Defective design, and the three ways of treating it

There is one clause on which the usefulness of a construction policy depends more than on any other, and it sits in a few lines in the middle of the exclusions. It is the one saying how the contract treats defective design, material or workmanship. Three wordings circulate, they look alike on a quick read, and the gap between the first and the third can amount to the whole loss. An underwriter who does not know which one they signed does not know what they bought.

The logic of the exclusion is worth grasping before arguing about it. A construction policy insures fortuity, not the quality of the work: if it paid to redo everything done badly, it would become a performance guarantee, that is, insurance of the contractor's contractual duty to its client. The insurer would then be covering a risk the insured controls entirely and profits from directly by economizing on execution. The defect exclusion is therefore not drafting meanness, it is the boundary between insurance and commercial warranty. The whole difficulty is that a defect does not stay at home: it breaks something else.

The first wording is the flat exclusion. It excludes the defect and everything flowing from it: the badly designed item, and the damage it caused to the rest of the works. Its consequence is brutal and rarely measured at placement. A tank whose stress calculation was wrong bursts, floods a machine hall and destroys two pumps: nothing is covered, neither the tank nor the pumps, because everything descends from the same defect. That wording turns the policy into cover against external events only, storm, fire, dropped load, and leaves out the most frequent family of losses on a complex site.

The second wording, by far the most common on large projects, excludes the defect but covers consequential damage. The badly designed item stays with the contractor; what it damaged around it is covered. On the previous example, the tank is not indemnified, the two pumps are. The boundary is clear in the statement and formidable in application, because it forces a definition of what the defective item is. A faulty weld on a pipe: is the weld defective, in which case almost all the loss is covered, or the whole pipe, in which case the excluded share becomes considerable? Wordings speak of damaged unit, defective portion, defective item, and those words do not cover the same ground. The split of the amount turns on that carve-up, not on the principle.

The third wording goes further and covers the making good itself, excluding only the betterment cost. The insurer pays to replace the defective item, but not what it costs to replace it with something better than what was specified. If the tank was to be six-millimeter steel and is rebuilt in eight, the policy pays the six-millimeter equivalent and the contractor bears the difference. That wording is the only one leaving the insured real protection against a design error, and it is priced accordingly. It also carries a difficulty of its own: establishing what the correct item would have cost means reconstructing a quotation that never existed, and that reconstruction is the real ground of discussion.

A fourth question runs through all three wordings and always arises too late: is the defective design the insured's, or anyone's? On a large project, design is the work of an engineering firm that is neither the project owner nor the contractor, sometimes of an equipment supplier on the other side of the world. A policy written in the name of all parties, which is the rule on project-wide policies, makes the exclusion apply to a defect committed by any one of them, including the party the insured did not choose and could not supervise. Conversely, a policy excluding only the insured's own defect leaves a neighbor's defect covered. That nuance does not appear in the exclusion clause: it appears in the definition of the insured, thirty pages earlier.

Finally there is the articulation with the French statutory regime, which does not operate on the same clock. The ten-year warranty and the proper-functioning warranty apply after acceptance, to disorders compromising the soundness of the works or making them unfit for their purpose, and they weigh on the constructors. The defect clause plays during the works. One and the same design flaw can therefore fall under the defect clause if it shows up before acceptance, and under the ten-year warranty if it shows up after, with two different insurers, two different limitation periods and two different logics of proof. The date of manifestation, once again, decides everything, which is why an emerging disorder is documented day by day rather than at the moment it becomes visible.