A construction policy assumes a construction site. It is written for works that progress, on an occupied site, guarded, lit, where people move about every day and where an anomaly is spotted within half a day. That assumption is so obvious it is never stated, and yet it stops being true more often than one thinks: a liquidation, a termination, litigation suspending a permit, a financial dispute, an administrative moratorium. The site stops, the firms leave, and what remains is unfinished works nobody watches any more.
The exposure of a stopped site is not lower than an active one's, it is different and often greater. What disappears are the risks tied to execution: machines no longer overturn anything, cranes no longer drop loads. What appears is everything else. Excavations stay open and fill with water. Unsupported slopes destabilize at the first wet season. Young unprotected concrete freezes. Steel structures erected without permanent bracing wait for a gust. And above all, an unguarded site becomes, within weeks, a target for copper theft and intrusion, which is the leading cause of loss on stopped projects.
Policies anticipate this through a suspension or cessation of works clause, and what it does must be known. It typically provides that cover ceases, or is heavily restricted, beyond an interruption expressed in days, thirty or sixty depending on the wording, and it often makes its continuation conditional on notification to the insurer. The count is calendar-based, it runs with nobody keeping it, and it starts on a day that is hard to determine, since a site almost never stops by a dated decision: it slows, a few crews stay, one firm comes back for three days, and one can no longer say when the interruption actually began.
That question of date is complicated by a distinction the wordings make and practice ignores. A planned seasonal shutdown, a year-end closure, an interruption programmed for a technical reason are not interruptions within the meaning of the clause, because they are part of the site's normal course. An imposed stoppage, of indefinite duration, is one. Between the two lies the frequent case: a site officially suspended for a fortnight over a supply problem, which lasts four months because the supplier filed for insolvency. Each week, resumption is announced for the next, and no notification is given since the stoppage is supposed to end shortly.
The main contractor's liquidation adds a difficulty of its own, and it is not a physical one. On a policy bought by that contractor, the policyholder's disappearance raises the question of the contract's continuity, of payment of subsequent premiums and of who, from now on, is the insurer's counterpart. On a single policy bought by the project owner, that problem does not arise, which is an argument rarely cited in favor of that configuration and which nonetheless weighs heavily on the day it applies. In both cases, the body conducting the liquidation has neither mandate nor budget to maintain a site, and guarding is the first expense a liquidation stops paying for.
What becomes of the cover calendar during the stoppage must also be seen, because time keeps passing for everyone. The policy period runs to its contractual expiry, which the stoppage does not push back. A site interrupted for twelve months therefore consumes twelve months of its cover without advancing a meter, and resumption happens on a policy with far less left than the new schedule requires. Extension is then negotiated from the worst possible position: after a stoppage, on degraded works, by an insured whose financial soundness has just been called into question by the very events that caused the stoppage.
What to retain for action comes in two stages. At the moment of stoppage, do three things that cost almost nothing: notify the interruption without waiting to know how long it will last, because a premature notification has no adverse consequence; establish a dated record of the state of the works, which will be the only reference available at resumption; and maintain guarding and weatherproofing, which are exactly the conditions on which restricted cover remains in force. At the moment of resumption, do one: compare what is left of the policy period with the schedule of remaining works, and deal with the gap before reopening the site rather than after.
The main contractor on a housing scheme is placed in liquidation on February 8. The site, structure complete and finishing works started, stops gradually: the last crews leave on March 2, guarding ceases on March 15 for want of a payer. The single policy, bought by the project owner, expires on November 30 and carries a cessation of works clause restricting cover to theft, fire and natural events beyond sixty days of interruption, subject to guarding and weatherproofing being maintained. No notification was given. On May 12 an intrusion removes 340,000 euros of copper cabling and technical equipment; on June 3 water ingress damages installed partitions to the tune of 480,000 euros. A new contractor resumes on September 1, with ten months of works remaining. What happens?
The start of the sixty days is the first question, and it is arguable: February 8 of the liquidation, March 2 of the last crews leaving, or a date in between? The most defensible is March 2, the physical fact, and it puts expiry of the period at May 1. Both losses are therefore later, and restricted cover applies to them, which is rather good news since it targets theft precisely. But it is conditional on guarding being maintained, stopped on March 15, and on weatherproofing. The May 12 theft therefore falls within the restriction whose condition is unmet, and the June 3 ingress does not even fall within the three retained perils. The reflex is to conclude on a double loss of 820,000 euros and stop there. The third gap is heavier and nobody looks at it: the policy expires on November 30, and the September 1 resumption starts on ten months of works. Three months of cover remain for ten months of construction, and the extension will have to be negotiated on works left six months unwatched, degraded and unrecorded, which is the worst possible position. The non-obvious point is that the act that would have paid most was not guarding, for which nobody had a budget, but the dated record of March 2: free, it would have established the state of the works when the crews left, separated what degraded during the stoppage from what was already so, and given the September negotiation the one reference it will not have.
- 01A stopped site is not less exposed but differently: execution risks disappear, open excavations, young concrete, unbraced structures and above all copper theft remain.
- 02The cessation of works clause counts calendar days from a date that is hard to determine, a site almost never stopping by a dated decision.
- 03A planned seasonal shutdown is not an interruption within the meaning of the clause; an imposed stoppage announced for a fortnight that lasts four months is one, and notification is never given.
- 04The policy period does not extend itself: twelve months of stoppage consume twelve months of cover without advancing a meter, and extension is then negotiated from the worst position.
- 05The act that pays most at the moment of stoppage is free: a dated record of the state of the works, the only reference available at resumption to separate what degraded during it.