excess of loss reinsurance
/ XL = ex-ell /
Instead of sharing a fixed slice of everything, what if the reinsurer only stepped in when a loss got truly big? That is the idea behind excess of loss reinsurance. The cedant pays all of every loss up to a chosen level it can comfortably absorb, and the reinsurer pays only the part above that level, up to an agreed ceiling. It is like a high-deductible policy for the insurer itself: small and medium losses are the cedant's problem, but the rare monster is the reinsurer's.
The level at which the reinsurer starts to pay is the attachment point (or retention); the most it will pay is the limit; together they define a layer, written as 'limit excess of attachment', for example '4 million excess of 1 million'. On a 600,000 loss the reinsurer pays nothing — it is below the 1 million attachment. On a 3 million loss the reinsurer pays 2 million (the part from 1 million up to 3 million), while the cedant keeps its 1 million. On a 7 million loss the reinsurer pays its full 4 million limit and the cedant is back on the hook for the 2 million above the layer. Excess of loss comes in flavours: per-risk XL responds to one large individual policy loss, while per-occurrence (or per-event) XL responds to the total of all losses from a single event, such as one storm hitting many homes.
Excess of loss is non-proportional: premium is not a fixed share of original premium but a separately calculated price for the layer, and there is no ceding commission. It is the natural tool when the worry is severity — one huge claim or one catastrophic event — rather than the routine churn of small claims, so it targets protection precisely where capital is most at risk. The pricing is genuinely hard: large losses are rare, so there is thin data in the very region the cover protects, and reinsurers lean on exposure models and judgement. A common confusion: per-risk and per-occurrence layers are different covers; a per-risk treaty will not pay just because a storm produced a large aggregate if no single risk pierced the attachment.
A per-risk layer of 4 million excess of 1 million. Loss of 800,000: reinsurer pays 0. Loss of 3 million: reinsurer pays 2 million, cedant keeps 1 million. Loss of 6 million: reinsurer pays its full 4 million, cedant keeps 1 million below and 1 million above the layer.
The reinsurer pays only the slice of each loss inside the layer: limit excess of attachment.
Per-risk and per-occurrence XL are not interchangeable. A per-risk cover ignores how many policies a single event hit; a per-occurrence cover sums them. Buying the wrong one leaves a gap exactly where it hurts.