On a large project, the first insurance decision is not the choice of a cover, it is the choice of an architecture. Either the project owner places one policy covering all parties, or each company brings its own. Both structures exist, both are defensible, and they produce losses that look nothing alike. An underwriter who has never watched a file unwind under each regime always underestimates what the second costs in time.
The project-wide policy brings together under a single contract the project owner, the main contractor, subcontractors of every tier and often suppliers and engineering firms. Its first advantage is arithmetic: one limit, one deductible, one insurer, therefore a loss settled without having to decide who was at fault. The second runs deeper and rests on a legal consequence. A policy insuring several persons for the same interest bars the insurer from exercising recourse against one of them after indemnifying another: one does not turn against one's own insured. Waiver of subrogation between co-insureds is therefore not a commercial concession, it is the very mechanics of the structure, and it is what removes the liability war that ordinarily follows any construction loss.
That same mechanics produces its drawback, and it is symmetrical. By removing recourse, the structure also removes the most effective incentive to work well. A company whose subcontractor broke something knows the policy will pay and that nobody will come asking, which imperceptibly changes the attention paid to method. Insurers answer with high deductibles borne by the responsible company, which restores part of the spur without restoring the litigation. The deductible then stops being a simple sharing of risk and becomes an instrument of discipline, which explains levels that look unrelated to expected loss experience.
Separate policies invert everything. Each party insures its own liability and its own property, recourse stays open, and every loss starts with a question of apportionment. The practical consequence is less the cost than the delay: a two-million loss on which three companies pass the cause around can immobilize a site while no insurer advances funds, for want of knowing which one owes. On a structure where every week of stoppage has a price, that delay often costs more than the apportionment itself. That is why the heaviest projects move toward the project-wide policy, and it is not a question of premium.
Between the two regimes opens a gap for which nobody is responsible and which everyone discovers at the same moment. Each policy has a scope defined by dates, by a list of insureds and by a description of the covered works. A package awarded mid-project to a company that arrived after placement, equipment delivered to site before the cover period starts, a party falling under none of the definitions retained: each time, the damage falls between two contracts. The most frequent case is off-site storage, where equipment sometimes waits for months in a third-party warehouse, under a construction policy covering the site only, and under no transit policy since nothing is moving.
Waiver of subrogation then deserves to be read for what it actually carries, rather than for what its heading suggests. It operates between the persons named as insureds, which means checking that the naming really covers everyone, including companies that joined later, and not merely the main contractor and its direct subcontractors. It does not operate toward third parties, and an equipment supplier may well sit outside the list while being the origin of the loss. It generally does not operate in case of intentional fault either. Finally, it does not strip the project owner of its contractual rights: it extinguishes the subrogated insurer's recourse, not the client's own action against its contractor for what is not indemnified, starting with the deductible and with delay penalties.
Those two remaining heads are in fact what projects handle worst. The project-wide policy removes the dispute over physical damage, but it does not touch the construction contract, which carries delay penalties, a performance bond and liability for uninsured overruns. A loss settled without argument by the insurer can therefore be followed, six months later, by proceedings between the same parties over the delay it caused. Believing that waiver of subrogation has pacified the project is a common misjudgment: it pacified one layer, the damage layer, and left intact the contract layer, which is often the heavier one on a structure whose commissioning drives revenue.
A project owner builds a data center. It places a project-wide policy naming as insureds the project owner, the main contractor and its subcontractors, with waiver of subrogation between insureds and a 250,000 euro deductible per loss borne by the responsible company. The generator sets, bought directly by the project owner from a supplier not named in the policy, are stored for eleven weeks in a warehouse rented fifteen kilometers from site, for want of a finished slab. On April 9, a fire destroys the warehouse and the six sets, worth 4.3 million euros. Investigation establishes that the loss originates in a defective battery charger supplied by the vendor. Commissioning of the center is pushed back seven months. Who pays what?
The first reflex, invoking the project-wide policy, runs into a question of scope rather than of cover: the sets were not on site. Almost all construction policies cover off-site storage, but on express conditions, with a sub-limit and prior declaration of the address. Eleven weeks of storage without declaration is the commonest route to a decline, and it is the first point to check before any other. If the extension exists and was declared, the policy pays up to its sub-limit, which is almost always well below 4.3 million. The 250,000 euro deductible then raises an unexpected question: it is borne by the responsible company, yet no insured company is responsible here, the triggering event coming from a supplier of the project owner. The deductible therefore falls back on the project owner itself, which is not what the clause meant but is what it says. On the favorable side, the supplier is not a named insured, so the waiver of subrogation does not protect it, and the paying insurer will be able to turn against it, which also gives the project owner its own action for the uncovered share. Finally, the seven months of delay do not fall under this policy but under advance loss of profits cover, if placed, and that cover will require the underlying physical damage to be itself covered: if off-site storage was undeclared, the decline on the equipment mechanically carries the decline on the delay. An eleven-week administrative omission thus governs two thirds of the file.
- 01The project-wide policy removes the liability war because an insurer does not turn against its own insured: waiver of subrogation is the mechanics of the structure, not a favor.
- 02With separate policies, what costs is not the apportionment but the delay: nobody advances funds while it is unknown who owes.
- 03The gap sits at scope boundaries, and off-site storage is its commonest case: on a defined site, outside transit, with no declared address.
- 04The waiver operates only between named insureds: a supplier outside the list stays open to recourse, and the insured keeps its contractual action for the deductible and penalties.
- 05A clause placing the deductible on the responsible company says nothing about the case where the responsible party is not an insured, and it then falls back on whoever placed the policy.