infolaw (@infolaw.co.uk)

Gatwick Investment Ltd and others v Liberty Mutual Insurance Europe SE; Bath Racecourse Company Ltd and others v Liberty Mutual Insurance Europe SE and others - Find Case Law - The National Archives View download options Gatwick Investment Ltd and others (Appellants) v Liberty Mutual Insurance Europe SE (Respondent); Bath Racecourse Company Ltd and others (Appellants) v Liberty Mutual Insurance Europe SE and others (Respondents) Appellants – (1) Gatwick Investment Ltd (2) Millcroft Management Ltd (3) Sal Hotels Ltd (4) Serena Investments Ltd (5) Southampton Row Hotel LLP (6) London Victoria Hotel No 2 Ltd Jeffrey Gruder KC Josephine Higgs KC (Instructed by Edwin Coe LLP) Respondents – (1) Liberty Mutual Insurance Europe SE (2) Allianz Insurance Plc (3) Aviva Insurance Ltd David Scorey KC David Walsh KC Paul Davies (Instructed by DWF Law LLP, formerly by DAC Beachcroft LLP at the hearing) Appellants – Bath Racecourse Company Ltd and others Adam Kramer KC William Day Amelia-Rose Edwards (Instructed by Stewarts Law LLP) LORD HAMBLEN, LORD LEGGATT AND LORD BURROWS (with whom Lord Reed and Lord Briggs agree): In 2020 and 2021, the Covid-19 pandemic and resulting public health measures taken by the UK Government caused heavy financial losses to businesses around the country. Thousands of businesses made claims under insurance policies providing various types of business interruption cover. Issues arose as to whether Covid-19 related losses were covered by these policies. This court addressed a number of such issues in a test case: Financial Conduct Authority v Arch Insurance (UK) Ltd [2021] UKSC 1; [2021] AC 649 (“the FCA test case”). These appeals raise issues not decided in the FCA test case about the effect on business interruption insurance claims of furlough payments made by the UK Government under the Coronavirus Job Retention Scheme (“CJRS”). Payments made under this scheme reimbursed employers for a significant proportion of the wages and related costs of employing staff who were furloughed during the pandemic. The question is whether these payments go to reduce sums otherwise payable by insurers under a particular type of business interruption cover. The answer depends on the effect of “savings” clauses in the relevant policies. 3. The relevant policies contain two slightly different forms of savings clause, though it is agreed that nothing turns on the differences between them. Both are variants of a standard wording published by the Association of British Insurers. In what have been called the “Arena proceedings”, the relevant savings clause provides: “If any of the charges or expenses of The Business payable cease or reduce in consequence of the Damage such savings during the Indemnity Period shall be deducted from the amount payable.” In the “Gatwick proceedings”, the savings clause states that the amount payable as indemnity shall be: “less any sum saved during the Indemnity Period in respect of such of the charges of the Business payable out of Gross Revenue as may cease or be reduced in consequence of the incident.” The courts below held that, under these clauses, furlough payments made to policyholders under the CJRS are to be deducted from the amounts payable by insurers because these payments (1) reduced the charges or expenses of “the Business” payable out of “Gross Revenue” and (2) did so in consequence of the “Damage” / “incident”. We are told that this reflects the position taken by the insurance market generally and that an estimated £1 billion has been deducted on this basis. The appellant policyholders contend that the courts below were wrong so to decide. In outline, they submit that: (1) as wages and other employment costs had to be paid before being reimbursed by payments under the CJRS, these “charges” did not “cease” and were not “reduced”; and (2) the payments were not received “in consequence” of the insured peril since (a) proof of the insured peril was irrelevant to the entitlement to receive furlough payments and/or (b) the payments were collateral benefits of a gratuitous character and not to be regarded in law as caused by the insured peril. The 22 claimants and appellants in the Arena proceedings are subsidiary companies within the Arena Racing group. They operate racecourses, greyhound racing tracks, golf clubs, hotels and in one instance a pub in England and Wales. Companies in the Arena Racing group including the appellants entered into a composite contract of insurance for the 2020 calendar year (“the Arena policy”) underwritten by the respondent insurers. 7. The Arena policy covered material damage to insured premises and associated business interruption. The business interruption cover also included a “Denial of Access” clause, which so far as relevant provides: “This Section extends to include any claim resulting from interruption of or interference with The Business carried on by The Insured at The Premises in consequence of action by the Police Authority and/or the Government or any local Government body or any other competent authority following danger or disturbance within a one mile radius of The Premises which shall prevent or hinder use of The Premises or access thereto …” The insurers have admitted that they are in principle liable to indemnify the appellants under this cover for loss of “Gross Revenue” resulting from interruption of their businesses in consequence of actions taken by the UK Government in response to the Covid-19 pandemic. But they maintain that, in calculating the amount payable, credit must be given under the savings clause for furlough payments. Each of the six claimants and appellants in the Gatwick proceedings owns or operates a hotel in England. Each appellant entered into a separate Commercial Combined Policy with Liberty Mutual providing various forms of insurance cover for its hotel for the period from 8 October 2019 to (as extended) 20 October 2020. Each Gatwick policy provided cover against business interruption. This was extended by an endorsement to cover “Prevention of Access (Non Damage)” as follows: “Under Business Interruption loss following interference with the Business carried out by the Insured in consequence of action by the Police or other Statutory Authority following danger or disturbance within 1 mile of the Premises which shall prevent or hinder use of the Premises or access thereto or, interference with the Business carried out by the Insured.” Liberty Mutual originally denied cover under this clause on the ground that actions taken by the UK Government in response to Covid-19 were not actions by a “Statutory Authority”. That point was decided against Liberty Mutual by Jacobs J in his judgment dated 26 January 2024, referred to below, and is no longer pursued. But it remains in issue in these proceedings, as in the Arena proceedings, whether in calculating the amount payable as indemnity credit must be given for furlough payments. This is described in detail in the judgment of Jacobs J, at paras 16–48. The following is a short outline. On 20 March 2020, the Prime Minister and the Welsh First Minister gave instructions that public venues including cafés, pubs, bars and restaurants (including those in hotels) should close that night. Those instructions were implemented the next day in England and Wales by statutory regulations. On 23 March 2020, the Prime Minister announced the first nationwide lockdown. The Welsh First Minister made a similar statement the same day. On 25 March 2020, the Coronavirus Act 2020 was enacted. In broad terms this Act established emergency arrangements in response to the Covid-19 pandemic including measures in relation to health workers and food supply, and powers in relation to potentially infectious persons and to control events or gatherings. Section 76 of the Act specified that HMRC was to have such functions as the Treasury may direct in relation to Covid-19, and section 71 modified the signature requirements for Treasury Instruments. On 26 March 2020 new regulations were made which replaced those issued on 21 March 2020. These regulations required the continued closure of businesses that had already been required to close (including restaurants, cafés and bars, which were prohibited from selling any food or drink other than for consumption off-premises) as well as the closure of further business premises. The regulations also prohibited hotels from offering accommodation to guests save for certain very limited and specific exceptions. On 4 July 2020, the first lockdown ended in England and new regulations were introduced imposing more limited restrictions. Pubs, bars and restaurants were permitted to serve food and drink for consumption on the premises; hotels could provide accommodation to all; and gatherings of up to 30 people were permitted. However, there were strict social distancing and cleansing requirements which limited the number of guests and the Government was still recommending that people avoid gatherings larger than six. The July regulations were revoked on 14 October 2020, when a three-tier regional system was introduced. On 31 October 2020, the Prime Minister announced a second lockdown which was implemented on 5 November 2020 by further regulations. These were in turn replaced by regulations issued on 2 December 2020, together with a “Winter Plan” published by the UK Government on 23 November 2020. It is common ground on these appeals that the measures mentioned above were taken by the Government and other “statutory” or “competent” authorities in response to the Covid-19 pandemic and that they prevented or hindered use of, or access to, relevant premises insured under the Arena policy and under the Gatwick policies during the policy periods. The precise restrictions varied, depending on the time, location and nature of the premises. The CJRS was first announced by the Chancellor of the Exchequer, Rishi Sunak, in a speech on 20 March 2020, when he said: “This week, the Government has taken unprecedented steps to fight the coronavirus. We have closed schools. We have told people to stay at home to prevent the spread of infection. We are now closing restaurants and bars. Those steps are necessary to save lives. But we don’t do this lightly—we know those measures will have a significant economic impact. I have a responsibility to make sure we protect, as far as possible, people’s jobs and incomes. Today I can announce that, for the first time in our history, the government is going to step in and help to pay people’s wages. We’re setting up a new Coronavirus Job Retention Scheme.” The CJRS was implemented, extended and amended by a series of Treasury Directions made under sections 71 and 76 of the Coronavirus Act 2020. As stated in the Treasury Directions, the purpose of the CJRS was “to provide for payments to be made to employers on a claim made in respect of them incurring costs of employment in respect of furloughed employees arising from the health, social and economic emergency in the United Kingdom resulting from coronavirus and coronavirus disease”. 22. The terms of the CJRS were first set out in a Treasury Direction dated 15 April 2020, which introduced the scheme with effect from 1 March 2020 until 31 May 2020. To qualify for payments under the CJRS, an employer had to have a scheme registered for Pay As You Earn (“PAYE”). Payments could be claimed in respect of furloughed employees. As defined in the Treasury Direction of 15 April 2020 (para 6.1), an employee was a “furloughed employee” if: the employee has been instructed by the employer to cease all work in relation to their employment, the period for which the employee has ceased (or will have ceased) all work for the employer is 21 calendar days or more, and the instruction is given by reason of circumstances arising as a result of coronavirus or coronavirus disease.” The expenditure to be reimbursed was defined as (para 8.1): the gross amount of earnings paid or reasonably expected to be paid by the employer to an employee any employer national insurance contributions liable to be paid by the employer arising from the payment of the gross amount the amount allowable as a CJRS claimable pension contribution.” Under the Treasury Direction of 15 April 2020, an employer was entitled to make a claim for payment/reimbursement from HMRC of 80% of the wages of a qualifying employee (up to £2,500 per month) together with employer national insurance contributions and any amounts allowable as CJRS claimable pension contributions. Subsequent Treasury Directions extended the duration of the scheme on broadly similar terms but with variations relating, in particular, to the level of support provided to employers (for example, the percentage of the wages claimable reduced and reimbursement of employer national insurance contributions and pension contributions could not be claimed for periods after 31 July 2020). The CJRS ended on 30 September 2021. To make a valid claim for payment under the CJRS (or to avoid incurring an obligation to repay the sum claimed), the employer had to pay the full amount of the wages and other costs of employment claimed in respect of a qualifying employee. At various times from April 2020 onwards, each appellant in the Arena and Gatwick proceedings successfully claimed payments (covering, for example, 80% of wages as set out in para 24 above) under the CJRS. To qualify for these payments, the appellants “furloughed” employees by instructing them to cease work. It is an agreed fact that, had it not been for the receipt of payments under the CJRS, the appellants would have made some employees redundant, thereby saving on the various costs of employment. A trial of preliminary issues in the Arena and Gatwick proceedings took place before Jacobs J. Judgment was given on 26 January 2024: [2024] EWHC 124 (Comm); [2025] Lloyd’s Rep IR 82. The judge decided the issues about the effect of the savings clauses in favour of the insurers. An appeal from the judge’s decision on these issues was dismissed by the Court of Appeal (Sir Julian Flaux C, Popplewell and Phillips LJJ) for reasons given in a judgment handed down on 21 February 2025: [2025] EWCA Civ 153; [2025] Lloyd’s Rep IR 353. The lead judgment was given by the Chancellor, with which the other members of the court agreed. The policyholders appeal, with the permission of this court, on the grounds that: Under the savings clauses, employee costs were not “reduced” by payments made by the Government to the appellants under the CJRS (the “construction issue”); and Those payments were not legally caused by the insured peril and so were not “in consequence of” it because: (a) proof of the insured peril was irrelevant to the entitlement to claim CJRS payments; and/or (b) the payments were of a gratuitous, benevolent or voluntary nature and should for that reason be treated as collateral benefits (the “causation issue”). 31. On this issue, Jacobs J followed the earlier decision of Butcher J in Stonegate Pub Co Ltd v MS Amlin Corporate Member Ltd [2022] EWHC 2548 (Comm); [2023] Bus LR 28 (“Stonegate”), with which he expressed agreement. Addressing a savings clause in materially similar terms to those in issue on these appeals, Butcher J rejected the policyholders’ argument that employment costs had not been “reduced” because the policyholders had incurred and paid those costs, albeit the costs had then been reimbursed by furlough payments. He held as follows, at para 258: “In my judgment, employment costs were at least ‘reduced’ pro tanto by reason of the payment of corresponding amounts under the CJRS. I consider that the natural meaning of [the savings clause] is that it is referring to costs to the business. Insofar as such costs were defrayed by the Government, I consider that they were ‘reduced’. That, in my view, reflects the net financial effect of payments under the CJRS and the commercial reality.” 32. The Court of Appeal in the present proceedings agreed with this interpretation. The Chancellor stated as follows, at para 174: “Looking at the substance, the commercial and economic reality by reference to which these commercial insurance contracts should be construed, is that the effect of the CJRS payments reimbursing the insureds for 80% of their wages bill was indeed to reduce that bill by 80%, which constituted a saving within the savings clauses. It is no answer to that commercial and economic reality, as the insureds seek to contend, that the insureds still paid the wages and that the funds reimbursing the insureds came from a third party, the Government. The bottom line at the end of the day is that the insureds did not have to bear the expense of the wages bill and to that extent, the charges or expenses of the business were reduced.” 33. The appellants argue that the courts below (and Butcher J in Stonegate) misconstrued the savings clauses. In outline, they contend that, on the natural and ordinary meaning of the words, the clauses are only triggered where a liability of the insured (a “charge” or “expense” which is “payable”) is not incurred (“cease”) or is incurred in a lesser amount (“reduced”). The following points are made: The savings clauses refer to “charges” (and, in the Arena policy, to “expenses”) which are “payable”. That denotes payments contractually or otherwise legally due to another (see the online Oxford English Dictionary definition of “payable”: “that is to be paid; due, owing; falling due”). A liability for a “charge” or “expense” will only “cease” or “reduce” where it is not incurred, fully or partially. In the present cases, the liability of a policyholder to pay its employees’ wages and related costs was incurred and was discharged by payment. Later reimbursement by payments under the CJRS did not change that. At all times the liability to pay the wages and related costs was unaffected and did not “cease” or “reduce”. The furlough payments represented “other income” and did not alter the policyholder’s liability for the cost of wages. The wages bill remained constant. That that is the commercial reality is borne out by the fact that such payments were treated as income by the UK tax authorities and under the Financial Reporting Standard applicable in the UK (FRS 102). Such income falls outside the policies since furlough payments were not “Gross Revenue” because they were not money payable “for work done and services provided”. Taking reimbursement of paid wages and related costs into account would lead to arbitrary results which cannot have been intended. For example, it would mean that whether or not the saving was deductible would depend on the fortuity of the timing of the reimbursement and whether or not it fell within the policy period. Moreover, in a case where the furlough payments were repaid by employers it would mean that a charge or expense which had been “reduced” by reimbursement would somehow “unreduce”. 34. The appellants also submit that the courts below erred in placing undue reliance on the general principle that a contract of insurance is a contract of indemnity. The recoverable loss under a business interruption policy is an intangible unlike, for example, property damage. To minimise dispute, the loss is defined by an agreed formula. That formula may over- or under-indemnify but it is to be applied according to its terms. As Branson J stated in Polikoff Ltd v North British & Mercantile Insurance Co Ltd (1936) 55 Ll L Rep 279 at 284: “… one has to remember that this specification is a formula upon which the parties have agreed, and it is not for the Court to try to find out a better formula, or one which would give a more favourable result to the one party or to the other than that to which the parties have set their hands.” See also, to the same effect, the decision of the New South Wales Court of Appeal in Mobis Parts Australia Pty Ltd v XL Insurance Co SE [2018] NSWCA 342; [2019] Lloyd’s Rep IR 162 (“Mobis”), para 147. The appellants contend that their case is further supported by the decision of M Osborne J in the Supreme Court of Victoria (Commercial Court) (Australia) in Princess Theatre Pty Ltd v Ansvar Insurance Ltd [2024] VSC 363 (“Princess Theatre”), in which it was held that payments made under the Australian JobKeeper scheme did not fall within the savings clause. The judge stated, at para 466: “The wages were not reduced and did not cease. The premise of the making of JobKeeper Payments was to enable employers … to be able to pay wages and expenses of their employees. Because the wages and expenses were maintained … the ‘sum saved’ component has no application. The fact that the JobKeeper Payments constituted in effect a new revenue stream which could be utilised to pay these expenses is a different issue.” The respondents support the reasoning and conclusion of the courts below and of Butcher J in Stonegate. They emphasise that a reasonable person in the position of the parties to these insurance contracts would look at the terms of the savings clauses in a pragmatic and commonsensical way, not in an overly legalistic way. They would look at whether, in substance and as a matter of economic reality, a particular charge or expense had ceased or reduced. They would not be fixated with whether, as a matter of law, the liability for the particular charge or expense had been affected. Nor would they draw distinctions between legal liabilities and funding for those legal liabilities. Counsel for the respondents also emphasise that payments made under the CJRS were not analogous to a general government grant. The payments had to be spent on wages; the employer had no discretion. To that extent, they were hypothecated and the Government was bearing a specified share (initially 80%) of the cost of the employees’ wages. By doing so, those costs were “reduced” pro tanto within the meaning of the savings clause. It is not in dispute that the correct approach to construction is as summarised in the FCA test case, at para 47. Having referred to the discussion in the judgment of Lord Hodge JSC in Wood v Capita Insurance Services Ltd [2017] UKSC 24; [2017] AC 1173, this court stated: “The core principle is that an insurance policy, like any other contract, must be interpreted objectively by asking what a reasonable person, with all the background knowledge which would reasonably have been available to the parties when they entered into the contract, would have understood the language of the contract to mean.” There was some discussion at the hearing of the present appeals of whether an observation made in para 77 of the FCA test case judgment involved a development of this standard approach. It was there said that: “the overriding question is how the words of the contract would be understood by a reasonable person. In the case of an insurance policy of the present kind, sold principally to SMEs, the person to whom the document should be taken to be addressed is not a pedantic lawyer who will subject the entire policy wording to a minute textual analysis. It is an ordinary policyholder who, on entering into the contract, is taken to have read through the policy conscientiously in order to understand what cover they were getting.” (Internal citations omitted.) The context in which this observation was made was an argument that the cover provided by the disease clause in the business interruption section of the policy in question had been cut down by wording buried away in the middle of a general exclusion of contamination and pollution risks at the back of the policy. The above passage was explaining why such a restriction would need to be expressed transparently. It was not seeking to alter or refine the “core principle” set out in para 47 of the judgment. In particular, it was not seeking to suggest that the objective approach to construction involves focusing on one party, the “ordinary policyholder”, rather than the construct of a reasonable person in the position of the parties. That said, it is true that in most cases—including the insurance contracts with which we are concerned—such a person will not be a pedantic lawyer. A further issue raised at the hearing was the relevance of the principle that a contract of insurance is generally intended to be a contract of indemnity. 42. Butcher J in Stonegate and Jacobs J and the Court of Appeal in the present case saw this principle as supporting their conclusion on the proper construction of the savings clauses. Reliance was placed on the decision of Flaux J in Synergy Health (UK) Ltd v CGU Insurance plc [2010] EWHC 2583 (Comm); [2011] Lloyd’s Rep IR 500 (“Synergy”). That case also concerned the construction of the savings clause in a business interruption policy. The issue was whether a deduction from gross profit for depreciation of plant and machinery, which ceased as a result of the insured peril (a fire at the premises), was a charge or expense “payable” out of gross profit, given that depreciation is an accounting entry and is never paid to anyone. In holding that it did fall within this wording, Flaux J relied on the fact that otherwise the insured would recover more than a full indemnity and stated, at para 258: “Although the defendants’ construction stretches the word ‘payable’ somewhat, it seems to me that it is to be preferred to [the insured]’s construction, which leaves the saving in respect of depreciation out of account. My principal reason for that conclusion is that it seems to me that, as a matter of principle, a policy should be interpreted as providing an indemnity for the loss suffered not for more than such an indemnity. Of course if the wording is incapable of any other construction, a court might be driven to the conclusion that something in excess of a full indemnity was intended, but given the unlikelihood and unreasonableness of such a conclusion, the court should not arrive at it unless no other conclusion is possible.” The appellants contend that this “indemnity principle” illegitimately introduces a presumptive rule of interpretation when the primary focus should always be on the words used without any presumption as to what they may mean. We do not so interpret this passage. In our view, Flaux J was not purporting to articulate or apply a special legal rule. He was simply having regard to the object or purpose of the contract as being to indemnify the policyholder against loss. It is an entirely conventional approach when the language of a contract has more than one possible meaning to prefer the meaning which better accords with its commercial purpose: see, eg, Rainy Sky SA v Kookmin Bank [2011] UKSC 50; [2011] 1 WLR 2900, paras 21 and 43. 44. In Mobis the New South Wales Court of Appeal considered that Flaux J in Synergy had placed too much emphasis on the indemnity principle and reached a different conclusion as to whether depreciation was deductible under the savings clause in that case. Meagher JA recognised, at para 146, that, since the general object of the policy was to indemnify against loss of gross profit, “the prospect of under- or over-indemnification may colour the meaning of the language used”, referring to Brett LJ’s judgment in Castellain v Preston (1883) 11 QBD 380, 386. He considered, however, that Flaux J’s reasoning in Synergy “gives the indemnity principle unwarranted effect in the face of the language of the policy, and the specific objects of the provisions for the assessment of loss”. He further stated, at para 149: “A reasonable businessperson seeking to understand these lengthy clauses would not begin by assuming that they mean nothing more than the expression “full indemnity for actual loss to gross profit”, and then proceed to enquire whether anything in the language required otherwise. His or her attention would remain fixed on the sense of the language describing the method for ascertaining the loss as coloured by its immediate and commercial context …” 45. In Hyper Trust Ltd Trading as the Leopardstown Inn v FBD Insurance plc [2023] IEHC 455 (“Hyper Trust”), McDonald J in the Irish High Court considered the apparently contrasting views of Flaux J and Meagher JA about the role of the indemnity principle. He concluded as follows, at paras 70 and 72: … I can readily accept that the indemnity principle is a factor to be taken into account in interpreting the terms of a policy. But, if it is clear from the language of the policy that the parties have agreed to accept either something less—or something more—than a perfect indemnity, the ordinary rules of contractual interpretation will require that effect be given to the solution expressly chosen by the parties. … I accordingly accept that the indemnity principle is relevant in construing the [insurance] policy in so far as it may colour the meaning of the language used. However, … the indemnity principle does not require that a strained or artificial meaning should be given to the words of a policy …” 46. We consider that the differences between the judgments of Flaux J in Synergy and Meagher JA in Mobis have been overstated. Both judgments recognise that the indemnity principle is to be taken into account in interpreting the terms of an insurance policy. Both judgments also recognise that there will be no room for the indemnity principle to apply if the policy language clearly requires otherwise. In our view, it is both permissible and correct to take into account the indemnity nature of a contract of insurance when construing it. Indemnification is the purpose of most types of insurance policy, including business interruption policies. Such policies may contain a formula by which the indemnity is to be calculated but this does not detract from that underlying purpose. That being so, in circumstances where a clause relating to the quantification of loss has more than one possible meaning, it is appropriate to have regard to the purpose of indemnification and to adopt the construction which is more consistent with that purpose. There is no dispute that wages and related costs are “charges” or “expenses” under the relevant policies and that they were “payable”. There is also no dispute that “cease” means “come to an end” and that “reduce” means “lower, diminish or lessen” (as stated in the online Oxford English Dictionary). In the present case we are concerned with whether the relevant charges or expenses were reduced rather than with whether they ceased. The appellants contend that charges or expenses which are “payable” are only “reduced” if the liability to pay them is lowered, diminished or lessened. They say that the clause is concerned with, and refers to, legal liability for the charge or expense. The respondents contend that charges or expenses are “reduced” if the amount of the charge or expense that the insured has to bear is lowered, diminished or lessened. They say that the clause is concerned with and refers to the economic burden of the charge or expense. We consider that, as a matter of language, either construction of the savings clauses is a possible meaning. The appellants’ case has some linguistic support from the use of the word “payable”. On the other hand, the clauses do not refer to liability for the charge or expense or to the incurral of such a charge or expense, either of which would more clearly suggest that the clause was only concerned with legal liabilities. As it is, the language used in the savings clauses could be understood as referring either to: (i) whether as a matter of fact the charge or expense is reduced; or (ii) whether liability for the charge or expense is reduced as a matter of law. Although either meaning is a possible meaning of the words, we have no doubt that the respondents’ construction of the clause is to be preferred for the following eight reasons. First, no reason has been put forward to explain why a reasonable person in the position of the parties should choose to differentiate between the incurring of a charge or expense and the bearing of a charge or expense. The economic effect of paying a charge and expense and being funded or reimbursed for that payment is the same as when the charge or expense is paid by another party on behalf of the obligor or the obligation to pay is waived. So, for example, in relation to national insurance, the furlough scheme could well have provided that the contribution otherwise due from the employer need not be paid. Instead, it provided for the contribution to be paid and reimbursed. That, however, is a matter of mechanics. It does not alter the economic outcome. Second, the general context is that the policy is providing business interruption insurance. It is a policy which is concerned with the economic effects of insured perils on the policyholder’s business. As such, one would expect a clause addressing savings to the business to be concerned with the economic effect of the saving in question. 54. Third, the specific context is a clause concerned with savings. The evident purpose of the savings clause is to prevent over-indemnification—ie to prevent the policyholder from recovering more than the actual loss suffered as a result of the insured peril. Because of the way in which the amount payable by the insurers is calculated in such policies, in this case based on “Gross Revenue”, such over-indemnification will occur if the interruption actually reduces costs during the indemnity period and these “savings” are not deducted from the amount calculated as payable. To take an example given in Riley on Business Interruption Insurance, 11th ed (2021), at para 2.79, if some of the buildings forming part of the premises are destroyed and no alternative accommodation can be found several expenses relating to the destroyed buildings will cease. If the policyholder were simply to recover the revenue lost as a result of the destruction of the premises without taking account of the expenses saved: “This would be more than an indemnity; it would in fact be making a profit out of the incident. Hence, the insertion of the savings clause to redress the position by stipulating that any sum saved in this way shall be deducted in the computation of the claim.” The purpose of avoiding over-indemnification is served if all savings which are in fact made are taken into account, but it is defeated if savings actually made are ignored because they did not reflect a reduced legal liability. Fourth, as discussed above, where there are two possible meanings of a clause in an insurance contract it is appropriate to prefer the meaning which better fits the purpose of indemnification. This is all the more so where that is also the purpose of the specific clause in question. Fifth, the appellants’ construction leads to arbitrary results which are most unlikely to have been intended. For example, it is agreed that, had it not been for the receipt of furlough payments, the appellants would have made some employees redundant thereby saving on the various costs of employment which would have reduced their insurance claim. On the appellants’ case, they not only get the benefit of those payments, but they can also get full insurance recovery without regard to the savings that would have been realised if those payments had not been made. That would be a perverse and wholly uncommercial bargain to make. Sixth, the reality that the Government was paying employees’ wages is borne out by the Government’s own description of the CJRS scheme. When announcing the scheme on 20 March 2020 Rishi Sunak described how the scheme meant that “the government is going to step in and help to pay people’s wages”: see para 20 above. Seventh, the respondents’ patently more commercial construction of the savings clause has been unequivocally preferred by five experienced former and current judges of the Commercial Court (Butcher and Jacobs JJ, Sir Julian Flaux C, Popplewell and Phillips LJJ). It has also been adopted by McDonald J in the Irish High Court in Hyper Trust who expressed the following opinion, with which we agree, at para 50: “I can see that a lawyer might construct an argument that the effect of the Government supports is to make good outlays already expended by the insured rather than to save or reduce those outlays in advance. However, I do not believe that this is how the savings clause would be construed by a reasonable person in the position of the parties.” 60. Eighth, the artificiality of the contention that CJRS payments did not reduce employment costs is further highlighted by the fact that payments could be made to the employer under the scheme before the amounts were paid to the employee. As noted in Stonegate, para 265: “The Treasury Direction of 15 April 2020, by para 8.1 specified that CJRS payments might reimburse ‘the gross amount of earnings paid or reasonably expected to be paid by the employer to an employee’. There is similar wording in para 8.6; para 8.1(b) refers to national insurance contributions ‘liable to be paid’; and para 12 refers to ‘earnings paid or payable’ by employers to furloughed employees.” (Emphasis added by the judge.) We agree with Butcher J’s observation, at para 266, that the question of whether CJRS payments fall to be taken into account under the savings clause cannot depend on whether payments were received before or after the payment to the employee. 61. In relation to particular points raised by the appellants not already addressed: Reliance on how the payments under the CJRS are treated by the tax authorities or by accountants is beside the point. What matters is what has been agreed by the parties, not how the payments may be treated by other parties for other purposes. In any event, as noted by Butcher J in Stonegate, paras 260–261, under both US GAAP and International Financial Reporting Standards, as adopted by the EU, furlough payments could be offset against wage expenses rather than being recorded as “other income”. The point made by the appellants in relation to the timing of reimbursements (see para 33(vi) above) is correct but is simply a reflection of the line drawn by a finite policy period. It is a commonplace that this may result in apparently arbitrary consequences in relation to claims or expenses. To the extent that the judge’s reasoning in the Princess Theatre case (in the short passage quoted at para 35 above) differs from our own, we disagree with it. For all these reasons we agree with the conclusion of the courts below that, on the proper construction of the savings clauses, payments under the CJRS “reduced” the charges or expenses of the policyholder’s business. On this basis we turn to the second issue raised on these appeals: whether this reduction occurred “in consequence of” the “Damage” or “incident”. It is not in dispute that the word “Damage” (in the Arena policy) and the word “incident” (in the Gatwick policies) are to be construed as referring to the insured peril: here the peril covered by the prevention of access clause in the relevant policy. 64. It is clear that the words “in consequence of” import a causation requirement. It is common ground between the parties that the requirement is one of proximate causation and that the same test of proximate causation applies both to the link between the insured peril and loss under the insuring clause and to the link between the insured peril and the reduction of expenses under the savings clause. As explained in the FCA test case, at paras 162–163, the concept of “proximate” causation was developed by the common law (and codified in section 55(1) of the Marine Insurance Act 1906) to explain the connection that must generally be shown between loss and the occurrence of an insured peril. In this case, we are not dealing directly with causation of loss, but rather with gains received (the furlough payments) by the insured. But we agree with the parties that, logically, the same test of causation applies to gain as it does to loss. Therefore, whether savings were made “in consequence of” the insured peril turns on whether the insured peril was a proximate cause of the gain. Because it provides the context for the appellants’ argument, it is helpful to explain how issues of causation of loss were decided in the FCA test case. In general terms, the “disease clauses” considered in the FCA test case provided cover for business interruption loss caused by any occurrence of a notifiable disease within a specified distance (usually described as a “radius”) of the policyholder’s business premises. The majority of the Supreme Court decided that, on the proper interpretation of these clauses, each case of illness sustained by an individual resulting from Covid-19 was a separate “occurrence” of a notifiable disease: see paras 67–74 of the judgment. 66. A central plank of the insurers’ case was that, to establish the required causal connection between one or more cases of disease occurring within the radius and business interruption loss, it was necessary to show, at a minimum, that the loss would not have been sustained but for the occurrence of the insured peril: see para 177. The insurers argued that this “but for” test of causation could not be met because, on the facts, the lockdown and other measures taken by the UK Government which led to business interruption losses were taken in response to a national emergency affecting the country as a whole. It could not be shown that, but for cases of Covid-19 occurring within a particular local area, the policyholder would not have suffered the losses that it did. To the contrary, it was evident that the same or similar losses would have been sustained anyway because in this hypothetical situation the Government would still have imposed the same nationwide restrictions as it did in response to all the other cases of Covid-19 elsewhere in the country: see para 179. So, the insurers argued, the losses were not caused by the insured peril. 67. The Supreme Court rejected this argument. The court held that, on the proper interpretation of the disease clauses, in order to show that loss from interruption of the insured business was caused by one or more occurrences of illness resulting from Covid-19, it was enough to prove that the interruption resulted from 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: para 212. The basis for this conclusion was that each individual case 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). That analysis, in the court’s view, best gave effect to the presumed intention of the parties to the insurance contracts. A key part of the court’s reasoning (set out in paras 193–197 of the judgment) was that it would be contrary to the commercial purpose of the cover to treat the causative impact of the insured peril (cases of a notifiable disease occurring inside the specified radius) as diminished or outweighed by matters (cases of the disease occurring outside the radius) which could naturally be expected to occur concurrently with the insured peril. 68. This analysis was also applied to “prevention of access” and “hybrid” clauses similar in structure to the clauses under which the present claims are made. Under these clauses, the occurrence of disease within a specified radius of the insured premises is merely the first element of a composite insured peril. To attract an indemnity, the elements of the insured peril must occur in a causal sequence which progressively narrows its scope. Taking the “Denial of Access” clause in the Arena policy as an example, the elements of the insured peril are: (A) “danger or disturbance” (a description that includes any case of Covid-19) within a one mile radius of the premises, which causes (B) action by the Government or any other competent authority, which causes (C) prevention or hindrance of use of, or access to, the premises, which causes (D) interruption of or interference with the policyholder’s business. Consistently with the interpretation of the disease clauses, the first step in this causal chain is satisfied by showing that the relevant Government action was taken in response to cases of Covid-19 which included at least one case occurring within the radius. In relation to the further steps in the chain, the insurers in the FCA test case again argued that a “but for” test of causation should be applied. They asserted that, to show that the relevant Government action caused an inability to use the insured premises and consequent business interruption loss, the policyholder had to prove that the business interruption loss would not have occurred but for the Government action in so far as it prevented use of those particular premises. On this approach the relevant counterfactual was one in which the measures actually taken by the Government were taken but did not apply (exceptionally) to the policyholder’s business premises: see paras 221–227. 70. Again, the court rejected this approach, observing that no reasonable policyholder would have understood the insurance cover which it was getting to be insurance against such a narrow and fanciful risk: para 227. The insurers’ argument had the same essential vice as the argument concerning the causal impact of the occurrence of disease. It sought to rely on matters which were inherently likely to occur as a result of the same underlying fortuity to restrict the scope of the indemnity. Thus, it was entirely predictable and to be expected that if an outbreak of a notifiable disease that included cases within the specified radius led to the Government taking measures that required the policyholder’s business premises to close, those measures would have other adverse effects which would have caused interruption of the business even if the premises had not been required to close. As stated at para 237: “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.” At para 239 of the judgment, the principle was stated as being that: “where insurance is restricted to particular consequences of an adverse event … the parties do not generally intend other consequences of that event, which are inherently likely to arise, to restrict the scope of the indemnity.” Applying this principle—which for convenience we will call the “same underlying fortuity” principle—the following conclusion was drawn, at para 243 of the judgment. Properly interpreted, a clause such as a prevention of access clause providing cover against a composite peril: “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.” A similar analysis was applied to the “trends clauses” found in many policies. The court held that, in applying the trends clauses to claims arising out of the Covid-19 pandemic, adjustments to turnover should be made only to reflect trends or circumstances unrelated to the pandemic and not to reflect other (uninsured but non-excluded) consequences of the pandemic which would naturally be expected to occur concurrently with the insured peril: see paras 285 and 288. As already noted, the relevant insuring clauses in the Arena and Gatwick policies are similar in structure to the “prevention of access” and “hybrid” clauses considered in the FCA test case. It is accepted on these appeals that all the elements of the composite insured peril occurred and that the policyholders are therefore entitled to be indemnified for business interruption losses in accordance with the analysis in the FCA test case, summarised above. The starting point, therefore, is that (A) at least one case of Covid-19 occurring within a one mile radius of the policyholder’s premises caused (B) action by the Government (comprising the lockdown and other restrictions outlined at paras 12–18 above), which caused (C) prevention or hindrance of use of, or access to, the premises, resulting in (D) interruption of the policyholder’s business. On the face of it, the causal connection between the business interruption loss falling within this cover and the savings from furlough payments is straightforward. Take the example of a hotel closed as a result of Government action (such as the regulations issued on 26 March 2020) following the occurrence of one or more cases of Covid-19 within a mile of the hotel premises. An inevitable consequence of the closure of the hotel was that the policyholder was liable to continue paying wages and other costs of employing staff whose services were now not needed. In the absence of the CJRS, a natural response to this situation would have been to make some employees redundant. Indeed, it is an agreed fact on these appeals that, had it not been for the receipt of furlough payments, this is what the policyholders would have done, thereby saving on various costs of employment which would have ceased or reduced (see para 27 above). As it was, the existence of the CJRS altered the calculus—as it was designed to do—by making it financially advantageous for employers to “furlough” relevant employees and claim reimbursement under the scheme. Thus, in light of the CJRS, the predictable and direct consequence of the occurrence of the insured peril was that the policyholder instructed employees to cease work but continued to employ them while claiming reimbursement of the costs of doing so (or a large proportion of those costs) under the CJRS. The expenses of employing the relevant employees were thereby reduced, just as they would have been if the employees had been made redundant. The effect of the savings clause was to require the sums saved in this way to be deducted in calculating the business interruption losses recoverable under the policy. On the face of it, the sums saved were a consequence of, ie were proximately caused by, the insured peril. Had it not been for the skill of the appellants’ counsel in arguing otherwise, we would have thought this conclusion indisputable. That was evidently the view taken by the policyholder in Stonegate, where it was conceded by the end of the hearing before Butcher J that furlough payments had been a consequence of the insured event: see para 256. But in these proceedings the conclusion is disputed. Two arguments are advanced. They are that the furlough payments were not received in consequence of the insured peril because, in short: (1) the existence and proof of the insured peril was irrelevant to the policyholders’ entitlement to the payments (the “irrelevance argument”); and (2) the payments were a gratuitous, benevolent or voluntary conferral of a benefit by a third party (the “collateral benefits argument”). 78. The first argument relies on the fact that the criteria for claiming payments under the CJRS were not the same as the requirements for claiming an indemnity under the prevention of access cover. As described earlier in this judgment (see paras 22–23 above), the main requirements for claiming payments under the CJRS were that the employer (1) had a registered PAYE scheme and (2) had “furloughed” the relevant employee(s) by instructing them to cease work because of “circumstances arising as a result of coronavirus or coronavirus disease”. This description clearly included circumstances which gave rise to a claim under the prevention of access cover. But the entitlement to claim furlough payments was not limited to such circumstances. In particular, it was not a condition of eligibility to claim payments under the CJRS that access to, or use of, the employer’s premises had been prevented or hindered, or its business interfered with, by Government action. Nor was it a requirement that at least one case of Covid-19 had occurred within one mile of the employer’s premises. The appellants submit that, in these circumstances, the furlough payments which they received under the CJRS cannot be regarded as having been proximately caused by the composite insured peril. This is said to follow from the fact that such payments could be claimed irrespective of whether use of or access to the premises had been prevented or hindered, and irrespective of whether any case of Covid-19 had occurred within a one mile radius of the premises. Accordingly, it is said, the existence and proof of the insured peril was “irrelevant” to the entitlement to receive furlough payments. To reinforce this argument a hypothetical example is posited of a policyholder who owns a hotel in a rural area which was subject to the nationwide restrictions imposed by the Government and suffered a reduction in revenue as a result of the impact of Covid-19, but in circumstances where it so happened that no case of Covid-19 occurred within a one mile radius of the hotel. Such a policyholder would have no claim under the prevention of access cover. But business prudence would no doubt still have led this hotelier to furlough some or all of its staff and claim payments under the CJRS. This is said to show that there was no “inherent causal link” between such payments and the insured peril. 81. The appellants also rely on a decision of the Full Court of the Federal Court of Australia which considered the effect of payments under the Australian JobKeeper program. In LCA Marrickville Pty Ltd v Swiss Re International SE [2022] FCAFC 17; (2022) 290 FCR 435 (“LCA Marrickville”) a travel agency (“Meridian”) had business interruption cover which applied when an outbreak of disease occurred within a 20 kilometre radius of its premises. The indemnity provisions of the policy included a savings clause similar to those in issue here. A question arose as to whether JobKeeper payments received by Meridian were received “in consequence of” the interruption resulting from the insured peril and so satisfied the causal requirement in the savings clause. Reversing the decision of Jagot J on this point, the court answered this question by saying, at para 461: “As a matter of the application of the policy’s provisions, they were not. The criteria for eligibility for JobKeeper payments were financial ones; they did not depend on whether or not there had been an outbreak within 20 km of the premises of the business. Meridian was entitled to the JobKeeper payments regardless of whether or not there was an outbreak within 20 km of its premises. Conversely, had Meridian not met the financial tests for JobKeeper, it would not have been entitled to JobKeeper payments, even if the insured peril in [the disease clause] occurred.” The appellants submit that this reasoning should be applied here in preference to the conclusion reached by the courts below. In considering the appellants’ argument, it is important to keep in mind that the sole test is one of proximate causation: were the furlough payments received “in consequence of” the insured peril? The criteria for claiming payments under the CJRS are therefore relevant only in so far as they bear on the answer to that question. Two preliminary points may be made. First, the courts below addressed a different question of causation which does not need to be decided. They found that the existence of the CJRS was caused by Government actions which were themselves caused by cases of Covid-19 that included cases falling within the prevention of access clauses in the Arena and Gatwick policies. For example, Jacobs J said, at para 445 of his judgment, that: “the effect of the decision of the Supreme Court in the FCA test case is that, when considering the operation of the insured peril, a concurrent causation analysis is to be applied … It is therefore sufficient to show that the CJRS (and thus the payments made pursuant to that scheme) was brought into being in consequence of a combination of government restrictions affecting the business of each claimant policyholder in combination with restrictions affecting the business of other policyholders.” In the Court of Appeal, the Chancellor, at para 183, quoted the above passage with approval and also said, at para 185, that: “it was the general prevalence of Covid-19 (including cases within the relevant radius) which led to the restrictions imposed by the Government. The furlough scheme was announced at the same time that those restrictions were imposed and was intended to mitigate the effects of those restrictions, so that the incidence of Covid-19 and the restrictions imposed as a consequence were a sufficient effective cause of the furlough scheme.” We see no reason to doubt this analysis of what caused the Government to introduce the CJRS. But we think it adds unnecessary complication. There is no need to show that the furlough scheme itself was a consequence of the insured peril. All that matters is whether the payments made to the policyholders under the CJRS were made in consequence of the insured peril. The answer to that question does not depend on whether the CJRS was brought into being as a result of prevention of access to the policyholder’s premises caused by Government restrictions which were themselves caused by cases of Covid-19 occurring within the relevant radius of the premises. The analysis would be just the same if the furlough scheme had been a pre-existing scheme established before the relevant restrictions were imposed. 87. The other preliminary point concerns what is meant by the appellants’ assertion that the existence of the insured peril was “irrelevant” to the policyholders’ entitlement to receive furlough payments. As already noted, the circumstances which gave rise to a claim under the prevention of access cover were relevant to the entitlement to receive furlough payments in that they were “circumstances arising as a result of coronavirus or coronavirus disease” which therefore brought an employee instructed to cease work within the scope of the CJRS (see para 22 above). Those circumstances were also causally relevant to the payments in that the prevention of access to the policyholder’s premises in circumstances that fell within the cover induced the policyholder to furlough employees and claim payments under the CJRS. But when they use the term “irrelevant”, what counsel for the appellants evidently mean is just that it was unnecessary for the policyholder to prove that all the elements of the prevention of access cover were met in order to claim payments under the CJRS and that the policyholder would have been entitled to claim, and would have claimed, such payments even if one or more of those elements had not been present. The appellants’ argument comes down to the proposition that the policyholders would have received payments under the CJRS irrespective of whether use of or access to their premises had been prevented or hindered and irrespective of whether any case of Covid-19 had occurred within one mile of the premises. This is just another way of saying that a “but for” test is not satisfied. The appellants’ argument assumes that, to show that the furlough payments were proximately caused by the insured peril, it must be shown that the payments would not have been made but for the occurrence of the insured peril. They argue that proximate causation has not been established because this test is not met. The flaws in the irrelevance argument are patent. They are, first, that the appellants’ case is inconsistent with what was decided in the FCA test case and, second, that it is internally inconsistent. The appellants’ case is inconsistent with the decision in the FCA test case because, to avoid giving credit for furlough payments, they are seeking to rely on the very “but for” test of causation rejected by this court. Their case is also internally inconsistent because they need to rely on this court’s analysis of causation in the FCA test case, including the rejection of the “but for” test of causation for which the insurers contended, to establish that they have cover for their business interruption losses. These points may be illustrated by the appellants’ example of a policyholder who owns a hotel located in a rural area where (so it is assumed) no case of Covid-19 occurred within one mile of the premises. For that reason, the policyholder would not have cover under the prevention of access clause. Yet the nationwide restrictions imposed by the Government would have prevented or severely hindered the use of this hotel in the same way as every other hotel, causing loss of revenue and likely causing the policyholder to furlough employees and claim payments under the CJRS. There is no reason to suppose that the presence or absence of a case of Covid-19 within the radius would have made any difference to the amount of business interruption loss suffered. The appellants use this example to illustrate that they could, and would, have suffered business interruption losses and received furlough payments even in the absence of the insured peril. This shows, so they argue, that the furlough payments were not proximately caused by the insured peril. But if the argument were a good one, it would apply equally to the reduction in revenue claimed by them under the policies. That reduction in revenue could and would have been suffered even in the absence of the insured peril. On the appellants’ argument, it would follow that the loss of revenue was not proximately caused by the insured peril. That would mean that the appellants have no cover for their losses under the prevention of access clause in the policies. Fortunately for the appellants, this argument—which the insurers were then making—failed in the FCA test case, so they do have cover under the prevention of access clause. But they cannot have it both ways. They cannot, on the one hand, assert in reliance on the decision in the FCA test case that their losses were proximately caused by the insured peril even though they would have been suffered in the absence of the insured peril and, on the other hand, maintain that their savings from a consequent reduction in expenses were, for that reason, not proximately caused by the insured peril. They have themselves accepted—as they were logically bound to do—that the same test of causation applies both to the link between the insured peril and loss under the insuring clause and to the link between the insured peril and the reduction of expenses under the savings clause (see para 64 above). 93. Although the courts below dealt with it briefly, they identified this fundamental flaw in the appellants’ argument. It was, we think, the point that Jacobs J was making when he said, at para 446, that “what works on one side of the line should also work on the other” and that it is not appropriate to take a different approach to causation in the context of savings. In the Court of Appeal, the Chancellor made the point succinctly when he said, at para 187 of the judgment: “The insureds’ arguments in effect depend upon a but for test of causation: it is said that the insureds would have been entitled to CJRS payments in circumstances in which the insured perils had not occurred and so would be entitled to such payments, without the occurrence of the insured perils, and so the savings were not caused by the insured perils. But these are the very but for causation arguments which were rejected by the Supreme Court in the FCA test case, both in relation to the operation of the insured perils, and the trends clauses. They are equally inapposite to the related causation question which applies to the savings clauses, in which the proximate cause test should be the same.” The appellants attempt to overcome this fundamental objection to their argument by relying on an example given in the FCA test case to illustrate the limits of the same underlying fortuity principle. The example was that of a travel agency with prevention of access cover which lost almost all its business because of the travel restrictions imposed as a result of the pandemic. We said, at para 244 of the judgment, that: “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.” In their written case the appellants draw an analogy with a policyholder who owns a hotel in an inner city area with many cases of Covid-19 within the one mile radius, which was subject to restrictions during the pandemic, but which was in any event being renovated during the period and would not have opened throughout that period. The appellants submit—and we agree—that in this case the sole proximate cause of the loss of revenue (and savings in expenses) would be the renovations and not the insured peril. The appellants observe that the travel agency in that example would (if it was a qualifying employer with a registered PAYE scheme) have been entitled to furlough employees and claim payments under the CJRS. They submit that what applies to loss must also apply to savings, so that the furlough payments received by the travel agency would also not be proximately caused by the insured peril. The appellants then argue that their position is analogous to that of the travel agency and that, as in the travel agency example, the sole proximate cause of the furlough payments which they received was the Covid-19 pandemic and not the insured peril. It is difficult to understand how the appellants think that this argument assists their case. The point of the travel agency example was to illustrate a situation in which business interruption loss would not be covered by a prevention of access clause even though all the elements of the composite insured peril were present. If the appellants are right, therefore, that their situation is analogous to that of the travel agency, the conclusion is that they are not entitled to an indemnity. It would, as they submit, also follow that furlough payments would not fall within the savings clause because such payments would not be caused by the insured peril. But it is hard to see why the appellants consider this to be of comfort. If their business interruption loss is not covered because it was not proximately caused by the insured peril, the fact that furlough payments cannot be deducted from their loss does not make them any better off. 99. The reasoning underpinning the travel agency example was not spelt out in the FCA test case judgment as fully as it might have been. But the aim was to illustrate loss of business which, arguably at least, would not be covered because it was not caused by the same underlying fortuity as the insured peril. The reason would be that it was not naturally to be expected, or inherently likely, that an outbreak of disease which included cases of illness occurring within a mile of the insured’s premises and led to Government action which prevented or hindered use of or access to the premises would also result in restrictions on international travel. On this view, consequences of the travel restrictions should not be disregarded in deciding whether business interruption losses (or savings) were caused by the insured peril. This might lead the court to conclude that, even though the insured peril had arisen (because government restrictions imposed in response to cases of Covid-19, including cases within the radius, had prevented or hindered use of or access to the business premises), part if not all of the loss of revenue suffered by the business was not proximately caused by the insured peril. The Australian case of LCA Marrickville (see para 81 above) relied on by the appellants involved an actual claim by a travel agency raising a similar issue. As mentioned earlier, one of the policyholders in that case, Meridian, was a travel agency which had cover for business interruption loss caused by an outbreak of disease occurring within a 20 kilometre radius of its premises. It was an agreed fact that there was a relevant “outbreak” of Covid-19 within 20 kilometres of Meridian’s premises in Melbourne by 30 March 2020. At first instance Jagot J found that this outbreak was a proximate cause of restrictions imposed by the Government of Victoria which prevented potential customers from attending Meridian’s premises. The judge left open for further evidence the question whether this had caused business interruption loss, while observing that a reduction in business transacted by telephone or internet would not be caused by the outbreak (para 420). Before the outbreak occurred, the Commonwealth Government had already introduced an overseas travel ban which effectively prevented Meridian’s customers or potential customers from leaving Australia (para 407). International travel bookings accounted for around 90% of Meridian’s revenue and domestic travel bookings for the remaining 10% (para 404). 103. Applying the same analysis as this court had adopted in the FCA test case, the judge found that restrictions imposed by the Victorian Government and the Commonwealth Government’s overseas travel ban did not have the same underlying cause. That conclusion was upheld on appeal by the Full Federal Court. The court reasoned, at para 441: “Here, the insured peril was the outbreak of a disease within 20 kilometres of the insured’s [premises]. It may well be expected that when such a circumstance arises, the authorities will require businesses to close their doors and restrict the free movement of residents. That being so, each of those events, if not otherwise part of the insured peril itself, can be ignored as competing causes of the insured loss. However, the imposition of nationwide international travel restrictions is not something which the parties would naturally expect to occur concurrently with the localised outbreak of a disease. That is consistent with the primary judge’s findings that those restrictions were motivated by factors other than the outbreak of the disease in Victoria. They, therefore, do not have the necessary characteristics to be causes arising from the same underlying fortuity such that their causative impacts cannot be set up against the insured peril. As the Commonwealth Government travel bans effectively curtailed or destroyed Meridian’s business … by detrimentally impacting that 90% of its business related to international travel, it could not be said that the insured peril was a proximate cause of those losses. Accordingly, there is no basis on which to upset the primary judge’s reasons in this respect.” This was the context in which the court considered whether, on the assumption that Meridian was entitled to an indemnity under the disease clause for loss of domestic travel business, payments received under the JobKeeper program established by the Commonwealth Government were to be taken into account under the savings clause in calculating the amount payable by the insurer. We have quoted the court’s reasons for concluding that JobKeeper payments were not received “in consequence of” the insured peril at para 81 above. On the findings made by Jagot J and affirmed by the Full Federal Court, that conclusion is readily understandable. To be eligible to participate in the JobKeeper program, an entity’s turnover had to have reduced by a relevant percentage (30% in the case of a business the size of Meridian) (para 452(2)). The dominant cause by far of the reduction in turnover by more than 30% which entitled Meridian to claim JobKeeper payments was loss of international travel business which the court had found was not proximately caused by the local outbreak of disease. Accordingly, the receipt of the JobKeeper payments was not proximately caused by the insured peril. This reasoning and conclusion do not apply or assist the appellants here. It is common ground that the losses suffered by their businesses were proximately caused by the insured peril operating concurrently with other (uninsured but non-excluded) consequences of the same underlying fortuity. The same must equally apply to the associated savings from receiving furlough payments. Unlike Meridian or the hypothetical travel agency postulated in the FCA test case, the policyholders in the present proceedings are agreed on these appeals to have sustained business interruption loss proximately caused by the insured peril. Applying the same test of causation which applies to the link between the insured peril and loss under the insuring clause to the link between the insured peril and the reduction of expenses under the savings clause, the occurrence of the insured peril likewise caused them to furlough employees and receive payments under the CJRS. Subject to the collateral benefits argument which we are about to consider, the savings were therefore made “in consequence of” the insured peril. The appellants’ second argument on the causation issue relies on the proposition that a gift or other gratuitous, benevolent or voluntary conferral of a benefit by a third party on the victim of an injury is not regarded in law as caused by the injury; and that, for this reason, such a “collateral” benefit does not reduce the loss (it is “res inter alios acta”) for which the victim may claim compensation or an indemnity. As developed, the principal argument was that, in insurance law, as in the law on damages for torts and breach of contract, a benefit that is conferred on the insured by a third party gratuitously, benevolently or voluntarily is collateral and is to be left out of account; ie in that situation it is the third party’s intervention that is the proximate (or legal) cause of the gain and not the insured peril. 109. In Swynson Ltd v Lowick Rose LLP [2017] UKSC 32; [2018] AC 313, para 11, Lord Sumption explained the concept that such a benefit does not reduce loss in this way: “The general rule is that loss which has been avoided is not recoverable as damages … To this there is an exception for collateral payments (res inter alios acta), which the law treats as not making good the claimant’s loss. It is difficult to identify a single principle underlying every case. In spite of what the Latin tag might lead one to expect, the critical factor is not the source of the benefit in a third party but its character. Broadly speaking, collateral benefits are those whose receipt arose independently of the circumstances giving rise to the loss. Thus a gift received by the claimant, even if occasioned by his loss, is regarded as independent of the loss because its gratuitous character means that there is no causal relationship between them.” 110. Adam Kramer KC and William Day, who presented the appellants’ case on this issue with intellectual dexterity, rooted this idea in the theory of causation developed by Hart and Honoré in their seminal work Causation in the Law, 1st ed (1959) and 2nd ed (1985). Hart and Honoré maintained that the “common sense” concept of causation used in ordinary life can be seen, on analysis, to rest on certain statable principles and that the same concept of causation is also used in legal reasoning. One of the core principles identified by Hart and Honoré was that voluntary human action has a special place in causal inquiries and, more particularly, that “the free, deliberate and informed act or omission of a human being, intended to exploit the situation created by [the] defendant, negatives causal connection”: see 2nd ed (1985), pp 43, 136. They described one aspect of this principle as being that “voluntary conduct negatives causal connection not only with loss but with gain” (p 137), so that “where the immediate source of the gain is an indemnity, compensation, or gift from a third person the rule is that it cannot be taken into consideration if the third person acted voluntarily” (p 141). It is also worth noting their comment that “the principle is elastic and the voluntary nature of an act to some extent a matter of degree” (p 138). An initial question is whether it is even open to the appellants to make the collateral benefits argument if, as we have concluded, the irrelevance argument fails. In the courts below, the insurers successfully maintained that it is not. The Court of Appeal was persuaded that “once it has been determined, in relation to Ground 2 [the irrelevance argument], that the reduction in the charges and expenses of the business is in consequence of the insured peril, the argument that the payments which led to that reduction were collateral must fail”: see para 188. Jacobs J also appears to have taken this approach, at paras 454–455 of his judgment. We do not accept that the argument can be side-stepped in this way. The contention that payments made under the CJRS are not to be regarded in law as caused by the insured peril because of their gratuitous, benevolent or voluntary character is analytically distinct from the contention that such payments are not to be regarded in law as caused by the insured peril because they do not satisfy a “but for” test. The rejection of the latter argument does not rule out the former. They are concerned with different aspects of causation. The benevolence argument therefore needs to be considered separately. The insurers respond to this by saying that there is no scope for an argument based on the general law because the parties have agreed in the insurance policies how the amount of the business interruption loss recoverable under the policy is to be calculated—including what benefits are to be deducted as part of this calculation—and this agreement supplants or displaces the general law which would otherwise apply. We agree that, for this reason, there is no scope for applying the general law directly. The policy provisions specifying how the amount payable by the insurers is to be calculated are, for this purpose, a complete code. The relevant provision here is the savings clause. If payments received by policyholders under the CJRS come within the wording of the savings clause, they must be deducted from the amount otherwise payable. If they do not come within that wording, they must not be deducted. There is no other legal basis on which they can be taken into account. But it does not follow that the general law has no role to play. As noted earlier, it is common ground between the parties that the words “in consequence of” in the savings clause are to be interpreted as meaning “proximately caused by”. The term “proximate cause” is used in insurance law to indicate that whether a loss is to be regarded as caused by an insured peril is a legal as well as a factual question. The relevant legal principles include principles concerning when voluntary conduct, such as the gratuitous conferral of a benefit, negatives causal connection. It is logical and common ground that the same principles that apply to causation of loss also apply to causation of gains. It can thus be said that such principles, recognised as part of the general law, have been imported by the language of the savings clause and are to be applied in determining whether savings from the receipt of a benefit were made “in consequence of” the insured peril. If they were, the savings clause requires them to be deducted. Central to the appellants’ collateral benefits argument is a line of insurance cases referred to, loosely, as “subrogation” cases. The leading cases are Burnand v Rodocanachi Sons & Co (1882) 7 App Cas 333 and Castellain v Preston (1883) 11 QBD 380. The appellants submit that general principles of proximate (ie legal) causation are the correct explanation for the principle applied in these cases. 117. The context for this line of cases is the general rule that indemnity insurance functions only as a backstop form of protection, such that the insurer is entitled to the benefit of any sum received or receivable from any third party in diminution of the loss. Three types of case can be distinguished. In the first, the insured receives compensation from a third party before being indemnified by the insurer. In this situation the insurer’s liability is reduced by the amount received from the third party. In the second type of case, the insured receives compensation from a third party after being indemnified by the insurer. In this situation the insured is liable to account to the insurer for the sum received from the third party, which is subject to an equitable lien in the insurer’s favour: see Lord Napier and Ettrick v Hunter [1993] AC 713. The only difference between the first and second type of case is one of timing, which in principle should not affect the result. The question whether the insurer is entitled to the benefit of a payment received from the third party should not, and does not, depend on whether the payment is received by the insured before or after the insured has been paid by the insurer. In the third type of case, the insured is indemnified by the insurer but has an unsatisfied right to claim compensation from a third party for the insured loss. Here the insurer is entitled to take over the insured’s right of action against the third party and enforce it in the name of the insured. Although the term “subrogation” has commonly been used to refer to cases of the second as well as the third type, it would be clearer to confine the term to the third type of case, where the insurer is “subrogated” to (ie entitled to take over) the insured’s rights against the third party (a point well made in Charles Mitchell and Stephen Watterson, Subrogation Law and Practice (2007), ch 10). The cases in the line of authority cited by counsel for the appellants are all cases of the first or second type referred to in para 117 above and are therefore not cases of subrogation in the strict sense. They are in fact cases in which the insured person had no right of action against a third party to which the insurer could have been subrogated but where the third party nevertheless made a payment to the insured relating to the subject matter of the loss. The question in each case was whether the insurer was entitled to the benefit of this payment. In two early cases—Randal v Cockran (1748) 1 Ves Sen 98 and Blaauwpot v Da Costa (1758) 1 Eden 130—vessels seized by Spanish raiders were insured against loss. In each the shipowner was indemnified by the insurer for the loss of the vessel but later received a payment out of funds set up by the King (from reprisals against Spanish vessels) for losses caused by the Spanish raids. In both cases it was held that the insurer was entitled to the benefit of this sum. Those cases were distinguished in Burnand v Rodocanachi. The facts were that a cargo of tobacco insured for an agreed value against risks that included war risks was destroyed by the Confederate cruiser Alabama during the American Civil War. The underwriters paid the agreed value of the cargo, which was less than its actual value, to the cargo owner. The cruiser had been built for the Confederacy in Liverpool, and after the war the United States demanded compensation from the British Government for the damage to shipping caused by the Alabama. An international arbitration resulted in an award of US$15.5 million, which was distributed to shipowners and others who had suffered losses under an Act passed by the US Congress. From this fund the cargo owner received a payment assessed to represent the difference between the insured value and the actual value of the cargo. The issue which reached the House of Lords on appeal was whether the underwriters were entitled to this sum. The House of Lords held that they were not. This was because the Act of Congress specifically provided that no payment out of the fund was to be allowed “for any loss or damage for or in respect to which the party injured … shall have received compensation or indemnity from any … insurer”, but that if any compensation or indemnity received from any insurer “shall not have been equal to the loss or damage so actually suffered, allowance may be made for the difference”: see (1881) 6 QBD 633, 634. Although the sum paid by the underwriters to the cargo owner had been the agreed value of the cargo, that agreement could not be treated as conclusive given the express basis on which the payment from the Alabama fund had been made. “Here it is admitted that there is in the Act of Congress everything said and done which a supreme legislature could possibly say or do for the purpose of excluding the present claim and attributing that fund which has been appropriated in this case to the sufferers by the capture, not to the valued part but to the unvalued part of the loss. That distinction, which in my opinion does exclude for this purpose the part covered by the valuation of the policy of insurance, is made by the Act of Congress. It was a true and bona fide valuation but it did not cover the actual loss. The fund awarded by the Act of Congress of the United States is only for that part of the actual loss which the valuation did not cover and which the insurers have not paid.” It was accepted that the payment by the United States was made voluntarily and not pursuant to any legal obligation. But Lord Blackburn in particular went out of his way to emphasise that this was not a reason to reject the underwriters’ claim. As he said, at p 341: “the question is not whether the money was voluntarily paid or not voluntarily paid, but whether de facto the money which was paid did reduce the loss.” The reason the money did not reduce the loss was that “In the present case the Government of the United States did not pay it with the intention of reducing the loss”. 125. All the Law Lords distinguished Randal v Cockran and Blaauwpot v Da Costa. Lord Blackburn said of those cases, at pp 339–340: “It was, certainly, I think, a voluntary gift on the part of the Crown, and was for the benefit of the sufferers. But then I think that that gift being made, as it was made, for the benefit of those who had suffered from the captures, and the money being paid for that purpose, it did diminish the loss; and consequently the benefit of it enured to the persons who were bound to indemnify; and it was so decided in those two cases. It was not because the King was bound to pay the money—he was not: it was not because there was a moral obligation to pay it … it was because de facto there was a payment which prevented, or diminished pro tanto, the loss against which the insurers were bound to indemnify the assured.” Hence the distinction between Burnand v Rodocanachi, on the one hand, and Randal v Cockran and Blaauwpot v Da Costa on the other, turned on whether the third party provider of the fund had, or had not, made clear whether it intended the payment to enure for the benefit of the insurers. In Randal v Cockran and Blaauwpot v Da Costa, there had been no indication that the money paid was to benefit only the insured and was not to benefit the insurer. The money paid by the King was therefore deducted from what the insurer was bound to pay under the insurance policy. In contrast, in Burnand v Rodocanachi it was made explicit, as an aspect of the legislative scheme, that the money was not given to benefit the insurer. The money paid therefore did not go to reduce what the insurer was bound to pay under the insurance policy so that, having made payment, the insurer was not entitled to any reimbursement. In the second leading case of Castellain v Preston a house was damaged by fire after its owner had exchanged contracts to sell it for £3,100 but before the sale was completed. The seller had insured the property against fire and recovered £330 in respect of the damage from the insurers. The purchaser afterwards paid the purchase price of £3,100 in full without any reduction on account of the fire damage. The insurers then brought an action against the seller to recover the sum of £330. The claim failed at first instance but succeeded in the Court of Appeal. 128. In this case the money paid by the third party (the purchaser of the property) was not paid voluntarily. The purchaser had a legal obligation to pay the purchase price without deduction. But that obligation arose under a contract (the contract of sale) which had nothing to do with the fire and the payment under that contract of sale was not made for the purpose of compensating the seller for the fire damage or for any reason connected with the fire. At first instance Chitty J held that in these circumstances the insurers were not entitled to receive any part of the purchase price. But the Court of Appeal disagreed. Bowen LJ (whose judgment best articulates the reasons for the decision) said, at p 404: “It is insisted that only those payments are to be taken into consideration which have been made in respect of the loss. I ask why, and where is the authority? If the payment diminishes the loss, to my mind it falls within the application of the law of indemnity.” Later in his judgment, at p 405, Bowen LJ emphasised that it did not matter that “the contract of insurance was a collateral contract wholly distinct from, and unaffected by, the contract of sale”. All that mattered was that “The beneficial interest of the vendors in the house depends on the contract [of sale] being fulfilled or not, and the fulfilment of the contract lessens the loss, its non-fulfilment affects it”. It was argued on behalf of the insured—and, according to Cotton LJ at p 395, “was put to us most strongly”—that “if any part of the purchase-money is to be taken into account, why is a gift not to be taken into account? That may be said to diminish the loss as well as a contract of sale”. To this, Cotton LJ responded: “The answer is that when a gift is made afterwards in order to diminish the loss, it is bestowed in such terms as to shew an intention to benefit the assured, and to give the insurer the benefit of that would be to divert the gift from its intended object to a different person.” 130. Bowen LJ dealt with the suggested analogy with gifts in a similar way. He said, at p 404: “With regard to gifts, all that is to be considered is, has there been a loss, and what is the loss, and has that loss been in substance reduced by anything that has happened?” He accepted that “in the vast majority of cases, it is difficult to conceive a voluntary gift which does reduce the loss”. But he explained that the reason is that such a payment is not generally intended to do so. Bowen LJ gave an example, at pp 404–405, which has been cited in later cases: “Suppose that a man who has insured his house has it damaged by fire, and suppose that his brother offers to give him a sum of money to assist him. The effect on the position of the underwriters will depend on the real character of the transaction. Did the brother mean to give the money for the benefit of the insurers as well as for the benefit of the assured? If he did, the insurers, it seems to me, are entitled to the benefit, but if he did not, but only gave it for the benefit of the assured, and not for the benefit of the underwriters, then the gift was not given to reduce the loss, and it falls within Burnand v Rodocanachi. If it was given to reduce the loss, and for the benefit of the insurers as well as the assured, the case would fall on the other side of the line, and be within Randal v Cockran, to which allusion has been made.” Castellain v Preston exemplifies the general position that an insurer is able to recover the monies it has paid to the extent that the insured’s loss has been diminished by the receipt of money from a third party (here the purchaser). But the reasoning indicates that, had the third party paid the money gratuitously or benevolently, with the intention of benefiting only the insured and not the insurer, the position would have been different. In that situation the third party’s payment would not be deducted from the loss to be indemnified under the insurance policy and the insurer would have no right to recover the money paid. The distinction drawn in Burnand and in Castellain between payments which reduce the insured loss and payments which do not reduce the loss because they were intended to benefit only the insured and not the insurer has consistently been applied in later cases. 133. In Stearns v Village Main Reef Gold Mining Co Ltd (1905) 10 Com Cas 89 the Transvaal Government in South Africa had commandeered gold belonging to a mine owner just before the outbreak of the Second Boer War. The gold was insured against loss, and the mine owner recovered the full value of the commandeered gold from the underwriters. In response to a request from the mine owner and to avoid the shutting down of the mine, the Government later paid back a sum representing approximately a third of the value of the gold seized. The underwriters successfully claimed this sum from the mine owner. In affirming the decision, the Court of Appeal held that the fact that the payment by the Government was in the nature of a gift or an act of grace did not prevent the underwriters from claiming the benefit of it. That was because it was clear that the payment was made with the object of diminishing the loss which was covered by the insurance and there was no condition attached to the payment showing that it was being given for the purpose of benefiting the mine owner personally. Romer LJ emphasised that the test was an objective one. He said, at pp 95–96: “The question whether the Transvaal Government, in returning this money, were thinking of the insurers appears to me to be immaterial, if they imposed no condition or trust or obligation upon the money as between themselves and the defendants when it was returned. Probably the Transvaal Government were not thinking of the insurers at all. But on the facts I have stated, it appears to me that … in the absence of any circumstances negativing that view, the insurers would be able to say: ‘We are entitled to avail ourselves of that diminution of the loss which we insured against.’” In Merrett v Capitol Indemnity Corpn [1991] 1 Lloyd’s Rep 169 the opposite inference was drawn. An insurance broker gratuitously paid to its client (the insured) part of a sum claimed from insurers. In an arbitration the insurers were held liable to indemnify the insured for its loss but the arbitrators deducted from the award the amount paid by the broker, finding that this had not been a loan but an out and out payment made for the commercial purpose of keeping the client’s goodwill. On an appeal from the award, Steyn J held, at p 171, that the arbitrators had adopted a wrong legal approach by simply asking whether the payment was an out and out payment and not addressing the question whether the payment was intended solely for the benefit of the insured. On the primary findings of fact made by the arbitrators, it was clear that the broker expected to be reimbursed by the insurer and that the payment was made solely for the benefit of the insured and not for the benefit of the insurer. 135. In Colonia Versicherung AG v Amoco Oil Co [1997] 1 Lloyd’s Rep 261 ICI Chemicals & Polymers Ltd (“ICI”) bought a cargo of naphtha which had been contaminated before shipment. ICI claimed damages from the owner of the refinery from which the cargo was shipped (Amoco), alleging that the contamination was due to its negligence. Amoco made a payment in settlement of the claim and, as part of the settlement, took an assignment of ICI’s cargo insurance policy. One of the questions in proceedings between Amoco and the insurers was whether the insurers were entitled to deduct the payment made by Amoco to ICI from the sums otherwise due under the policy. The Court of Appeal held that, on the authorities, the key question was the intention of the parties to be derived from the relevant circumstances, in this case the settlement agreement, and that the insurer was entitled to the benefit of moneys paid to the insured unless the payment was intended to benefit the insured to the exclusion of the insurers. On the proper construction of the settlement agreement, Amoco’s intention had not been to benefit ICI to the exclusion of the insurers. The payment therefore went to reduce the insured loss. Hirst LJ (with whom Peter Gibson and Pill LJJ agreed) said at p 270: “In Burnand v Rodocanachi, as the judgments show, the critical factor was the clearly expressed intention of the US Congress to compensate the beneficiaries for their uninsured losses, together with the express exclusion of any claim by the insurers in their own right or that of the assured. … In Castellain v Preston, on the other hand, no such intention to exclude the insurers could be derived from the purchaser’s payment of the full purchase price without abatement on account of the fire damage. … It follows that the crucial question is whether, on the construction of the deeds, … it was the intention of Amoco to benefit ICI to the exclusion of the [insurers].” A similar conclusion was reached in Talbot Underwriting Ltd v Nausch, Hogan & Murray Inc (The “Jascon 5”) [2006] EWCA Civ 889; [2006] 2 Lloyd’s Rep 195, where a vessel sustained damage from flooding while being fitted out at a shipyard. The shipyard repaired the damage at its own expense. One of many issues decided by the Court of Appeal turned on whether the shipowner was entitled to an indemnity under its insurance for the damage. The argument failed on the ground that at the time of the flooding the vessel was at the risk of the yard which had to repair the damage to complete its contract with the shipowner and obtain payment. There was nothing to suggest that the yard carried out the repairs to the vessel with any intention other than to complete the work under the contract and obtain payment of the price: see para 66. Thus, no inference could be drawn that the work was intended to benefit the shipowner to the exclusion of the insurers. The last case cited in this line of authority is Atlas Navios-Navegação LDA v Navigators Insurance Co Ltd (The “B Atlantic”) [2014] EWHC 4133 (Comm); [2015] 1 Lloyd’s Rep 117. This concerned a disputed claim under a marine insurance policy for the loss of a vessel, which included a claim for suing and labouring expenses. The case raised many issues. One point taken by the insurers was that some US$1.2 million of the legal fees for which the shipowners were claiming an indemnity had been funded by a third party (the shipowners’ P&I Club, Gard) on an ex gratia basis. The insurers contended that, in the circumstances, the owners were not entitled to recover that sum. Flaux J dealt with the point shortly observing, at para 348, that it was unpleaded but was a bad point anyway because the facts were similar to those of Merrett v Capitol Indemnity. As in that case (where the payment was intended solely for the benefit of the insured), the payment did not diminish the insured loss. The principles illustrated by these cases were, in our view, accurately summarised by Butcher J in Stonegate, at para 284. In short, any payment made by a third party to the insured in respect of the subject matter of the insured loss, even if made voluntarily or gratuitously, will diminish the loss and enure to the benefit of the insurer except where the intention of the third party in making the payment, expressly stated or inferred from the circumstances, was to benefit only the insured to the exclusion of the insurer. If that test is applied here, furlough payments made under the CJRS to policyholders, even if they could be characterised as voluntary, plainly fall within the rule and not the exception. The furlough payments reduced costs of employing staff which the policyholder would otherwise have borne as a result of continuing to employ those staff during the indemnity period (the general assumption on which business interruption loss is calculated under the policies). Prima facie, therefore, the furlough payments reduced the losses insured against. The exception does not apply as there is nothing in the terms of the CJRS or the surrounding circumstances to suggest that payments made under the scheme were intended to benefit only the employer who made the claim for reimbursement to the exclusion of any insurer who was liable to indemnify the employer for its business interruption loss. In contrast to Burnand, payments made under the CJRS were not expressed to be limited to uninsured losses. Nor was there any term of the scheme or statement by the Government suggesting that the intention was to benefit employers to the exclusion of insurers. This was so notwithstanding—as Butcher J noted in Stonegate, at para 286—that the Government was aware that some companies had business interruption insurance, as evidenced by a Treasury Fact Sheet of 18 March 2020. Such an intention is in any case an improbable one to impute to the Government. Had it been thought appropriate to make special provision for losses covered by insurance, the logical provision to make would have been to exclude such losses from the CJRS. Alternatively, if that option was rejected because of the potential delay and uncertainty involved for the employer in claiming under its insurance, the scheme could have provided for furlough payments to be refunded by the employer or the insurer to the extent that they were recovered or recoverable from the insurer. It is improbable that the Government would have intended the employer to enjoy the benefit of being reimbursed twice for the same expense. 142. The appellants assert that that was the Government’s intention. This assertion is based on a letter dated 25 September 2020 from Mr John Glen MP, Economic Secretary to the Treasury, to the Association of British Insurers. The letter contained a statement that: “It is the Government’s firm expectation that grant funds intended to provide emergency support to businesses at this time of crisis are not to be deducted from business interruption insurance claims.” The grant funds to which this statement was referring were specified as being the Coronavirus Small Business Grant Fund; the Retail, Hospitality and Leisure Grant Fund; the Local Authority Discretionary Grant Fund; and their equivalents in the devolved nations. Mr Glen’s letter was sent in response to a letter from the Association of British Insurers which conveyed a commitment on behalf of many insurers that they would not be making deductions for grants made under these schemes when adjusting business interruption insurance claims. It is doubtful that the letter from Mr Glen on behalf of the Government amounts to more than a statement of subjective expectation or belief to which no weight can be attached in determining the objective intention behind the schemes in question. But even if the letter is considered relevant to that determination, it does not assist the appellants’ case because it did not relate to the CJRS. The grants referred to in the letter were one-off payments which, as the letter indicates, were intended to provide general support to businesses considered to be in particular need of financial aid at a time of crisis. Unlike payments under the CJRS, such grants did not reimburse specific costs of the business. In these circumstances it would have been difficult for insurers to argue that they reduced the loss defined in the policy. What is most notable about the letter, in our view, is that the Government sought no similar commitment from insurers and expressed no similar expectation that payments made under the CJRS were not to be deducted from business interruption insurance claims. In summary, therefore, the exception recognised in Burnand does not apply to the facts of this case because there was no express stipulation or other clear indication that the Government intended the furlough payments to enure solely for the benefit of the insured and not the insurers. The appellants dispute the analysis of the “subrogation” line of cases which we have outlined above. They submit that it is wrong to derive from those cases a principle that a payment by a third party to the insured will be taken into account unless it can be established that the third party, in making the payment, intended to benefit only the insured to the exclusion of the insurers. As mentioned in para 116 above, they argue that the correct explanation of these cases lies in general principles of proximate (ie legal) causation. They maintain that what determines whether a benefit conferred by a third party diminishes the loss is whether the benefit is in law to be regarded as caused by the event insured against; and a payment made by a third party voluntarily or gratuitously or benevolently is collateral (ie the gain is not legally caused by the insured peril) and is to be left out of account. It was made clear in Burnand and Castellain and the line of cases following them that the fact that a benefit is conferred voluntarily or gratuitously does not determine whether it is to be treated as diminishing the insured loss. The explanation for why gifts often do not reduce the insured loss lies in the fact that they are commonly intended—either expressly or by implication from the circumstances—to benefit only the recipient to the exclusion of any insurer. That test is not one of causation, legal or otherwise. This is illustrated by Castellain where, as noted at para 128 above, there was no causal connection between the fire damage which was the peril insured against and the payment of the purchase price under the contract of sale. The benefit conferred by the third party purchaser was therefore entirely collateral. But that did not prevent the payment from being treated as diminishing the loss insured against. 148. However, although the subrogation cases do not assist them, the appellants also make a more general submission that, as a matter of principled interpretation of the words “in consequence of”, the payments under the CJRS ought not to be regarded as proximately caused by the insured peril because they were voluntary, gratuitous or benevolent and were therefore collateral benefits. The example of the brother’s gift posed in Castellain was used by Bowen LJ to illustrate the test of intention which determines whether a benefit conferred by a third party reduces the loss under the general law. But, so it is argued, the example can also be used to illustrate circumstances in which a benefit conferred by a third party might be said not to be proximately caused by the insured peril even though occasioned by it. Suppose that in that example the insurance policy had provided that the amount payable by the insurer was to be calculated as the cost of repairing the damage insured against less any sum saved in consequence of the damage. The policyholder could argue that the brother’s gift, though made in response to the damage, should not be regarded in law as a consequence of it because of the voluntary or gratuitous or benevolent nature of the act. Bringing the example closer to the present cases, counsel for the appellants posited a hypothetical case of a racing enthusiast who made a philanthropic donation to the Arena appellants for the specific purpose of helping them maintain the workforce at their racecourses during the pandemic lockdowns. In such a case the donation, although it reduced the wage bill, arguably did not reduce it “in consequence of the incident”. The argument would be that the proximate cause of the reduction was an independent act of generosity by the donor and not the insured peril. On this basis the payment would not fall within the savings clause. These examples do not assist the appellants, however, because the facts of the present cases are not analogous. The rock on which the collateral benefits argument founders is that it is untenable to characterise payments made under the CJRS as voluntary donations. As the appellants recognise in their written case, a payment is not voluntary or gratuitous in the sense required to negative causal connection if it is made pursuant to a legal obligation to make good the loss. Payments made under the CJRS were made pursuant to such a legal obligation. There was no discretion or choice about whether to make a payment or as to the amount payable under the scheme. An employer whose claim met the qualifying conditions was entitled to be reimbursed by HMRC for expenditure covered by the scheme. That CJRS payments were a matter of legal entitlement and obligation, and not largesse, is clear from the terms of the Treasury Directions. It is explicit, for example, in para 2.3 of the Treasury Direction of 15 April 2020, which provided that a claim under the CJRS must be made in such form and manner and contain such information as HMRC required “to establish entitlement to payment under CJRS”. Similarly, para 14.2 provided that: “Entitlement to a payment under CJRS is without prejudice to any entitlement to a payment under any similar scheme arising from a direction under section 76 of the Coronavirus Act 2020.” The subsequent Treasury Directions all contained similar provisions. 152. For the appellants, Mr Kramer KC sought to overcome this fundamental objection by arguing that the whole CJRS was a voluntary, gratuitous or benevolent intervention by the Government. He also emphasised that no consideration was given by employers for the right to receive payments under the scheme. We do not consider that these points affect the analysis. As discussed earlier, it is not necessary to show that the CJRS itself was a consequence of the insured peril, only that the payments made to policyholders under the CJRS were made in consequence of the insured peril. The fact that the payments were made as a matter of legal obligation and not voluntarily is therefore decisive. It is not relevant that the obligation was itself undertaken voluntarily. In this context it would be fallacious to suggest that a payment made as a matter of contractual obligation is a voluntary payment causally unconnected to the circumstances which triggered the obligation because the obligor chose voluntarily to enter into the contract. The position is no different where the source of the legal obligation is, as here, an instrument made under statutory authority. 153. As for absence of consideration, that would be relevant if it signified an absence of legal obligation (as where no contractual obligation is created because no consideration was given for a promise). But here it does not. As the CJRS was a statutory and not a contractual scheme, consideration in the technical sense which that term has acquired in the law of contract is not a relevant concept. If the question is asked, more broadly, whether payments under the CJRS were made in return for anything, they plainly were. Most directly, the employer in return for receiving CJRS payments in respect of an employee had to continue to employ that person and continue to pay their wages and other costs of employment despite having instructed the employee to cease all work in relation to their employment. After the initial period which ended on 31 July 2020, the employer also had to pay sums which were not fully reimbursed, in the form of employer national insurance contributions, any pension contributions and, during certain periods, a proportion of the employee’s wages. As well as these detriments to the employer, the furlough payments were made in order to obtain economic and public health benefits for the Government and the public at large. The aims of the CJRS included minimising the detrimental economic impact of the restrictions imposed to reduce the spread of Covid-19 and “maintain[ing] the UK economy’s productive capacity through the crisis” by preserving “employer-employee matches”. The CJRS also aimed to incentivise businesses to “follow government guidelines and, where necessary, ask employees to stop working in order to help reduce the spread of infection”: see HMRC, Coronavirus Job Retention Scheme: Evaluation Plan (December 2020), pp 4 and 8. The CJRS was not an exercise of gratuitous or benevolent giving. Accordingly, we do not accept that the CJRS itself, let alone payments made under it, can properly be characterised as gratuitous or voluntary or benevolent in nature. On no view can CJRS payments be likened to gifts or philanthropic donations for which it may be said that the sole cause in law is the donor’s voluntary decision to confer the benefit. The payments flowed as a matter of legal obligation from the circumstances arising from the Covid-19 pandemic which entitled the employer to claim furlough payments. In support of their contention that payments made under the CJRS were voluntary or gratuitous or benevolent and should therefore be characterised as collateral benefits, the appellants rely on various cases in the field of damages for torts and breach of contract. These cases do not assist the appellants and it is unnecessary to examine them. But four points may usefully be made. First, caution is needed in extrapolating from the law on damages for torts and breach of contract to the analysis of proximate causation in insurance cases because there are potentially significant contextual differences. One obvious difference is that an insurer who has agreed to indemnify the claimant against loss caused by a peril insured against has not committed any wrong. It cannot therefore be said that the effect of treating a benefit conferred by a third party as reducing the loss is to relieve a wrongdoer from a liability to pay damages. Another difference is that in the law of tort and contract, damages are calculated as if the claimant had taken any steps which the claimant would reasonably be expected to take to mitigate its loss and costs incurred in taking such steps can in principle be recovered as damages. No such mitigation rule applies in the context of non-marine insurance unless the policy provides for it. 158. Second, it is a matter of controversy whether or to what extent the question, as to when benefits conferred by third parties are to be taken into account in assessing damages for torts and breach of contract, rests on principles of legal causation rather than on principles and policies that can be rationalised on other grounds. Even if the principle is the same as under the general law of insurance, so that whether such benefits are to be taken into account depends on whether a payment which in fact reduces the loss was intended to benefit the claimant to the exclusion of the defendant, that principle is distinct from legal causation. That was the view adopted by the High Court of Australia in National Insurance Co of New Zealand Ltd v Espagne (1961) 105 CLR 569, where Windeyer J said, at p 598: “The most satisfying of the reasons that have been given for refusing to diminish damages because of voluntary gifts is that they are given for the benefit of the sufferer and not for the benefit of the wrongdoer.” He further explained, at p 600, in a passage cited in Hodgson v Trapp [1989] AC 807, 821, that: “the decisive consideration is, not whether the benefit was received in consequence of, or as a result of the injury, but what was its character: and that is determined … [in the case of gratuitous payments] by the intent of the person conferring the benefit. The test is by purpose rather than by cause.” The third point that we would make is that, however the matter is analysed, there is no relevant analogy between payments under the CJRS and voluntary gifts or gratuitous payments from a benevolent fund intended to alleviate the plight of victims of misfortune. As already discussed, the CJRS was not an exercise of charity or benevolence and payments from the CJRS were not voluntary but made pursuant to a legal obligation. Fourth, if there is any relevant analogy, it is with the treatment of state benefits payable as of right to injured claimants. This has for many years been governed by a special statutory regime (in its current form enacted by the Social Security (Recovery of Benefits) Act 1997). But as made clear by the House of Lords in Hodgson v Trapp [1989] AC 807, the position at common law is that such benefits are to be deducted in assessing damages. For the reasons given, Jacobs J and the Court of Appeal were correct to hold that the savings clauses in the Arena and Gatwick policies require furlough payments received by the appellants under the CJRS to be deducted from their lost revenue in calculating the amount payable as indemnity. That is because, on the proper construction of the savings clauses, the furlough payments (1) reduced charges or expenses of the policyholder’s business and (2) did so in consequence of the insured peril. We would therefore dismiss the appeals. http://dlvr.it/TS9CpW