IN the law of obligations, to establish that Event (E) was a factual cause of Consequence (C), it must generally be shown that E was a necessary condition for the occurrence of C. It is asked: “but-for E would C have happened?” This is the classic “but-for” test of factual causation. It is a test of the necessity of E, but not its sufficiency, for C to happen.
In London International Exhibition Centre Plc v Allianz Insurance Plc [2024] EWCA Civ 1026 (“Allianz”) – dismissing appeals to a decision on preliminary issues in six expedited test cases – the Court of Appeal affirmed, after the landmark decision in Financial Conduct Authority v Arch Insurance (UK) Ltd. [2021] UKSC 1, [2021] A.C. 649 (“Arch”) that the but-for test is not the only way to establish that E was a factual cause of C, in contract law, in the insurance context. E may cause C even if C would have happened without E.
Allianz and Arch both concerned insurers resisting claims by policyholders for coverage for business interruption losses suffered in consequence of government measures during the COVID-19 pandemic. The principal issue in both cases was what the causal test between insured peril and loss was and, so, whether the policies covered the losses. Allianz shows that Arch was no anomaly.
It is useful to first briefly set out Arch. The Financial Conduct Authority brought proceedings against eight leading business interruption insurers. The clauses covered losses caused by the occurrence of a notifiable disease within a specified radius of the policyholder premises (“radius clauses”) (at [48], [161]). The claimants therefore needed to establish that the insured peril – COVID-19 cases within a specified radius of the premises – was a factual cause of their losses. The Supreme Court held that the but-for test could not have been intended because it would create “intractable counterfactual questions” (at [195]). It would also mean that cases outside the radius could negate the factual causal link between cases for which the claimants were insured within the radius. Thus, instead, if a COVID-19 case was a “separate and equally effective cause” of the loss then factual causation would be established (at [192]–[197]). A single case within radius, therefore, sufficed (at [212]).
Allianz was decided in this context. However, the clauses in Allianz covered losses caused by closures resulting from cases of notifiable disease at the premises (“premises clauses”). The Court of Appeal (Males L.J., Popplewell L.J. and Andrews L.J., one judgment) held that this was immaterial. As in Arch, the but-for test did not apply.
The Court of Appeal began by setting out the relevant principles, for which Arch is now high authority. First, as a contract, the meaning of an insurance policy is interpreted objectively by asking how a reasonable person would have understood the policy. Because the policies were principally sold to small or medium sized enterprises, that reasonable person is not a “pedantic lawyer”, but an “ordinary policyholder” (at [51]–[52]). Second, while it is presumed in the insurance context that the parties intended a causal test of “proximate” or “efficient” cause, this may be displaced if the policy, properly interpreted, calls for some other test (at [52] and see Leyland Shipping Ltd. v Norwich Union Fire Insurance Society Ltd. [1918] A.C. 350, 369). Ascertaining party intention, here, requires a “wider enquiry” and is unlikely to depend on “linguistic analysis” (at [52]–[53]).
Third, importantly, a loss may have multiple factual causes which, individually, are neither necessary nor sufficient to cause the loss. If one such cause is an insured peril, and the other combining causes are not expressly excluded, then a policyholder will generally recover (at [55]).
Fourth, it follows that factual causation may be satisfied when the but-for test is not. In fact, where multiple unnecessary and insufficient causes combine to bring about some consequence, the but-for test necessarily cannot be satisfied in respect of any one of them individually (at [57]–[58], Arch, at [183], [190]–[191]).
Applying these, the lodestar of the court’s interpretation was the nature of the insured peril, and what the parties must have intended in light of it. Notifiable diseases – such as cholera, the plague and now COVID-19 – often spread rapidly and widely and, consequently, were unlikely to be confined to single premises: “such occurrences come not in single spies but in battalions” (at [62]). This being the kind of peril in contemplation, it was unlikely that the parties would have expected that any government measures would have been in singular response to disease at only the insured’s premises. The measures would likely have been in response to a wider outbreak. It followed that the parties must have intended, contrary to the submissions of some insurers (at [42]–[45]), that an insured peril could factually cause loss covered by the policy, without being a but-for cause of that loss. If a but-for test did apply, this would entail the policyholders having to prove that the specific case(s) at their premises were necessary for the Government to take its measures. This would be “difficult or impossible” and cannot have been intended (at [64]).
In these circumstances, the parties must have intended that factual causation would be satisfied if a case at the premises was “one of a number of [separate and equally effective] causes” of the measure (at [64], [73]) – rejecting causal necessity and, to be sure, sufficiency (see J. Stapleton, “Unnecessary and Insufficient Factual Causes” (2023) 16(2) Journal of Tort Law 351, 352).
One insurer also argued that the relevant authority had to know of the specific occurrence of the disease at the particular premises if that disease was to be said to have “caused” the government measures, and losses (at [46]–[47]). The idea here is that a decision cannot be “caused” by something of which the decision maker has no specific knowledge. The Court rejected this as “unrealistic”. Measures would be in response to both cases “known and unknown”, and the parties would have contemplated this (at [66]–[71]).
Note that it seems right that something which is constitutive of a homogenous and general phenomenon, but not itself precisely known, can be a factual cause of a decision in relation to that general phenomenon. When someone walks to Liverpool Street to catch the tube, they usually have no precise tube in mind. But the general phenomenon of tubes leaving Liverpool Street, of which the precise tube they do end up catching is constitutive, has caused them to decide to walk to Liverpool Street. As well, during COVID-19, measures were taken because of the prospect of future cases. So, even something which does not yet exist it seems may cause some consequence.
The “one of a number of [separate and equally effective] causes” test was thus satisfied – and causation established (at [72]) – because the occurrence of COVID-19 cases at the premises was “one of a number of causes” of the government measures closing businesses.
There are three more points about the case.
First, what is the relationship between Arch and Allianz? The Court was clear in stating that it was interpreting the specific premises clauses before it, bespoke as they were, and not looking for some reason to follow, or not follow, Arch (cf. Jacobs J., below, at [171]). However, simultaneously, the court accepted that Arch’s reasoning should not be “ignored” (at [59]) and it considered (at [25]–[30]) and used (at [63]–[64]) much of Arch’s reasoning on its way to finding that the same causal test which applied to radius clauses applied to premises clauses, too. It seems difficult to deny Arch’s influence.
Second, it is interesting that Allianz affirmed the holding in Arch that such interpretation is “unlikely to depend on linguistic analysis” placing, instead, an emphasis on the intentions of the parties (Allianz, at [53]). Arch had held of phrases such as “following”, and “in consequence of”, that “it is not profitable to search for shades of semantic difference” (Arch, at [162]). This seems right. Such phrases reveal no preference as to causal necessity or sufficiency. Objective intention rightly assumes primacy here. However, it seems a bridge too far to make an almost oppositional distinction between the “meaning of words” and the “legal effect of the contract”, as Arch did, and Allianz seems to affirm (Arch, at [162]; Allianz, at [53]). In other insurance contexts, post-Arch, the linguistic analysis of causal words has been important (see Brian Leighton Ltd. v Alliance Insurance Plc [2023] EWCA Civ 8, [2023] Lloyd’s Rep. IR 380).
Third, a perhaps less obvious but significant dimension to the case is that it further advances the concept of factual causation beyond the but-for test in the law of obligations generally. Even within the contractual interpretation of the policies, the court was mindful to justify diverging from the “conventional ‘but for’ approach to causation” (at [64]). There are many examples where the but-for test leads to counterintuitive conclusions (see J. Stapleton, “Unnecessary Causes” (2013) 129 L.Q.R. 39). In consequence, there are increasingly instances where the law of obligations does not apply it. Consider duress (Barton v Armstrong [1976] A.C. 104 P.C.), and material contribution in tort (Williams v Bermuda Hospitals Board [2016] UKPC 4, [2016] A.C. 888). Allianz, and Arch, join these.
Allianz is a significant sequel to Arch. It sheds light on the interpretation of causal language in insurance policies, in the important context of the COVID-19 pandemic. It should also be recognised as advancing the concept of unnecessary – beyond “but-for” – and insufficient causes, in contract and in the law of obligations. Permission to appeal has been refused.
