Skip to content
←Back to glossaryClaims & settlement

Large loss threshold

Its 5 cross-references, woven across the glossary's 18 categories

The amount above which a claim is handled separately, in claims management, pricing and reserving, because its severity follows a different law than ordinary claims and mixing it with them distorts all statistics.

Definition

Not all claims are alike. Most are frequent and small; a few, rare, reach amounts a hundred or a thousand times larger. Insurers therefore set a threshold above which a claim is classed as large and handled under its own rules. In claims management, it is assigned to experienced handlers, monitored by management and reported to reinsurers within agreed deadlines. In pricing, large losses are capped at the threshold in the analysis of ordinary experience, then their cost is added back through a loading calculated on a broader base, often the whole market or a long period, because their occurrence in a given segment is largely random. In reserving, they are valued case by case, and their late development follows a different profile. Choosing the threshold involves a trade-off: too low, it removes a significant part of ordinary experience; too high, it lets exceptional losses distort tariffs. Extreme value theory methods help locate it, by identifying the point from which the distribution of amounts takes the shape of a heavy tail.

Example

Textbook case: a motor insurer sets its large loss threshold at 150,000 euros. These claims account for 0.3 percent of the number but 25 percent of the cost. In the tariff, each zone is analyzed on claims capped at 150,000 euros, then receives a uniform loading for large losses, so that a single bodily injury accident does not send a municipality's premium soaring.

Related terms
Also known as

large loss, large loss threshold, major loss

Stay informed

Follow the research

One letter a week. First, the week's underwriting note, when one is published: a dated, argued decision, signed by Paul Bernardski. Then what the observatory published on cyber risk, political risk and insurance. No account, no tracking.

No spam, one-click unsubscribe, no data ever sold.