One policy or several: what the choice really decides

A large project can be insured in two ways, and the choice is made early, often by default. Either the project owner buys a single policy covering all parties and all the works, or each contractor brings its own for what concerns it. Put that way, the subject looks like a question of administrative organization. It is not one. The choice decides who answers at the time of the loss, how fast the works will be repaired, and where the gaps will sit, because there are gaps in both configurations, simply not in the same places.

The single policy has an advantage that explains its popularity on large projects and that is not the one usually cited first. Economy and uniform terms are put forward; the real strength is elsewhere, in removing the question of attribution. Damage to the works is covered because it is the works, without first having to establish which contractor caused it. On a site where thirty firms work at the same time on the same square meters, that question can take two years, and for those two years the works are not repaired. Pay first and attribute later, or not at all, is the service the project owner is actually buying.

The flip side is that this policy stands in for liability without abolishing it, and the parties get that wrong. A contractor covered as an insured under a single policy keeps its contractual liability toward whoever engaged it. It remains liable for the deductible, for delay penalties, for everything above the limits, and for what the policy excludes. That residual liability is nothing like theoretical on a large project where deductibles run to hundreds of thousands of euros per loss. A contractor that cancelled its own cover because the project supplied one then discovers it has nothing left for the share that remains with it, and that it no longer has it at the moment it is needed.

Separate policies produce the reverse configuration, with qualities of its own. Each contractor stays covered by its usual insurer, which knows its trade and its losses, and the project owner carries no administrative burden. The price of that dispersion is that any damage becomes first a question of attribution: the responsible party must be identified before anyone pays. The insurers of three neighboring firms watch each other, each with an interest in the cause sitting with another, and the repair waits. The project owner, party to none of those contracts, watches its works stay as they are while others argue.

The most characteristic gap in that configuration is not the slowness, however, it is the seam between two scopes. Each policy covers its insured's works. Damage caused by one contractor to works performed by another therefore falls between two definitions: they are not the responsible party's works, and the victim's insurer covers its works but may assert its own recourse. Damage caused by nobody, an infiltration whose origin stays undetermined between two packages, finds simply no contract that says it is concerned. The single policy handles both cases without thinking about it, because it defines the works as a whole; separate policies handle each as litigation.

An intermediate configuration is worth knowing because it is frequent and carries the worst of both: a single policy for the structural works, and separate policies for the technical packages. The boundary then follows the split of the contracts, which is not the split of damage. A fire starting in an electrical board, a technical package, that damages the structure, structural works, sits astride, and each of the two contracts reads as if the other did not exist. It is not an intended configuration, it builds up by sedimentation when a package is added late or when a firm refuses to join the project policy.

A word finally on coinsurance, which overlays that choice without merging with it. A single policy on a large project is almost always carried by several insurers sharing the risk in fixed shares, led by a leader. That structure changes nothing about the cover, each coinsurer owing its share on the same terms, but it changes the handling: the leader adjusts, the others follow, and a disagreement among them shows up as a partial settlement rather than a declinature. The useful distinction is therefore this: the choice between one policy and several bears on the number of CONTRACTS and decides attribution; coinsurance bears on the number of INSURERS on one contract and decides the pace of settlement.