Among the regime's perils, drought holds a place its discretion makes deceptive. It produces no images: no flooded street, no torn-off roof, no evening news. It produces cracks, spread across tens of thousands of houses, notified months or years after the summer that caused them, and it has become one of the regime's heaviest items. An underwriter who has never had to explain it to a committee usually discovers the subject through a reserve that is strengthened three years running.
The mechanism is well known and it is slow. Clay soils shrink as they lose water, swell as they take it back, and that differential movement under shallow foundations cracks masonry. What matters for a portfolio is the conjunction of two attributes, and of two only: a sensitive clay soil, and a building whose foundations are insufficiently anchored. Exposure is therefore neither an elevation nor a distance to a watercourse: it is a geological map crossed with a construction type, and almost no insurance file carries the second.
The first difficulty for a portfolio is that this peril has no footprint in the sense the others have. A windstorm crosses in one night, a flood occupies a catchment for a week. A drought settles over entire regions for a season, and its effects appear on a calendar particular to each house, according to the depth of its foundations and the presence of a tree eight meters away. There is no event in the sense the rest of the trade means: there is a season, and a long tail of notifications.
The second difficulty follows directly and it is the costly one: the notification lag. A dry summer produces damage that appears the following autumn, becomes visible in winter, is notified in spring when the owner grows worried, and waits for an order recognizing the quarter concerned. An accounting year can therefore present itself as excellent on this peril while its burden is already formed and merely not yet notified. Treating that lag as good fortune is the classic error, and it is paid two years later on a reserve that has to be explained.
The third difficulty is the nature of the event in the regime's sense, which is administrative rather than physical. Recognition is granted by commune and by quarter: two neighboring communes on the same clay formation may not be recognized for the same period, and one house may fall under two successive orders for damage that never stopped progressing. For a portfolio, this means that the number of claims depends on a public decision whose rhythm is neither that of the climate nor that of the cracking.
One must add what prevention can and cannot do, because the question always comes. On existing buildings, the remedies concern the surroundings: controlling water discharge, removing or managing the thirstiest vegetation, avoiding works that dry one side and not the other. They reduce aggravation, they do not remove the cause. On new buildings, everything is decided at design, through a soil investigation and suitable foundations. A portfolio of nineteen-seventies houses on sensitive clay therefore carries a hazard that can no longer be corrected, only known.
The conduct to adopt for a portfolio differs from that of other perils and comes to three steps. Risks are attached to the map of clay formations rather than to an administrative grain, without which exposure is invisible. The burden of this peril is tracked over three rolling years and never on the current year, because the current year shows only what has already been notified. And for the most exposed segments, the year of construction and the foundation type are recorded, being the only two attributes that separate a portfolio that suffers from a neighboring one that does not.
A householders mutual closes its financial year. On the drought peril it has paid 4.2 million euros, against 9.8 the previous year, and the committee sees in this the end of an adverse cycle: the burden is halved, the reserve for incurred but not reported claims is held at last year's level, and the technical result of the householders line turns positive again. Three facts appear elsewhere in the same year's documents. The summer was the driest in fifteen years across two of the three regions where the mutual operates. Forty-one communes in its portfolio have filed a request for drought recognition, and none has yet received an answer. Finally, 38% of the portfolio sits on sensitive clay formations, and the year of construction is recorded for only one contract in four. What must be said to the committee?
The fall in burden is not an improvement in the risk, it is the lag proper to this peril, and the committee is reading a year through an instrument that does not measure what it thinks. The 4.2 million paid this year are, in the main, the tail of previous summers; what the dry summer has just built up will appear only as cracks become visible, owners notify, and orders are issued. The file's three facts all say the same thing and corroborate one another: an exceptional summer over two regions of presence, forty-one communal requests pending, and a portfolio more than a third of which sits on sensitive clay. Holding the reserve for claims not yet notified at last year's level therefore amounts to reserving an ordinary year after a season that is not one. The practical consequence is twofold and is not settled by a single step. On reserving, the quantity to look at is not the year's burden but that of three rolling years, and the trigger for setting the reserve is not the order but the dry season: a burden awaiting an administrative act to be recognized is a burden already incurred. On portfolio knowledge, the absence of a construction year on three contracts in four is the real gap, because it makes it impossible to separate what is exposed from what is not: two houses on the same clay, one from 1975 on shallow foundations and the other from 2015 after a soil investigation, do not carry the same risk and count here as two identical lines. The committee may decide to change nothing in its reserve; it cannot decide it believing the cycle is over, because half the information it is ruling on has not yet arrived.
- 01Drought exposure is neither an elevation nor a distance to a watercourse: it is a geological map crossed with a construction type.
- 02This peril has no footprint in the sense the others have: there is a season, then a long tail of notifications spread over years.
- 03A year can look excellent while its burden is already formed and not yet notified: the lag is not good fortune.
- 04The event is administrative: recognition by commune and by quarter, on a rhythm that is neither the climate's nor the cracking's.
- 05Year of construction and foundation type are the two attributes separating a portfolio that suffers from a neighboring one that does not.