The hours clause, or who decides what counts as one event

A per occurrence excess of loss aggregates every loss arising from one event before meeting the retention. The treaty's whole value therefore rests on a word no formula defines, and that word is event. A windstorm does not arrive at an instant across a whole territory: it comes in over a coast, crosses for two days, weakens, and a second depression may follow the same track forty hours later. Losses arrive continuously, the retention is a single threshold, and someone has to decide where the sequence starts and where it ends. The hours clause is the contractual answer to that question, and it is in no way incidental: it decides how many times the retention applies.

Its mechanics are deceptively simple. The clause opens a window of an agreed number of hours, and every loss occurring inside that window counts as a single occurrence, whatever their number and whatever their geographic spread. The usual durations follow the physical duration of the peril rather than accounting convenience: seventy-two hours for wind and cyclones, one hundred and sixty-eight hours, that is seven days, for perils that unfold more slowly, seismic aftershocks, river flooding, freeze episodes. One cedant therefore routinely carries several hours clauses within a single program, one per family of perils, and confusing them when reporting a loss is a commonplace and expensive mistake.

The point textbooks pass over quickly is whose hand places the window. It is neither the reinsurer's nor a third party's: it is the cedant's, after the fact, once the sequence has ended and the amounts are known. It picks the starting hour, the window must be continuous, and the choice is exercised once only. It is thus a genuine economic lever, exercised in full knowledge of the figures, and the treaty grants it because refusing would make the recovery depend on a meteorological convention neither party controls. The cedant places the window where it captures the most, and the reinsurer knows this when quoting.

What "the most" means is where intuition goes wrong. Aggregating always looks favorable, since a single occurrence means a single retention instead of two. That stops being true the moment the aggregated sequence exceeds the top of the layer: the cedant then pays one retention, exhausts its limit, and bears net everything above the layer. By splitting, it pays two retentions, but it presents an unsaturated layer twice, and total recovery can be higher. What settles the two readings is therefore not the retention, which argues for aggregating, but the limit, which argues for splitting. And a third quantity joins them: how much capacity is left for the rest of the year.

The European market learned this in a dated case. Windstorms Lothar and Martin, in December 1999, cost French insurers roughly six billion euros over three days. Several cedants then found that their seventy-two hour clause split the sequence into two separate occurrences, and therefore two retentions instead of one: one mutual carrying a fifteen million euro retention absorbed thirty million net. The 2000 and 2001 renewals saw the clause extended to ninety-six and then one hundred and sixty-eight hours on European windstorm layers. The lesson in form counts as much as the lesson in substance: an hours clause is a market term, it is renegotiated like a price, and it is renegotiated after being tested.

The clause makes sense because a hurricane has a beginning, an end and a name. It becomes absurd applied to a peril with no natural occurrence. A software compromise introduced one day, dormant for months, then activated in successive waves cannot be dated: no one can say which day the event took place, and shifting the window by a few hours then changes the number of occurrences with no external fact requiring it. This is a live problem rather than a theoretical one, since cyber catastrophe bonds are issued on a per occurrence basis, which applies an occurrence notion, and therefore an hours clause, to a peril that carries none. The disagreement is not discovered at inception, it is discovered at the claim.

Four points of the wording therefore decide what the clause is actually worth, and they are read before signing. What opens the window: the first loss, or a meteorological fact named and dated by a third party body. The scope of the window: a single window for the whole book, or one window per zone and per country, which multiplies occurrences on a storm crossing several markets. The combination of perils: a sequence where wind precedes flood may count as one occurrence or as two depending on whether the clause treats them together or apart. And the agreement of definitions: if the policies the cedant issues define the event differently from the treaty, the gap stays with the cedant, and it finds that out while settling.