The module on proof established that an insurer invoking the defect exclusion must show two distinct things: that the design was defective, and that the defect played a part in the damage observed. The first limb is technical and is handled with documents. The second is logical, it is almost always neglected, and it is often the harder. A badly designed structure that collapses for a reason unrelated to its defect remains covered, and that sentence, which sounds theoretical, decides files where the defect is established without dispute.
The test structuring the demonstration is simple to state: without the defect, would the damage have occurred in the same way? If it would have occurred identically, the defect played no causal role, however grave it may be. If it would have occurred later, or smaller, the defect played a partial role, and that partial case is the most frequent on a construction site. The answer to that question is not an observation but a reconstruction, which explains why it rests on counterfactual calculations of which each side produces its version.
The configuration making the question acute is concurrent causes, and a construction site manufactures them continuously. An undersized slope collapses during an exceptional rainfall event. The sizing is insufficient, the rain is real, and the collapse happened only once both came together. It is not a matter of knowing which is the true cause, which would make no sense, but of knowing what the contract does with their meeting. Three answers coexist in wordings, and they are nothing like equivalent: the exclusion carries everything as soon as it contributes, the exclusion operates only if it is the determining cause, or the indemnity is shared proportionally.
The first of those three must be named for what it is, because it is the harshest and often the least visible. A clause excluding damage caused directly or indirectly by a defect, or excluding whatever a defect contributed to in any way, turns a minor contribution into a total exclusion. On works where a design defect played for ten percent and a storm for ninety, that wording leaves the insured with nothing. The words directly or indirectly, which read like boilerplate, are here the opposite of boilerplate: they are the substance of the clause.
The wording adopting the determining cause produces the opposite result and raises a difficulty of another nature. The predominant cause must be designated among several, which is not a measurement but a judgment, and two competent experts acting in good faith can differ. It is recognized by producing all or nothing: the whole damage is covered or wholly excluded depending on which way the assessment leans. Proportional sharing, rarer, avoids that tipping at the price of a negotiation over a percentage nothing allows to be measured objectively.
One case deserves isolating because it inverts intuition: the worsening of covered damage by a defect. A covered fire spreads further than it should have because compartmentation was badly designed. The initial loss has nothing to do with the defect, but the extent of the damage owes it something. The question is no longer whether the defect caused the loss, it did not, but whether it caused the part of the damage exceeding what would have happened without it. The answer depends on the same mechanics as above, applied not to the loss but to its margin, and it is a two-stage reasoning few files carry through.
What to retain for action lies in one methodological and one documentary requirement. The first: when a defect exclusion is invoked, ask in writing which of the two limbs it rests on, and require the second to be demonstrated separately and not inferred from the first. An established defect establishes nothing about causal role, and the slide between the two happens through a simple juxtaposition of paragraphs. The second: document the external cause with the same energy one would put into contesting the defect, because in a concurrent-cause configuration it is the demonstration that the damage would have happened anyway that is worth most, and it rests on weather data, readings and measurements no longer obtainable two years later.
A temporary retaining slope collapses on October 14, damaging neighboring works under construction: 7.2 million euros. Expert examination establishes two facts. The slope was designed at 1 in 1.5 where the ground called for 1 in 2, which is a design defect. And between October 11 and 14 the equivalent of two months of rain fell, an event the nearest station puts at roughly once in thirty years. A counterfactual calculation shows a correctly designed slope would have withstood that event, but with a small margin. The policy excludes damage caused directly or indirectly by a design defect. What happens, and what would another wording have changed?
The first reflex, to plead the thirty-year rainfall as a sufficient external cause, must be resisted. It is not one here, and the counterfactual calculation says why: a correct slope would have held, so without the defect the damage would not have occurred, and the causal role is established. That is where many files stop believing they have an argument. The second point is the wording, and it is decisive. The words directly or indirectly read like boilerplate and are exactly the opposite: they make a contribution sufficient, without the defect needing to be the determining cause. With a wording adopting the determining cause, the argument would reopen, since a thirty-year event can claim that title and the small margin in the counterfactual points that way. With a sharing clause, a percentage would be negotiated. Three wordings, three outcomes, on identical facts. The non-obvious point lies elsewhere, and it concerns the small margin: it is the only element that could have helped, and it will not help here. It would have helped in the reverse configuration, where the correct slope would not have held either, because then the damage would have occurred without the defect and the first limb of the demonstration would have fallen. The difference between the two files is a few millimeters in a counterfactual calculation produced two years later, from weather records somebody had the presence of mind to request in October.
- 01An established defect establishes nothing about causal role: the two limbs are shown separately, and the slide from one to the other happens by simple juxtaposition of paragraphs.
- 02The test is counterfactual: without the defect, would the damage have occurred in the same way? Identically, no causal role; later or smaller, a partial one.
- 03On concurrent causes, three wordings give three opposite outcomes on identical facts: sufficient contribution, determining cause, or proportional sharing.
- 04The words directly or indirectly are not boilerplate: they make a ten percent contribution carry exclusion of the whole.
- 05The worsening of a covered loss by a defect is reasoned in two stages: the defect did not cause the loss, the question is whether it caused the excess portion of the damage.