The London market sells two products under the name political violence, and they must be told apart at once. The terrorism and sabotage policy, of the LMA3030 family, covers terrorism and sabotage, often riot, civil commotion and strikes accompanied by damage, and it excludes war. The full political violence policy also covers insurrection, rebellion, civil war and war itself, and typically excludes only war between the five permanent members of the Security Council. This module deals with policies of the first kind, and with those that, without being full cover, exclude war. A paradox follows that the buyer rarely discovers at inception and always at the loss: cover is widest on the smallest disorders and narrowest on the event that actually destroys a plant. That shape is no drafting failure, it reflects a constraint of pooling, an interstate war producing simultaneous losses on every asset in a zone, which a portfolio carries only under aggregate limits, at the price of full cover. But acknowledging the reason does not remove the need to measure the effect.
Everything then turns on one word, attribution, and it is worth seeing why that word has become the pivot of the subject. The same explosion on the same transformer is an act of terrorism, therefore covered, if a non state actor carried it out, and an act of war, therefore excluded, if a state carried it out or directed it. Nothing in the damage separates the two hypotheses. The classification depends not on what happened to the insured property but on the identity of whoever acted, information the insured has no means of establishing and which the states concerned have a converging interest in leaving uncertain.
That uncertainty is no accident, and this is the point the module means to install. Using forces without insignia, groups that can be disowned, means that create no formal commitment, is a military doctrine whose function is precisely to be unprovable. It was designed to complicate an adversary's diplomatic and legal response, with no thought for insurance, and it happens to fall exactly on the boundary political violence policies draw between the covered and the excluded. An insured can therefore lose cover for a reason having nothing to do with it, with its risk, or with its conduct.
The burden of proof accordingly becomes the decisive question, and the general law answers it in a way favorable to the insured. An exclusion is in principle raised and therefore established by whoever relies on it: it is for the insurer to show the event falls within war, and doubt about attribution ought to remain with it. That is the default rule, and drafting regularly moves it. The most widespread exclusion clause in London, of the NMA 2918 type, expressly reverses it: if the insurer alleges that the exclusion applies, the burden of proving the contrary is on the insured, and doubt about attribution moves to its side. It is the first line to look for in a policy. Modern definitions reach acts committed by or on behalf of any sovereign power, whether war be declared or not, which extends the exclusion to deniable forces; some policies add a clause making official statements conclusive as to the classification of the event. The evidential rule survives and its subject matter has been widened to the point of changing its reach.
The clause making official statements conclusive deserves a pause, because it produces a result few buyers anticipate. It makes the fate of a commercial loss depend on a communiqué written for diplomatic reasons. Yet states classify the same facts in opposite ways according to what they seek: calling terrorism what is the work of a state adversary avoids entering the regime of armed conflict, and calling war what is the work of domestic groups justifies emergency measures. Neither classification is pronounced with an insurance policy in mind, and either will nonetheless decide the claim. A professional must know the clause exists, know how to spot it, and know that it is negotiable.
Three places where wordings genuinely differ must then be named, because that is where review should concentrate. The definition of war itself, which may require a conflict between recognised states or may settle for action on behalf of a sovereign power, which is not remotely the same reach. The treatment of civil war, sometimes covered, sometimes excluded, sometimes covered up to a threshold of intensity nobody can measure while the facts unfold. And the possible presence of a write back, limited in amount, for physical damage occurring in a designated zone whatever the attribution, which is the only construction that actually addresses the problem rather than relocating it.
A practical consequence follows on the notification side and it is often played badly. The insured does not have to classify the event in order to notify, it has to describe what it observed: the date, the time, the damage, the witnesses, the statements published, the condition of the site. A notification asserting a terrorist act from the outset commits the insured to a classification it cannot establish and which an insurer will turn against it if attribution shifts. Describing precisely and classifying cautiously is no manoeuvre, it is the only tenable position when the classification depends on information one does not have and may never have.
The conclusion is austere and better stated plainly than left to be discovered. This defect is not repaired after the event: once the smoke settles, positions are already taken and attribution becomes a political object over which an insured has no purchase. It is repaired in one place and at one moment only, in the drafting, before signature, by reading the definition of war rather than the list of covered perils. It is the least spectacular gesture in this whole cycle and it is probably the one that decides the largest sums.
A European telecommunications operator runs a network of masts in a border province. Between 4 and 11 November 2025, eleven sites are destroyed by explosive charges laid at night, for 42 million euros of physical damage and a covered business interruption. Nobody claims responsibility. The country's government classifies the facts as terrorist acts and arrests fourteen people. The defence minister of the neighbouring state declares, four days later, that the province is the theatre of a liberation struggle his country supports morally and to which it supplies no means. The policy, which the operator calls its political violence policy and which is in fact a terrorism and sabotage policy, covers terrorism and sabotage, does not reverse the burden of proof, excludes war, defined as any hostile action carried out by or on behalf of a sovereign power, whether war be declared or not, and contains a clause making official statements conclusive. What should the insured do, and where is the file decided?
The first useful reflex is to decline the question as posed, because the insured can neither prove nor disprove attribution, and taking a position on that ground is the one certain way to lose. What it must do is describe: the eleven sites, the dates, the methods used, the technical findings on the charges, the local police reports, the identities and status of those arrested, the chronology of the statements. Those elements belong to it and will not degrade further. The file then turns on three points, to be handled in that order. First the burden of proof: since the policy does not reverse it, the exclusion is raised by the insurer, and it is for the insurer to establish that the acts were carried out on behalf of a sovereign power, and moral support publicly claimed by a neighbour, with no means supplied, does not establish it. A ministerial statement is a fact, not proof of attribution, and the insured must say so plainly. Next the clause making official statements conclusive, which is the real danger and which cuts both ways here: the host government classifies the facts as terrorism, the neighbour speaks of a supported struggle, and one must read precisely whose statements the clause makes conclusive, since wording aimed at statements by the government of the country where the damage occurs would lead to cover. Finally the possible presence of a write back for physical damage in a designated zone whatever the attribution, the only mechanism that settles the question instead of relocating it. What must be said to the insured, even if it does not wish to hear it at the time of loss, is that this file was decided at inception: a definition of war extended to action on behalf of a sovereign power, combined with a conclusive classification clause, is the pair of clauses that makes such an event arguable, and it could be read before signing.
- 01Under a terrorism and sabotage policy, cover is widest on the smallest disorders and narrowest on the event that actually destroys: that is a constraint of pooling, which only full political violence cover lifts.
- 02Attribution is the pivot: the same damage is covered or excluded according to the identity of the actor, information the insured cannot establish.
- 03Deniable forces are a military doctrine designed to be unprovable, and it falls exactly on the boundary the policy draws.
- 04The burden of proving the exclusion lies on the insurer, unless a clause of the NMA 2918 type reverses it, and a definition extended to action on behalf of a sovereign power widens its subject matter.
- 05A clause making official statements conclusive makes a commercial loss depend on a diplomatic communiqué: read whose statements it makes conclusive.