” as one of its grounds of appeal to argue that the extent of the indemnity provided is only in respect of losses proximately caused by the insured peril alone and nothing else, its argument is misplaced. In any event, it would in our view take clearer words than these to justify an interpretation which would treat the cover as so limited and uncommercial in its scope.

229.         To illustrate this further, we can take the example of a restaurant (with no takeaway or delivery service) forced to close pursuant to the 21 and 26 March Regulations. The effect of closing the restaurant would be to reduce the turnover of the business to nil. We do not consider that it would be consistent with the intended scope of the cover for the insurer to reject a claim for the resulting loss on the basis that the turnover would have been reduced anyway because of other consequences of national measures taken in response to COVID-19, such as the prohibition on leaving home without reasonable excuse. Such matters might have been sufficient to cause business interruption loss in the absence of the insured peril. To that extent, the losses covered by the public authority clause include losses which would have occurred even without the public authority restrictions. But it does not follow that the losses are irrecoverable.

230.         Hiscox invited the court to regard the FCA's submission to this effect as obviously fallacious. Mr Gaisman described it as a “thirteenth chime” - by which we understood him to mean an assertion which “like the thirteenth stroke of a crazy clock … not only is itself discredited but casts a shade of doubt over all previous assertions”: see Lord Light LCJ in the fictional case of Rex v Haddock in AP Herbert, Uncommon Law (1935), p 28. In our view, however, Hiscox has miscounted. The reason why such losses would have occurred in any event is that there are two (or more) causes each of which would by itself have inevitably brought about the loss without the other(s). The case is therefore one where the loss is “over-determined”. It is analogous to the example discussed earlier of two fires which combine to burn down a house although each would have done so even without the other. The fact that (because of the other fire) neither fire satisfies the “but for” test does not mean that in identifying each fire as a cause of the loss anything has gone wrong with the clock mechanism.

231.         It is not every concurrent cause of loss, however, which (although not an expressly excluded peril) would not reasonably be regarded as limiting the scope of the indemnity provided by the public authority clause. Continuing with the restaurant example, counsel for the FCA postulated a case where the restaurant had a star chef who was due to leave on 1 April 2020 for reasons unrelated to the pandemic. In this case it would be unreasonable to require the insurer to indemnify the policyholder for loss of turnover resulting from inability to use the premises in so far as such turnover would have been reduced in any event by reason of the chef's departure.

232.         The distinction between the departure of the chef and the consequences of COVID-19 which would have reduced the turnover of the restaurant in any event is in our view to be found in other parts of the judgment under appeal.

The court's alternative reasoning

233.         In addition to the reasoning in para 278 of the judgment (quoted at para 218 above) which we cannot support, the court below gave other reasons for leaving out of account all the elements of the insured peril when assessing what loss is covered by the insurance.

234.         The main thrust of these reasons is that, in a situation where loss is caused by a composite peril - such as inability to use the insured premises caused by restrictions imposed by a public authority caused by an occurrence of a notifiable disease - the concurrent consequences of the different elements are “inextricably linked” and to attempt to separate them would be “artificial and unrealistic”. This point is made in a number of places in the judgment. For example, in discussing the prevention of access clause in the Arch wording (which we consider below), the court said (at para 348 of the judgment):

“… the approach advocated by the insurers of stripping out the government restrictions etc and their immediate effect … whilst leaving the pandemic and its economic and social effects is entirely artificial and ignores the inextricable connection between the various elements of the insured peril, both as a matter of legal analysis and as a matter of practical reality, given the nature of the pandemic emergency.”

235.         The phrase “inextricable connection”, as we read the judgment, is used by the court in two different senses. One point made is that different effects of the pandemic and Government restrictions may be extremely difficult or impossible to disentangle. The court did not consider it a reasonable interpretation of the policies that the policyholder can only recover such loss as would have occurred in the absence, for example, of Government restrictions when the very effect of the restrictions will almost inevitably be to make it impossible to say with certainty what the position would have been in that event (see para 280 of the judgment).

236.         We do not consider this argument to be a strong one. We agree with counsel for Hiscox that it is in the nature of business interruption claims that they can give rise to difficult questions of quantification, often concerning what would have happened in hypothetical circumstances. That might be so, for instance, in the case of the restaurant with a famous chef who was due to leave on 1 April. It might be very difficult and uncertain to estimate what the effect on turnover of the chef leaving would have been if the restaurant had not been required to close on 21 March. Yet such difficulty of proof is not a sufficient reason to ignore the effects of the chef's departure.

237.         The other sense in which the elements of the insured peril are inextricably connected is that those elements and their effects on the policyholder's business all arise from the same original cause - in this case the COVID-19 pandemic. It is inherent in a situation where the elements of the peril insured under the public authority clause occur in the required combination to cause business interruption that there has been an occurrence of a notifiable disease which has led to the imposition of restrictions by a public authority. It is entirely predictable and to be expected that, even if they had not led to the closure of the insured premises, those elements of the insured peril would have had other potentially adverse effects on the turnover of the business. We have already expressed our view that it would undermine the commercial purpose of the cover to treat such potential effects as diminishing the scope of the indemnity. The underlying reason, as it seems to us, is that, although not themselves covered by the insurance, such effects are matters arising from the same original fortuity which the parties to the insurance would naturally expect to occur concurrently with the insured peril. They are not in that sense a separate and distinct risk.

238.         The court below (at para 281 of the judgment) discussed an example, based on a different limb of the Hiscox public authority clause wording, of inability to use restaurant premises due to restrictions imposed by the local authority following the discovery of vermin dislodged from a nearby building site. The court said that, to interpret the indemnity as limited to loss that would not have been suffered but for the forced closure of the premises “would render the cover largely illusory, as insurers would argue that, as no one is likely to want to eat at a restaurant infested by vermin, all or most of the business interruption loss would have been suffered in any event.” The court considered that such cover “cannot have been intended and is not what we consider would reasonably be understood to be what the parties had agreed to”.

239.         We agree and consider the underlying explanation to be that, where insurance is restricted to particular consequences of an adverse event (such as in this example the discovery of vermin in the premises) the parties do not generally intend other consequences of that event, which are inherently likely to arise, to restrict the scope of the indemnity.

240.         This principle is not limited to a situation where a causal chain is specified in the insuring clause, as it is in the Hiscox clause. It applies equally to an originating cause of loss covered by the policy which is not expressly mentioned. In the case of the Hiscox 4 wording, for example, the first element in the specified causal chain is an occurrence of a notifiable human disease within one mile of the insured premises. Where there is such an occurrence, the disease is very likely also to have occurred elsewhere. In the present case the originating cause of any local occurrence of disease is the global COVID-19 pandemic. In circumstances where the policy does not exclude loss arising from such an event, other concurrent effects of the pandemic on an insured business do not reduce the indemnity under the public authority clause.

241.         An analogy is provided by the facts of IF P & C Insurance Ltd v Silversea Cruises Ltd [2003] EWHC 473 (Comm); [2004] Lloyd's Rep IR 217. That case concerned a cruise operator's claim for business interruption losses from a downturn in bookings arising from US State Department warnings issued following the 9/11 terrorist attacks. The insured peril was the State Department warnings regarding terrorist activities rather than the terrorist activities themselves. It was common ground that the 9/11 attacks and the warnings were concurrent causes of the downturn in bookings. An argument that the causative effects of the terrorist attacks and the State Department warnings could be separated out was rejected by Tomlinson J on the grounds that “[i]t is simply impossible to divorce anxiety derived from the attacks themselves from anxiety derived from the stark warnings issued in the immediate aftermath” (para 68). He further held that “since the consequences of the events of September 11 are not … excluded from the ambit of the cover, as opposed to being simply not covered, a claim under the policy must lie” (para 69).

242.         On appeal the insurers raised a new argument that the 9/11 events themselves were an excluded peril. The Court of Appeal at [2004] EWCA Civ 769; [2004] Lloyd's Rep IR 696 rejected this argument. Rix LJ (with whom the other members of the court agreed) held that, on the proper interpretation of the policy, at para 104:

“There is no intention under this policy to exclude loss directly caused by a warning concerning terrorist activities just because it can also be said that the loss was also directly and concurrently caused by the underlying terrorist activities themselves.”

Conclusion on the Hiscox wording

243.         The conclusion we draw is that, properly interpreted, the public authority clause in the Hiscox policies indemnifies the policyholder against the risk (and only against the risk) of all the elements of the insured peril acting in causal combination to cause business interruption loss; but it does so regardless of whether the loss was concurrently caused by other (uninsured but non-excluded) consequences of the COVID-19 pandemic which was the underlying or originating cause of the insured peril.

244.         This interpretation, in our opinion, gives effect to the public authority clause as it would reasonably be understood and intended to operate. For completeness, we would point out that this interpretation depends on a finding of concurrent causation involving causes of approximately equal efficacy. If it was found that, although all the elements of the insured peril were present, it could not be regarded as a proximate cause of loss and the sole proximate cause of the loss was the COVID-19 pandemic, then there would be no indemnity. An example might be a travel agency which lost almost all its business because of the travel restrictions imposed as a result of the pandemic. Although customer access to its premises might have become impossible, if it was found that the sole proximate cause of the loss of its walk-in customer business was the travel restrictions and not the inability of customers to enter the agency, then the loss would not be covered.

The Arch prevention of access clause

245.         The “Government or Local Authority Action” clause in the Arch wording indemnifies the policyholder in respect of reduction in turnover resulting from:

“Prevention of access to The Premises due to the actions or advice of a government or local authority due to an emergency which is likely to endanger life or property.” (Our emphasis)

As Arch correctly submits, under this clause the policyholder is only entitled to recover those business interruption losses which are the product of the specified causal sequence: that is to say, losses which are caused by (i) prevention of access to the premises, which is caused by (ii) the actions or advice of a government or local authority, which in turn is caused by (iii) an emergency which is likely to endanger life (or property).

246.         Mr John Lockey QC for Arch further submits - again, in our view, clearly correctly - that the Arch clause does not cover loss of turnover caused by an emergency or by Government actions or advice in response to an emergency but which is not brought about by prevention of access to the premises. It is only loss occasioned by prevention of access to the premises that is covered by the clause. He then, however, goes on to submit that in assessing that loss the correct comparison, to reflect the agreement to indemnify, is between the actual turnover of the business and what the turnover would have been on the assumption that the premises had been required to close but everything else had remained the same, including the COVID-19 pandemic (which Arch accepts is an “emergency which is likely to endanger life”).

247.         In our view, this contention involves the same error as the case put forward by Hiscox. It incorrectly treats the indemnity as confined to loss which would not have occurred but for the operation of the insured peril. In the restaurant example discussed earlier, it may very well be true that, if the premises had not been forced to close, turnover would have been reduced in any event as a result of other effects of the emergency, including Government actions and advice. But it is also true that, if there had been no other effects of the emergency, or of Government actions and advice, this turnover would still have been lost as a result of the premises being forced to close. To this extent, there were concurrent causes of loss, each sufficient to cause loss without the other. Furthermore, these concurrent causes arose out of the same underlying or originating cause, namely the COVID-19 pandemic. As with the Hiscox policies, on the correct interpretation of the Arch wording, such loss is in our view covered by the policy.

RSA 1

248.         In the Hiscox clauses there are four causally connected elements in the description of the insured peril and in the Arch clause there are three. In the hybrid clause in RSA 1 there are only two such elements, namely (i) closure or restrictions placed on the premises as a result of (ii) a notifiable human disease manifesting itself at the premises or within a radius of 25 miles of the premises. The number of elements in the clause, however, makes no difference in principle to the analysis.

249.         In its written case RSA submitted that, in determining whether business interruption loss is covered by the clause, it is appropriate to compare the actual position of the business with a counterfactual situation in which there was no case of the disease within 25 miles of the premises. In oral argument Mr Turner corrected this to align RSA with Hiscox's case by submitting that it is necessary to assume only that there was no closure or restriction placed on the premises as a result of any such case(s) of the disease. For the reasons already given in discussing the arguments made on behalf of Hiscox, we do not consider this to be the correct approach. Rather, properly interpreted, the clause covers loss caused by the two elements of the insured peril operating in the required causal sequence, but does so regardless of whether any other (uninsured but non-excluded) consequences of the same underlying fortuity (the COVID-19 pandemic) were concurrent causes of the loss.

250.         It is unnecessary to address other hybrid and prevention of access clauses in relation to which, as noted earlier, this issue does not affect the outcome of the proceedings. In principle, however, a similar analysis must apply to those clauses as to the clauses which we have specifically addressed.

VIII    The trends clauses

251.         All the sample policy wordings considered in these proceedings contain clauses of a kind generally known as trends clauses. Such clauses are part of the standard method used in insurance policies that provide business interruption cover for quantifying the policyholder's financial loss. All of the insurers who appeal against the decision of the court below do so on the issue of how the trends clauses apply in the circumstances of the present case. It is the insurers' contention that the trends clauses have the effect that they are not liable to indemnify policyholders for losses which would have arisen regardless of the operation of the insured perils by reason of the wider consequences of the COVID-19 pandemic.

252.         In so far as such an argument is relied upon as excluding the loss from cover as a matter of “but for” causation it has been addressed above. The remaining question is whether, and if so how, the trends clause in the policies affects the position.

253.         The standard method used in business interruption insurance to quantify the sum payable under the policy takes an earlier period of trading for comparison purposes. In most wordings this is the calendar year preceding the operation of the insured peril. A “standard turnover” or “standard revenue” is derived from the turnover of the business in this period. This figure is then compared with the actual turnover or revenue during the indemnity period. The results of the business in the comparator period are also used to derive a percentage of turnover that represents gross profit. The rate of gross profit is then applied to the reduction in turnover to calculate the recoverable loss. Increase in the cost of working during the indemnity period is also typically covered.

254.         Whilst the basic comparison between the turnover of the business in the prior period and in the indemnity period will produce a rough quantification of the lost revenue, there may be specific reasons why a higher or lower figure would be expected for the indemnity period apart from the operation of the insured peril. For example, the general trend in the business may be such as to make it likely that there would have been increased or decreased turnover during the indemnity period in any case compared with the previous year. Equally, there may be specific reasons why the turnover during the prior year was depressed, such as a strike that affected the business, or why it would be expected to have been depressed anyway during the indemnity period, such as a scheduled strike. The purpose of the trends clause is to provide for adjustments to be made to reflect “trends” or “circumstances” such as these. The aim is to achieve a more accurate figure for the insured loss than would be achieved merely by a comparison with the prior period and to seek to arrive at a figure which, consistently with the indemnity principle, is as representative of the true loss as is possible. The adjustment may work in favour of either the policyholder or the insurer, but it is meant to be in the interests of both.

255.         As an example, the relevant clause in the Hiscox 3 wording provides as follows:

256.         The MSA, QBE and RSA 3 wordings require adjustments to be made to provide “for the trend of the business and for variations in or other circumstances affecting the business” and the Argenta wording is to like effect. The Arch wording refers to “any trends or circumstances”. Having required the adjustment for such trends, variations or circumstances, all the clauses refer to the aim being to represent “as near as possible” or “as nearly as may be reasonably practicable” the results which would have been achieved “but for the damage” or “if the damage had not occurred”.

257.         The reference to “damage” is inapposite to business interruption cover which does not depend on physical damage to insured property such as the cover with which these appeals are concerned. It reflects the fact that the historical evolution of business interruption cover was as an extension to property damage insurance. It was held by the court below, and is now common ground, that for the purposes of the business interruption cover which is the subject of these appeals, the term “damage” should be read as referring to the insured peril.

258.         The Arch trends clause, as applied to the peril insured under the Arch prevention of access clause by substituting the words of that clause (in italics) for the word “damage” would accordingly read as follows:

“Rate of Gross Profit and Standard Turnover may be adjusted to reflect any trends or circumstances which

(i)        affect The Business before or after the Prevention of access to The Premises due to the actions or advice of a government … due to an emergency which is likely to endanger life [or]

(ii)       would have affected The Business had the Prevention of access to The Premises due to the actions or advice of a government … due to an emergency which is likely to endanger life not occurred.

The adjusted figures will represent, as near as possible, the results which would have been achieved during the same period had the Prevention of access to The Premises due to the actions or advice of a government … due to an emergency which is likely to endanger life not occurred.”

Approach to interpretation

259.         In considering the proper interpretation of the trends clauses, we would emphasise the following points.

260.         First, the trends clauses are part of the machinery contained in the policies for quantifying loss. They do not address or seek to delineate the scope of the indemnity. That is the function of the insuring clauses in the policy.

261.         Second, in accordance with the general principle referred to earlier (see para 77 above), the trends clauses should, if possible, be construed consistently with the insuring clauses in the policy.

262.         Third, to construe the trends clauses consistently with the insuring clauses means that, if possible, they should be construed so as not to take away the cover provided by the insuring clauses. To do so would effectively transform quantification machinery into a form of exclusion.

263.         A similar point was made by the court below (at para 121 of the judgment) when discussing the trends clause in RSA 3, where the court stated that the clause is:

264.         In the present case that means that, unless the policy wording otherwise requires, the trends clauses should not be construed so as to take away cover for losses prima facie covered by the insuring clauses on the basis of concurrent causes of those losses which do not prevent them from being covered by the insuring clauses.

Proper interpretation of the trends clauses

265.         As the insuring clauses in the various wordings were interpreted by the court below, no problem of inconsistency with the trends clauses arose. In the case of the disease clauses, the insured peril was identified as the notifiable disease or all individual cases of the disease. That meant that the relevant counterfactual for the purpose of making adjustments under the trends clause was what results would have been achieved by the business during the indemnity period but for COVID-19. In relation to the prevention of access and hybrid clauses, the court held that the correct approach is to seek to ascertain what the results of the business would have been if none of the elements of the insured peril had occurred. So, for example, in relation to the Arch wording set out above, this would require the assumption to be made that there had been (i) no emergency likely to endanger life; (ii) no consequent Government action or advice; and (iii) no consequent prevention of access to the premises. Since the emergency was the COVID-19 pandemic, this meant assuming that it had not occurred. The loss which would have occurred but for the insured peril therefore did not include the wider consequences of the COVID-19 pandemic.

266.         On what we consider to be the proper interpretation of the disease clauses, however, the counterfactual question raised by the trends clauses cannot be so answered. The trends clauses call for an inquiry into what the financial results of the business would have been if the insured peril had not occurred. We have held the insured peril to be each case of the disease within the radius rather than each case of disease wherever it occurs. Although we have rejected the applicability of the “but for” causation test as a matter of interpretation of the disease clauses, such a test is expressly called for under the trends clauses.

267.         Equally, in relation to the prevention of access and hybrid clauses, we have held that the composite peril involves interconnected elements and that the court below was wrong to treat the insured peril, not as the risk of all the elements occurring (in causal sequence), but as if it were the risk of any one or more of the elements occurring . This means that one cannot apply the “but for” counterfactual under the trends clause on the basis that none of the individual elements of the insured peril occurred.

268.         How then are the trends clauses to be construed so as to avoid inconsistency with the insuring clauses? In our view, the simplest and most straightforward way in which the trends clauses can and should be so construed is, absent clear wording to the contrary, by recognising that the aim of such clauses is to arrive at the results that would have been achieved but for the insured peril and circumstances arising out of the same underlying or originating cause. Accordingly, the trends or circumstances referred to in the clause for which adjustments are to be made should generally be construed as meaning trends or circumstances unrelated in that way to the insured peril.

History of trends clauses and market practice

269.         To construe trends clauses in this way is consistent with the historical evolution of such clauses which shows that their focus has been on trends or circumstances unconnected with the insured peril.

270.         The first edition of Macken, Insurance of Profits published in 1927 considers an early form of trends clause which provided for adjustments in turnover to be made “for all extraordinary and other circumstances of the business”. In discussing the rationale of such clauses, Macken stated at p 35 (with emphasis added):

271.         A clause closely resembling the modern trends clause was discussed in the third edition of Insurance of Profits published in 1939. The clause provided that:

“… such adjustments shall be made as may be necessary to provide for the trend of the business and for variations in or special circumstances affecting the business either before or after the damage or which would have affected the business had the damage not occurred, so that the figures thus adjusted shall represent as nearly as may be reasonably practicable the results which but for the damage would have been obtained during the relative period after the damage.”

272.         Macken explained that “trends” and “variations” reflected the natural growth or contraction of the business and changes to it, and “special circumstances” reflected specific occurrences, such as strikes. As he stated at p 86:

“[The clause] takes in not only special circumstances affecting the business (such as strikes and other spectacular occurrences) but ‘the trend of the business' - its natural growth or diminution - and any variations in it, either before or after the damage.”

273.         The same clause was discussed in the first edition of Riley, Consequential Loss Insurances and Claims (1956) in which the examples he gave of circumstances for which an adjustment should be made included matters such as: a trade recession; an industrial dispute; a special advertising campaign; a large new contract; new plant; the vagaries of weather and a general upward trend.

274.         In the fourth edition of Riley published in 1977, by which time the standard wording discussed referred to “other circumstances” rather than “special circumstances”, the author commented as follows at para 36 (with emphasis added):

“It is important to bear in mind that the indemnity in respect of reduction in turnover is qualified by the words ‘in consequence of the damage.' If, therefore, the reduction is attributable wholly or in part to causes not connected with the damage which would have affected turnover irrespective of the damage having taken place, an adjustment must be made to the figure of standard turnover in order to reflect as accurately as possible the loss solely due to the damage.”

Riley then gave an example of a reduction in turnover caused both by insured damage and by an “extraneous circumstance such as a strike” (our emphasis) and explained that an appropriate adjustment must be made to the figures for standard turnover to reflect the effect of the strike.

275.         In Hickmott, Interruption Insurance: Proximate Loss Issues (1990) an example is given of “what is considered to be the UK market intention of the cover” in a situation where there is damage both to the insured premises and to the surrounding area. The example is a case where both a hotel and a necessary access bridge (not owned by the hotel) are damaged by a storm. Had there been no damage to the hotel, it would have had no business anyway because of the damage to the access bridge until the bridge had been repaired. The author expresses the opinion that the business interruption loss would nevertheless be assessed on the basis of turnover that would have applied if the access bridge had not been damaged for such time as the damage to the hotel on its own would have affected the trading of the business: see para 35 at p 28 and Case Study IV at p 50.

276.         In the most recent, tenth edition of Riley on Business Interruption Insurance (2016) at para 15.21, reference is made to the approach adopted by the UK market in settling claims for business interruption following floods in Cockermouth in Cumbria in November 2009. The flooding was such that all the shops on Main Street were flooded to a depth of many feet and, after the water subsided, the street effectively became a building site for the next six months. If the trends clause had been treated as requiring an adjustment to be made to reflect what the turnover of each individual shop would have been but for the damage to that particular shop, the insurance recovery would have been substantially reduced on the basis that the business would have suffered a severe downturn in any event by reason of the wide area damage that affected other businesses. That was not, however, the approach taken by the market. Instead, claims were met up to the level that would have applied had the damage been restricted solely to the insured premises.

277.         The tenth edition of Riley (written by Mr Harry Roberts, an experienced loss adjuster) also discusses the decision in the Orient-Express case, which we review below, and doubts “whether it is actually a satisfactory outcome for either insurers or policyholders.” It is also said that the Orient-Express case “has served to highlight a potential difference between the wording of the policy and the original intention of insurers” and that the conclusion reached leads to “a potentially counter-intuitive result which is not likely to leave the insurance industry in a good light” (see pp 407-408).

278.         Our interpretation of the trends clauses is also consistent with the approach of the US courts. In US policy wordings it appears that the counterfactual is generally expressed in terms of what would have happened “had no loss occurred”. Where the underlying cause of the loss is a hurricane, it has been held that this requires the assumption to be made that the hurricane would not have occurred. Thus, in Catlin Syndicate Ltd v Imperial Palace of Mississippi Inc, 600 F 3d 511 (2010) the Fifth Circuit Court of Appeals had to decide how to determine loss under the business interruption provision of an insurance policy issued to the Imperial Palace casino in Mississippi. The casino was damaged by Hurricane Katrina which also caused damage to the surrounding area. When the casino re-opened, its revenue was much greater than before the hurricane as many nearby casinos remained closed and people who wanted to gamble had few choices. The casino argued that, in quantifying its business interruption loss, the requirement to consider what would have happened “had no loss occurred” meant assuming that no damage to the casino had occurred, but not that the hurricane had not occurred. This argument was rejected by the court (at pp 515-516) in the following terms:

“Imperial Palace asserts that Catlin asks us to interpret the business interruption provision in such a way that the phrase ‘had no loss occurred' morphs into ‘had no occurrence occurred'. Imperial Palace argues that instead, we should disentangle the loss from the occurrence and determine loss based on a hypothetical in which Hurricane Katrina hit Mississippi, damaged all of Imperial Palace's competitors, but left Imperial Palace intact: the occurrence occurred, but the loss did not. While we agree with Imperial Palace that the loss is distinct from the occurrence - at least in theory - we also believe that the two are inextricably intertwined under the language of the business-interruption provision. Without language in the policy instructing us to do so, we decline to interpret the business-interruption provision in such a way that the loss caused by Hurricane Katrina can be distinguished from the occurrence of Hurricane Katrina itself.”

279.         To similar effect is the decision of the Fourth Circuit Court of Appeals in Prudential LMI Commercial Ins Co v Colleton Enterprises Inc 976 F 2d 727 (1992) which concerned damage to a motel in South Carolina caused by Hurricane Hugo.

280.         The US cases illustrate that the suggested construction avoids the problem of what have been termed “windfall profits”. In both the Catlin case and the Prudential case the argument of the insured was that the loss should be adjusted by comparing the actual results of the business with what they would have been if there had been no damage to the casino/motel, but the hurricane had nevertheless occurred causing all the other damage that was in fact caused to the surrounding area. Adopting this basis of adjustment would have put the casino/motel in a position to claim increased “windfall” profits as a result of being the only undamaged casino/motel in the area.

281.         A further consequence of our interpretation of the insuring clauses is that, in calculating the amount which the insurer is liable to pay, the calculation must be confined to those activities of the business which were interrupted by the operation of the insured peril. Where a discrete part of the business was not interrupted by the insured peril, the relevant comparison is therefore between the actual turnover and the adjusted standard turnover only of the interrupted activities.

282.         To illustrate how this works in practice, we can take an example of a fashionwear business insured under the Arch wording which has a shop but also makes sales through a website. The effect of regulation 5(1) of the 26 March Regulations was to compel such a business to close its shop to customers but to allow it to carry on the part of its business which involved selling through its website. We may suppose that demand for fashionwear and hence turnover from orders placed though the website nevertheless fell as a result of the national lockdown. As discussed earlier (at para 141 above), we consider that in such a situation the policy covers the interruption of the business caused by the prevention of access to the shop premises even though it does not cover any effects of COVID-19 and the Government restrictions on business transacted through the website.

283.         In assessing the amount which the insurer is liable to pay , the first step is to identify which activities of the business were interrupted by the insured peril. In t his example, the business activity interrupted by the operation of the peril insured under the p revention of access clause wa s opening the shop to customers. On the other hand, selling through the website w as not interrupted by the occurrence of the insured peril .

284.         The next step is, for those activities interrupted by the insured peril, to identify the income actually earned from those activities during the period of interruption. This amount is then compared with the standard turnover, adjusted to reflect any trends or circumstances which affected those activities before the occurrence of the insured peril or which would have affected them had the insured peril not occurred. As discussed, for this purpose the trends or circumstances for which adjustments should be made do not include trends or circumstances arising out of the same underlying or originating cause as the insured peril, namely the COVID-19 pandemic. In the same way, the aim of the adjustments is to arrive at figures which represent, as near as possible, the results that would have been achieved during the relevant period but for the prevention of access to the premises due to Government actions due to an emergency likely to endanger life and any consequences of the COVID-19 pandemic which affected the relevant part of the business concurrently with the insured peril.

286.         The result will be that the amount recoverable by the policyholder will not be reduced if, had the shop not been compelled to close, its turnover would have been lower in any event as a result of other consequences of the pandemic. Nor will it be increased if, had the shop been allowed to remain open in circumstances where all its competitors were forced to close, its turnover would have been higher. At the same time the policyholder will not be able to recover any loss caused by a fall in website orders as such loss is not cover ed .

287.         For the reasons given, we consider that the trends clauses in issue on these appeals should be construed so that the standard turnover or gross profit derived from previous trading is adjusted only to reflect circumstances which are unconnected with the insured peril and not circumstances which are inextricably linked with the insured peril in the sense that they have the same underlying or originating cause. Such an approach ensures that the trends clause is construed consistently with the insuring clause, and not so as to take away cover prima facie provided by that clause.

288.         We therefore reach a similar conclusion to the court below, by a slightly different route. We consider, as they did, that the trends clauses do not require losses to be adjusted on the basis that, if the insured peril had not occurred, the results of the business would still have been affected by other consequences of the COVID-19 pandemic.

289.         In one respect, however, the court below did not carry through the logical implications of this analysis. As recorded in Declaration 11.4(c) and (d) of its order, the court held that:

290.         The significance of this point depends upon the extent to which there was “a measurable downturn in the turnover of a business due to COVID-19 before the insured peril was triggered.” It may be illustrated by an example of a claim under the Arch wording by the owner of a pub to which access was prevented when the Prime Minister announced the closure of various businesses, which included pubs, on the evening of 20 March and the 21 March Regulations then required these businesses to close. Suppose that, as a result of public concern about contracting COVID-19 and the advice given by the UK Government before 20 March 2020, the turnover of the pub in the week ending on 20 March was only 70% of its turnover in the equivalent week of the previous year. On the approach accepted by the court below, in estimating what the turnover of the business would have been during the indemnity period had the insured peril not occurred, account should be taken of this downturn. Assuming that the downturn would have continued during the indemnity period if access to the pub had not been prevented, the standard turnover used to calculate the loss should be adjusted downwards under the trends clause to 70% of what it would otherwise have been. It is this adjusted figure which should then be compared with the actual turnover of the business while the pub was closed in assessing the sum payable under the policy.

291.         The FCA and the Hiscox Interveners appeal against the court's decision on this point, arguing that it is inconsistent with the court's own conclusion (discussed earlier and recorded in Declaration 11.2 of its order) that “the correct counterfactual when calculating an indemnity is to assume that once cover under the policy is triggered none of the elements of the insured peril were present” - which in the case of the Arch prevention of access clause, for example, means assuming no prevention of access, no Government action and no emergency.

292.         Arch and the other insurers argue that making a downwards adjustment under the trends clause to reflect the results of the business before cover was triggered does not involve assuming that the emergency caused by COVID-19 or Government action taken in response to the emergency would have caused a reduction in turnover after access to the premises was prevented. It merely involves taking account of a measurable downwards trend in turnover which existed before cover was triggered and which therefore was not caused by the operation of the insured peril. Counsel for Arch submit that the effect of the FCA's argument is to compensate the policyholder for uninsured loss which began before the insured peril operated and which would have continued during the indemnity period.

293.         The court below accepted this submission and also held (at para 351 of the judgment) that any downturn in turnover before the date(s) when businesses closed pursuant to Government actions or advice was a trend or circumstance which affected the business before the occurrence of the insured peril for which an adjustment was required by the express words of the Arch trends provision (quoted at para 258 above).

294.         We cannot agree that such a downturn in turnover is a trend or circumstance for which an adjustment is permitted let alone required by the Arch trends clause (or any of the other trends clauses in issue on the present appeals). As we have interpreted the trends clauses, the trends or circumstances for which adjustments may be made do not include trends or circumstances caused by the insured peril (or its underlying or originating cause). Furthermore, the aim of any adjustment is to seek to ensure that the adjusted figures will represent as nearly as possible the results which would have been achieved during the indemnity period had the insured peril (and its underlying or originating cause) not occurred.

295.         Contrary to the insurers' submissions, making a downwards adjustment to reflect the effects on the business of the COVID-19 pandemic before cover was triggered plainly does involve assuming that those effects would have continued after the occurrence of the insured peril. It involves assuming that, even if there had been no prevention of access, turnover during the indemnity period would have been reduced as a result of other effects of the emergency. The fact that those effects were being felt before cover was triggered makes no difference in this regard. Nor is it correct that, as counsel for Arch submit, leaving such effects out of account in calculating the indemnity amounts to compensating the policyholder for uninsured loss which began before the insured peril operated. On the contrary, to reduce the indemnity to reflect a downturn caused by other effects of the pandemic, whenever they began, would be to refuse to indemnify the policyholder for loss proximately caused by the insured peril on the basis that the loss was also proximately caused by uninsured (but non-excluded) perils with the same originating cause. As discussed earlier, that is not permissible.

296.         Accordingly, we consider that the court below was wrong to hold that the indemnity for business interruption loss sustained after cover was triggered should be reduced to reflect a downturn in the turnover of the business due to COVID-19 which would have continued even if cover had not been triggered by the insured peril. The court had correctly concluded that losses should be assessed on the assumption that there was no COVID-19 pandemic. Consistently with that conclusion, the court should have held that, in calculating loss, the assumption should be made that pre-trigger losses caused by the pandemic would not have continued during the operation of the insured peril.

297.         Orient-Express Hotels Ltd v Assicurazioni Generali SpA [2010] EWHC 1186 (Comm); [2010] Lloyd's Rep IR 531 concerned a claim for business interruption loss arising from damage to a hotel in central New Orleans from wind and water as a result of Hurricanes Katrina and Rita in the autumn of 2005. The insurance policy, which was governed by English law, provided cover against physical damage to the property on an all risks basis. It also provided cover for loss due to interruption or interference with the business “directly arising from Damage”. The trends clause was in similar terms to those in the present case and provided that the adjusted figures “shall represent as nearly as may be reasonably practicable the results which but for the Damage would have been obtained during the relative period after the Damage”. The term “Damage” was defined in the policy to mean “direct physical loss destruction or damage” to the hotel.

298.         As a result of the damage to the hotel it was closed throughout September and October 2005. The surrounding area of New Orleans was also devastated by the hurricanes and there was a mandatory evacuation of the city. The city was only re-opened at the end of September and the beginning of October 2005. The hotel re-opened on 1 November 2005, although not yet fully repaired. Its business was severely affected both by the physical damage to the hotel and by the general impact of the hurricanes on the city. A dispute arose about the amounts which the insured (“OEH”) was entitled to recover which the parties agreed to submit to arbitration in London. A central issue was whether OEH could recover business interruption losses which resulted not only from the physical damage to the hotel but also from the wider damage to the city.

299.         Before the arbitrators the insurer submitted that OEH could only recover loss that it would not have suffered but for the physical damage to the hotel and not loss of business which the hotel would have suffered even if it had not been damaged at all (because of the effects of the hurricanes on the city). The arbitrators accepted the insurer's case, observing that counsel for OEH “did not, in the view of the tribunal, ever supply a convincing answer to this submission”. The tribunal held that the language of the policy:

“… requires OEH to establish that the cause of the loss claimed is the Damage to the Hotel. It is not necessary or relevant for this purpose to go behind the Damage and consider whether the event which caused the Damage also caused damage to other property in the City: the fact that there was other damage which resulted from the same cause does not bring the consequences of such damage within the scope of the cover.”

300.         The tribunal also rejected an argument based on concurrent causation, stating that:

301.         OEH obtained permission to appeal under section 69 of the Arbitration Act 1996 on two questions of law:

302.         On the appeal OEH were represented by different counsel who, in relation to question (1), argued that on the correct interpretation of the insuring clause the “but for” causation test should not be applied because this was one of those “very occasional” cases where “fairness and reasonableness require a relaxation in the standard” (para 24). This had not been argued before the arbitrators. The court recognised that there was “considerable force” in the argument (para 33), but pointed out that what “fairness and reasonableness” require “is very much a matter for the tribunal of fact, rather than for this court on an appeal limited to questions of law” (para 36). The court also observed that the issue was one on which no findings had been made by the arbitral tribunal since the argument was not put to it (para 37). In any event, the court concluded that it had “not been established that the tribunal erred in law in adopting the ‘but for' approach to causation which they did” (para 40).

303.         In relation to question (2) and the proper interpretation of the trends clause, the court rejected all of the nine arguments made on behalf of OEH and agreed with the conclusion reached by the tribunal. The court concluded as follows (at paras 57-58):

“57.     I agree with the tribunal that the clause is concerned only with the damage, not with the causes of the damage. What is covered are business interruption losses caused by damage, not business interruption losses caused by damage or ‘other damage which resulted from the same cause'. Nowhere in the Trends clause does it state that ‘variations or special circumstances affecting the Business either before or after the Damage or which would have affected the Business had the Damage not occurred' has to be something completely unconnected with the damage in the sense that it had an independent cause to the cause of the damage. The assumption required to be made under the Trends clause is ‘had the Damage not occurred'; not ‘had the Damage and whatever event caused the Damage not occurred'.

58.       I agree with Generali that OEH's construction effectively requires words to be read into the clause or for it to be re-drafted. Further, such a re-drafting of the Trends Clause, which would allow OEH to recover for the loss in gross operating profit suffered as a result of the occurrence of the insured event (ie the hurricanes) as opposed to the loss suffered as a result of the damage to the hotel, is inconsistent with the causation requirement of the main insuring clause which OEH accepts requires proof that the losses claimed were caused by damage to the hotel.”

304.         There are a number of general observations to be made about the Orient-Express decision.

305.         First, as with all cases, the decision and reasoning of both the arbitral tribunal and the court was framed by the arguments advanced by the parties. On these appeals we have had the benefit of far more detailed and wide-ranging argument than at the arbitration hearing or on the one-day arbitration appeal in the Orient-Express case.

306.         Second, the context of the decision was an appeal under section 69 of the Arbitration Act 1996, which is limited to questions of law “arising out of an award”. As illustrated by the argument raised on the appeal as to what “fairness and reasonableness” requires, this limits the permissible scope of argument.

307.         Third, unusually for an appeal under the Arbitration Act 1996, the court gave permission for a further appeal to the Court of Appeal and granted the requisite certificate that the appeal raised a point of law of general importance. The court thereby recognised that OEH's case on causation was well arguable and indeed had a real prospect of success. In the event the case settled before an appeal was heard.

308.         In the present case the court below considered that the Orient-Express decision was distinguishable but, if necessary, would have reached the conclusion that it was wrongly decided and would have declined to follow it. For reasons already given in addressing the causation and trends clauses issues, on mature and considered reflection we also consider that it was wrongly decided and conclude that it should be overruled.

309.         The main error occurred at the first stage of the analysis when considering causation under the insuring clause. Applying the analysis set out earlier under the heading “Causation”, business interruption loss which arose because both (a) the hotel was damaged and also (b) the surrounding area and other parts of the city were damaged by the hurricanes had two concurrent causes, each of which was by itself sufficient to cause the relevant business interruption but neither of which satisfied the “but for” test because of the existence of the other. In such a case when both the insured peril and the uninsured peril which operates concurrently with it arise from the same underlying fortuity (the hurricanes), then provided that damage proximately caused by the uninsured peril (ie in the Orient-Express case, damage to the rest of the city) is not excluded, loss resulting from both causes operating concurrently is covered. In the Orient-Express case the tribunal and the court were therefore wrong to hold that the business interruption loss was not covered by the insuring clause to the extent that it did not satisfy the “but for” test.

310.         If the tribunal or the court had held that the loss concurrently caused by both the damage to the hotel and the damage to other parts of the city was covered by the insuring clause, that would have fundamentally affected the approach to the interpretation of the trends clause. In any event, for the reasons set out above under the heading “The trends clauses”, we consider that the correct approach in the Orient-Express case would have been to construe the trends clause so as to exclude from the assessment of what would have happened if the damage had not occurred circumstances which had the same underlying or originating cause as the damage, namely the hurricanes.

311.         In reaching a different conclusion now to that which we both reached over ten years ago we would refer to what Justice Jackson said in McGrath v Kristensen, 340 US 162, 177-8 (1950), concurring in a decision which contradicted an opinion he had given when Attorney General:

“Precedent, however, is not lacking for ways by which a judge may recede from a prior opinion that has proven untenable and perhaps misled others. See Chief Justice Taney, License Cases, 5 How 504, recanting views he had pressed upon the Court as Attorney General of Maryland in Brown v Maryland, 12 Wheat 419. Baron Bramwell extricated himself from a somewhat similar embarrassment by saying, ‘The matter does not appear to me now as it appears to have appeared to me then.' Andrew v Styrap, 26 LTR (NS) 704, 706. And Mr Justice Story, accounting for his contradiction of his own former opinion, quite properly put the matter: ‘My own error, however, can furnish no ground for its being adopted by this Court. ...' United States v Gooding, 12 Wheat 460, 25 US 478. Perhaps Dr Johnson really went to the heart of the matter when he explained a blunder in his dictionary - ‘Ignorance, sir, ignorance.' But an escape less self-depreciating was taken by Lord Westbury, who, it is said, rebuffed a barrister's reliance upon an earlier opinion of his Lordship: ‘I can only say that I am amazed that a man of my intelligence should have been guilty of giving such an opinion.' If there are other ways of gracefully and good naturedly surrendering former views to a better considered position, I invoke them all.”

312.         We likewise invoke whatever ways by which we may “gracefully and good naturedly” surrender “former views to a better considered position.”

XI       Conclusion

313.         Although we have accepted some of the insurers' arguments on their appeals, in no case has that affected the outcome of the appeal. It follows that the insurers' appeals are dismissed. We allow the FCA's appeal on grounds 1 and 4, and on grounds 2 and 3 in the qualified terms set out in the judgment and in relation to those wordings where the appeal affects the outcome. We allow the Hiscox Interveners' appeal on ground 1 and on grounds 2 and 3 in the qualified terms set out in the judgment. The parties are encouraged to agree appropriate declarations and orders to be made in the light of the judgment.

314.         I have read in draft, with admiration, the joint judgment of Lord Hamblen and Lord Leggatt, with which Lord Reed agrees. I will refer to them as the majority. I agree with all the conclusions to which the majority have come. Subject to one major and one minor point, I also agree with their reasoning. The practical effect of their analysis is that all of the insuring clauses which are in issue on the appeal to this court (not including those clauses where the issues appealed by the FCA are academic) will provide cover for business interruption caused by the COVID-19 pandemic, and that the trends clauses will not cut it down in the calculation of the amounts payable.

315.         The contrary view, contended for by most of the insurers, was in the main that, either because of the narrow geographical limit of cover imposed by the radius limitations, or because of the express “but for” provision in most of the trends clauses, there was little or no effective cover for business interruption caused by the national response to COVID-19. This was not, of course, a disease which anyone could have had specifically in mind when the policies in issue were written and marketed. But it is clear from the use of the definition of a “Notifiable Disease” in most of the relevant clauses, and equivalent wording in the remainder, that COVID-19 (when it appeared) fell squarely within the types of disease for which all the relevant disease and hybrid clauses provided cover: see in particular paras 51 to 53 of the majority judgment. Thus it cannot be said that cover for business interruption attributable to the reaction to a disease like COVID-19 lay outside the purview of the policies in issue.

316.         The consequence therefore is that, on the insurers' case, the cover apparently provided for business interruption caused by the effects of a national pandemic type of notifiable disease was in reality illusory, just when it might have been supposed to have been most needed by policyholders. That outcome seemed to me to be clearly contrary to the spirit and intent of the relevant provisions of the policies in issue. It therefore comes as no surprise to me that all the judges who have considered these issues have been unanimous in rejecting that outcome, albeit that this court has done so, rightly in my view, more comprehensively than did the court below. This is not to mis-use the dubious benefit of hindsight when applying an insuring clause to events which could not have been contemplated by the parties at the time of their bargain. As the majority show, there were a number of well-known notifiable diseases (such as cholera, plague, typhus, yellow fever and SARS) to which the relevant clauses clearly applied, all of which were capable of spreading rapidly and widely, so as potentially to cause a threat to health on a national scale, and to threaten a national reaction by the responsible authorities, leading to business disruption on a national scale.

317.         The major point on which, left alone, I would have reached the same conclusion as the majority but by a different route concerns the construction of what I shall label the “radius limitations”, both in the disease clauses and in the hybrid clauses. By that I mean the provisions which delimit cover for the effects of a notifiable disease (in this case COVID-19) by reference to its incidence within a circle defined by a 25 mile or 1 mile radius from the insured premises (“the radius”). The insurers' case was that (i) the radius limitations meant that only occurrences of COVID-19 within the radius were within the insured peril and (ii) that the application of a “but for” causation test to identify the loss for which compensation was payable would be bound to produce a nil return, because the national response to the pandemic would have been exactly the same even if there had been no outbreak within the radius.

318.         The preferred response of the judges in the court below was to treat the radius limitation not as defining the insured peril, which they identified as the COVID-19 pandemic (at least within the whole of the UK), but as a condition for cover which required the disease first to have spread within the radius. In the rare cases where there was no such spread (probably confined to insured premises in the Isles of Scilly and possibly parts of the Western Isles of Scotland), there would be no cover at all even if those premises suffered serious business interruption because of national restrictions on movement. In relation to the overwhelming majority of premises the radius limitation would merely define the date from which business interruption could constitute the basis for a claim. But the court below adopted as a fall-back analysis the identification of every case of COVID-19 as a separate insured peril, each just as causative of the national reaction, and leading thereby to consequential business interruption, as any other.

319.         In this court the majority have rejected the primary analysis of the court below, but in substance accepted and applied the alternative. They have, in effect, rescued the policyholders from the at first sight sombre consequences of a narrow definition of the insured peril by a principled application of the doctrine of concurrent cause, where (in settled insurance law) the existence of one or more concurrent causes of loss, other than the insured peril itself, does not prevent cover provided that the concurrent causes are not themselves expressly excluded. Encouraged by the court below they have done so, rightly in my view, by overruling the only reported case on comparable facts, namely the Orient-Express case. What is striking about the present case is that each and every separate case of COVID-19 is identified as an equally operative concurrent cause of the national response to the pandemic, so that the typical number of concurrent causes in the authorities, say two or three, is increased by orders of magnitude to something approaching, or even exceeding, a million. But there is no reason why the essential logic of concurrent cause cannot be scaled up in that way.

320.         That is not to say that the majority have insulated policyholders from the unfortunate consequences of a bad bargain (properly construed) by the healing balm of purely legal rules of causation. On the contrary, and again rightly in my view, the majority ground their treatment of concurrent causation firmly within the process of construction. The question whether particular consequential harm to a policyholder is subject to indemnity is as much a part of the process of interpreting their bargain as is the identification of the insured peril. It is therefore a quite distinct process from, for example, applying the law about causation and remoteness of loss for the purpose of identifying the harm liable to be made good by tortfeasors to their victims. In terms intelligible to non-lawyers, the question is: for what loss have the parties agreed that the insurers should compensate the policyholders as the result of the occurrence of the insured peril? Both the insured peril and the covered loss lie at the very heart of the contract of insurance, and the process of construction requires that they be addressed together.

321.         Left to myself I would not have departed from the adoption of both these two alternatives by the court below save that, since they both lead to the same conclusion, it is not necessary to prefer one over the other. As already noted, the underlying driver for some workable analysis is the need to avoid a conclusion that, as the insurers contend, clauses with radius limitations are a completely illusory form of cover for a pandemic type of disease falling clearly within the covered class. I recognise the force of the linguistic analysis by the majority which leads them to reject, as outwith the insured peril, any occurrence of COVID-19 outside the specified radius. I accept that the use of “occurrence” and similar words may lend some weight to a conclusion that every individual case, rather than the outbreak of the disease as a whole, is a separate insured peril. I acknowledge that the recognition of a million concurrent causes of equal potency is consistent with established authority, although an extension of it into new territory.

322.         My main reasons for thinking that the alternative construction, which treats COVID-19 as a whole as falling within the insured perils once it spreads within the specified radius, is as persuasive as that of the majority, are as follows. First, construction requires the court to put itself into the mind of the reasonably informed reader of the contract in issue, so as to understand what that hypothetical person (rather than an insurance lawyer) would think that the parties meant by the words which they have used, in the relevant context. To my mind, that person would ask: do clauses with the radius limitations provide cover for the adverse business consequences of a national reaction to a national pandemic disease? They would likely answer yes or no. But I am less confident that they would answer: No, there is cover only for the consequences of individual cases within the radius, but since they are each as causally potent as any other case or cases, cover is in practice provided for the national consequences of a national outbreak, because of the law about non-excluded concurrent cause. They might find it as easy to conclude that the parties meant that cover for the national disease was there, provided it reached, spread, encroached or extended (call it what you will) within the radius.

323.         Secondly, I would not be confident that the hypothetical reader would necessarily attribute the case by case specificity to the word “occurrence” or its synonyms given to it by the majority. Depending upon context, the word “occurrence” can properly be applied to happenings which do not take place at a single specified time, in a particular way and at a particular location. Thus a hurricane, a storm or a flood may properly be described as an occurrence even though each may take place over a substantial period of time, and over an area which changes over time. It is not in my view an inappropriate word to use about a pandemic disease as a whole, although I accept that it may be a pointer of some weight to an individual case analysis.

324.         Thirdly, the reasoning adopted by the majority might be regarded by the hypothetical reader as somewhat circuitous. Its end product is, as already noted, that cover is effectively provided for business interruption caused by a national reaction to a national pandemic, provided that it extends within the radius. If so, the reader might ask, why not interpret it as doing so directly? An illustration of the length of that journey may be provided by reading the majority judgment, but stopping at para 95. At that point in the analysis, the reader might think that the policyholders with radius-limited disease clauses had lost.

325.         None of this means that I consider that the analysis of the majority is not a readily available construction of clauses which contain a radius limitation. If it were the only means of giving those clauses real rather than illusory cover for diseases of a type which they clearly comprehend, then I would agree with the majority without qualification. It is often said that there can only ever be one “true” construction of any contractual provision. Where different constructions produce different results it may well be necessary to choose between them. But where, as here, the practical outcome is the same, there is to my mind no such necessity.

326.         I can deal with my one minor point of departure from the majority very briefly. It is that I would be cautious about treating the cases about defence costs (dealt with by the majority at paras 186 to 188) as of any general application outside their specific field. In my view, and notwithstanding that they can be viewed as consistent with a concurrent cause analysis, cases about insurance cover for costs are better regarded as sui generis.