One claim or several: the aggregate, the country limit and the seam between layers

An insured remembers the figure printed at the head of its policy and believes it knows what is available. In political risk, that figure is almost always the least binding of the three that exist. There is the per-claim limit, which bounds one event; the annual aggregate, which bounds the sum of all claims in the period; and the country limit, which bounds what the insurer will expose on a given territory, all causes combined. The third is the one that most often bites, and it is the only one the insured sees nothing of day to day, because it lives in the insurer's underwriting policy and its reinsurance treaties rather than in the contract it signed.

The first question in a serious claim is therefore rarely the amount, it is the count: does a state measure hitting three subsidiaries, five contracts and two plants constitute one claim or several? The answer drives two things in opposite directions. A single claim carries one deductible and one waiting period, which serves the insured; but it opens only one per-claim limit, which harms it as soon as the total loss exceeds that limit. Several claims multiply the deductibles and multiply the limits. The same definition helps and harms the same insured, and what decides is not a principle but arithmetic: the position to defend depends on the relationship between total loss, per-claim limit and deductible, and it is computed both ways before being written into a notification.

Wordings settle this with two families of tools, one of which is borrowed from a neighboring line and adapts poorly. The first attaches to a single claim everything flowing from a common cause, which suits a single decree striking several assets well enough. The second applies an hours clause, inherited from natural catastrophe: everything occurring within a window of seventy-two or one hundred sixty-eight hours forms one event. That second family assumes a peril with a natural duration, which a hurricane has and a state measure does not. Applied to a creeping expropriation built over eighteen months, the hours clause produces either a fragmentation into as many events as there were measures, each with its deductible, or nothing at all, and neither reading describes the economics of the file.

The country aggregate introduces a difficulty of another kind, because it is shared with people the insured does not know. The capacity an insurer devotes to a country is finite, and it is often retroceded with a per-territory cap. When a measure strikes an entire sector, several insureds notify on the same event and against the same envelope. The diligent insured, whose file is complete and notified early, then competes with claims it cannot see, in a queue whose order nobody publishes. Its recovery therefore depends in part on other people's losses and on when each of them notified, which is a very concrete argument for notifying quickly, independent of any contractual consideration.

Then there is reinstatement, which produces the most counterintuitive effect in the whole arrangement. If the limit does not reinstate after a loss, the first event consumes the cover for the rest of the period. Yet a country that has just seen a sectoral expropriation or a closure of its FX market is precisely a country where the probability of a second event has just risen. The insured therefore finds itself uncovered at the moment the risk is highest, and this results from the normal operation of the contract rather than from any accident. The event that proves the risk is real is also the one that removes the protection.

That leaves the seam between layers, which is where the most money is lost for the most foolish reason. A political risk program of any size is built in successive layers, placed with different markets, on different wordings. Nothing guarantees that the primary layer and the excess layer define an event the same way. If the primary aggregates three measures into one claim by common cause, while the excess applies an hours clause that separates them, then none of the three claims as the excess sees them reaches its attachment point, and that layer never responds. The insured has paid a premium for capacity that structurally cannot trigger, and nothing in reading each contract on its own reveals it: the defect exists only in the relationship between the two.

From all this follows a method that is simple to state. Event definitions are reconciled across layers before the loss, because afterward each market reads its own text and nobody has the power to harmonize them. And at notification, the aggregation position is chosen deliberately, with the arithmetic run both ways, in the knowledge that a notification commits: a classification can be argued at length, but one very rarely walks back the division one announced oneself.