A construction policy is bought on a project that exists only on paper, and it covers three or four years during which that project never stops changing. It is a situation with no equivalent in other classes. A warehouse insured on January 1 is still the same warehouse on December 31; a site insured at contract signature no longer resembles, eighteen months later, what was described in the submission. Methods have changed, a package has changed contractor, a temporary phase has been added to claw back a delay, one more season has been spent with an open excavation. The contract has not moved, and the gap between the two widens with nothing to signal it.
Three things that ordinary language conflates must first be separated. The normal progress of a site, first: the structure goes up, present value grows, exposure shifts. That is the very unfolding of what was insured, and nobody has to declare it. A change to the project, next: an additional section, a new structure, an extension of scope, which calls for an endorsement because it bears on the subject matter of the contract. Aggravation of the risk, last, which is something other than the first two: the project stays the same, but the way of doing it exposes more. These three call for different acts, and confusing them produces either useless paperwork or a costly silence.
Aggravation is the hard category, because it is rarely recognized at the moment it is decided. It never presents itself as an insurance decision: it presents itself as a construction choice. Moving from tower crane lifting to mobile crane lifting on a public road, replacing a cofferdam with dewatering, executing in one phase what was to be done in two, carrying out earthworks in the wet season, densifying site facilities in one area because another has stayed occupied. Each of those decisions is taken in a works meeting, justified by schedule or cost, and shifts the insured risk. It is taken by someone whose subject it is not, and communicated to nobody whose subject it is.
The penalty for silence is not uniform, and that is what makes it poorly understood. Under French insurance law, an inaccurate statement or an omission is assessed according to whether it is in good or bad faith, and the consequences bear no comparison: bad faith leads to nullity of the contract, with premiums retained by the insurer; good faith leads to a proportional reduction of the indemnity, in the ratio of the premium paid to the premium that would have been due. Two regimes, one omission, and a boundary judged on facts. A site that changed method without giving it a thought and a site that kept quiet about an aggravation it knew to be significant are therefore not in the same position, and the difference is argued on written traces: minutes, method statements, emails to the broker.
Proportional reduction deserves to be looked at for what it is, because it is often presented as a mild sanction. It applies to the indemnity the ratio between the premium paid and the premium that would have been due at the insurer's rates for the actual risk. On a risk whose rating would have doubled, it halves the indemnity. An eight million loss settles at four, and this is not an expert disagreement but a calculation rule applied to a loss that is otherwise fully covered. The retained share is therefore silent: it appears in no exclusion, no deductible and no limit, that is, in none of the three places one usually looks to find out what a policy leaves behind.
A second asymmetry lies in the calendar of knowledge. The insurer learns of the aggravation at the time of the loss, that is, once it has produced its effect, and often learns it from a loss adjuster's report describing the actual method. The site knew of it months earlier and could still have declared it without consequence: a declared aggravation is a rating matter, it calls for an additional premium or a new condition, and an insurer unwilling to keep the risk must cancel with notice, which leaves the project time to replace the cover. Declared afterwards, the same aggravation is no longer a rating matter, it is a sanction matter. The content of the fact is identical, only its date has changed, and that date makes all the difference in treatment.
Finally, a construction policy has a feature that worsens the problem: its insureds are multiple. Project owner, main contractor, subcontractors, sometimes suppliers and design offices appear as insureds under the same contract. Knowledge of a fact by one of them is not necessarily attributable to the others, and the drafting of the declaration clause decides who must know and who must speak. Some wordings retain only the knowledge of the persons named as responsible for the insurance program, which protects insureds whose organization is a stranger to the contract; others retain the knowledge of any insured, which carries up into the contract what a site manager decided on a Tuesday morning. That reading is not improvised at the time of the loss, and it is what determines what the word omission covers on this particular contract.
An underground structure is insured on the basis of a submission describing diaphragm wall support and excavation in two phases with an intermediate slab. Fourteen months into the works, to claw back five months of delay, the main contractor decides to excavate in a single phase, without the intermediate slab, with reinforced dewatering. The decision appears in site meeting minutes and in a method statement signed off by the design supervisor. Neither the broker nor the insurer is informed. Eleven months later, a water inflow causes damage at the base of the excavation and damage to two neighboring buildings; the loss is assessed at 8.4 million euros. The adjuster reconstructs the actual method. The insurer establishes that its rating would have been increased by 90% for that method. What becomes of the indemnity?
The reflex is to look for an exclusion, and there is none: the loss is sudden damage to works under construction, exactly what the policy covers. The argument happens elsewhere, on disclosure. Bad faith must first be set aside, and the facts of the file help rather than hurt on that point: the decision is written down, signed off by the design supervisor, recorded in minutes. A contractor that conceals does not minute. Proportional reduction therefore applies, in the ratio of the premium paid to the premium that would have been due, a ratio of 1 to 1.9: the indemnity falls to roughly 4.4 million, and four million stays with the insureds without any clause of the policy saying so. The non-obvious point lies elsewhere. That change of method was decided to claw back five months, and eleven months later it produced a loss whose repair will cost more than the delay it avoided. The useful act required no legal expertise: the method statement existed, it was signed off, and a copy to the broker at the moment it was signed off would have turned a sanction matter into an additional premium matter, of an order of magnitude bearing no relation to four million.
- 01Three situations not to conflate: the site's normal progress, which is not declared; a change to the project, which calls for an endorsement; aggravation of the risk, which calls for a disclosure.
- 02An aggravation always presents itself as a construction choice, never as an insurance decision: it is taken in a works meeting, by someone whose subject it is not.
- 03A good-faith omission leads to proportional reduction of the indemnity, bad faith to nullity: the boundary is judged on written traces, minutes and method statements.
- 04Proportional reduction appears in no exclusion, no deductible and no limit: it is a silent retained share, invisible in the three places one usually looks.
- 05The same fact disclosed before is a premium matter, disclosed afterwards a sanction matter: only the date changes, and on a multi-insured policy the clause decides who had to know and who had to speak.