The tower: four contracts sold as a single number

A layered program exists for an arithmetical rather than a doctrinal reason: no single insurer carries alone the limit a billion euro deal requires, because its own underwriting discipline forbids it to commit beyond a fraction of its funds to one risk. The limit is therefore divided into successive tranches, each written by a different carrier, the first attaching above the retention and each of the others triggering only once the one below it is exhausted. A reader who has only seen mid-market deals meets here an object it believes behaves like a single policy. It does not, and this whole module is about why.

The first fact to hold is that a tower is not one contract but as many contracts as there are tranches. The primary layer is the one that underwrote: it received the reports, negotiated the exclusions, held the drafting discussion. The layers above adhere through a short policy stating that they follow the terms of the first, what practice calls following form. The phrase is convenient and it is misleading, because it almost always comes with reservations: the insurer of an upper tranche adds its own portfolio exclusions, its own dispute resolution clause, sometimes its own definition of what constitutes a recoverable loss. Following form means following form except for what I am telling you here, and what you are told sits on one page that is rarely read.

The second peculiarity concerns attachment, and it decides considerable sums. An upper tranche responds only once the one below is exhausted, but exhausted by what? By actual payment, or by an admission of liability, or by a settlement the lower insurer has accepted? The distinction looks academic and it is not: if the primary settles at eighty per cent of its tranche and the insured bears the balance, a strict exhaustion clause leaves the upper layer out of play, since the tranche below has not been paid in full. An insured who accepted an attractive settlement at the bottom of the tower has hollowed out the middle of its own cover, and only discovers it on the day it needs it.

Then comes the point this module means to install, because it costs more than the preceding ones and is handled worse: coordinating notifications. Each tranche is an autonomous contract, so each tranche carries its own duty to notify a circumstance capable of engaging the cover, and those duties are not discharged by one another. A claim is born small, it is notified to whoever seems concerned, that is to the primary layer, then it grows for eighteen months and ends up reaching the third tranche. The two upper layers then receive a notification whose triggering facts are eighteen months old, and they raise what their policy provides for: late notification, prejudice caused by delay, sometimes forfeiture depending on the law governing their own contract.

The natural objection is that an upper tranche insurer suffers no prejudice from hearing late, since it would have had nothing to do while the claim stayed below its tranche. The objection is sound in fairness and weak in law, for a reason a professional must be able to state: the upper tranche insurer loses the right to take part in defence strategy at the moment it is settled, that is in the first months, when one chooses to admit or to contest, when counsel are appointed, when the case theory is fixed. It receives a constituted file, a position already taken and fees already incurred. Prejudice by delay is no procedural fiction, it sits exactly there, and that is why serious wordings require notification to every tranche once the claim reaches a watch threshold rather than the threshold of that tranche itself.

A practical consequence follows, and it is the reflex this module wants to build: at the moment of notifying, one notifies the whole tower, at the address and in the form each policy prescribes, and keeps proof of despatch for each tranche separately. That looks heavy and it is very little work compared with what a lost tranche represents. The difficulty is never the effort, it is the mental picture: the person notifying believes it is addressing its insurance, in the singular, when it is addressing four undertakings with no contractual link between them, which pass nothing to one another.

It remains to say where terms actually stop following, because that is where reviewing a tower concentrates when time is short. Three places produce most of the divergence. The definition of recoverable loss, which may include at the bottom of the tower a head of damage that the tranche above excludes, with the disconcerting effect that one claim changes amount as it changes floor. Each carrier's portfolio exclusions, which answer to its own commitments and have nothing to do with the deal. And the dispute resolution clause, which may send the insured to a court for the first tranche and to arbitration for the third, on the same facts, with a risk of contrary decisions that nobody will coordinate.

The conclusion to draw is not that a tower is a bad object, it is that it demands work its commercial presentation conceals. It is sold as a limit, one reassuring number, whereas what is bought is a set of contracts whose coherence has been assumed rather than verified. The useful professional is the one who, before signing, reads the upper tranche policies rather than the table summarising them, and who, after signing, keeps the list of notification addressees current in the deal file. Neither gesture is difficult and neither happens by itself.