Attachment and Exhaustion Points
Definition
Attachment point is the loss level where an insurance-linked security starts taking losses. Exhaustion point is the level where the investor's principal is fully wiped out.
Why it matters
These two points define the insurance-loss layer an investor owns, the amount of protection supplied to the sponsor, and the events capable of impairing principal. A higher coupon is not meaningful without knowing how close attachment sits to modeled losses, how wide the layer is, whether losses accumulate across events, and which trigger converts an event into investor loss. Comparing attachment and exhaustion alongside expected loss, tail probability, peril concentration, and collateral terms is fundamental to pricing catastrophe bonds and collateralized reinsurance.
Common misconceptions
- •A high attachment point is not automatically safe because the relevant question is its distance from the modeled loss distribution for the covered peril, territory, and risk period.
- •Exhaustion does not describe the most likely loss; it identifies the contractual level at which the covered layer has been completely consumed.
- •Two securities with identical attachment and exhaustion amounts can have different risk when their trigger, exposure base, hours clause, aggregate treatment, reinstatements, or sponsor portfolio differs.
- •A layer does not always reset after one event. Aggregate structures can accumulate qualifying losses, while occurrence structures and reinstatement provisions apply different mechanics.
Technical details
Simple example
Attachment is where losses begin for the tranche.
Exhaustion is where the tranche is fully consumed.
If a cat bond attaches at $1 billion of covered hurricane losses and exhausts at $1.3 billion, investors absorb losses in that $300 million band.
What to compare
Compare attachment probability, expected loss, modeled tail loss, peril zone, sponsor, collateral terms, and whether the trigger is indemnity, industry loss, modeled loss, or parametric.
Layer loss calculation
For a simple linear layer, investor loss equals qualifying loss above attachment, capped at the layer width. The loss percentage is that amount divided by exhaustion minus attachment. Contract terms may instead apply franchise thresholds, binary principal reductions, step functions, event caps, or modeled-loss calculations.
Example: a $300 million layer attaches at $1.0 billion and exhausts at $1.3 billion. A qualifying loss of $1.12 billion penetrates the layer by $120 million, producing a 40% layer loss. A $1.4 billion qualifying loss exhausts the layer, so principal loss is capped at 100% rather than continuing above the layer.
Calculate with the contract's loss measure, not the economic damage reported in the news. Industry loss estimates, sponsor indemnity losses, modeled portfolio losses, and physical parameters can produce different results from the same catastrophe.
Occurrence, aggregate, and reinstatement structures
Occurrence layers respond to a qualifying event, subject to definitions such as peril, territory, and hours clause. Aggregate layers track covered losses across the risk period after deductibles or franchise amounts. A series of moderate events can therefore impair an aggregate layer even when no single event reaches an occurrence attachment point.
Reinstatement provisions can restore exhausted or partially used protection in exchange for additional premium, while annual aggregate limits cap total coverage. Investors should map how prior events reduce remaining limit and whether new premium compensates for renewed exposure.
Review event attribution carefully. Hurricanes, floods, wildfires, and severe convective storms can produce disputes about when one occurrence ends, whether secondary perils are covered, and which losses fall inside the specified window.
Modeling and relative-value analysis
Compare attachment probability, exhaustion probability, expected annual loss, occurrence exceedance probability, aggregate exceedance probability, modeled return period, and sensitivity to alternate catastrophe models. Attachment in nominal dollars is meaningless without the underlying exposure and loss distribution.
Spread should compensate for expected loss, model uncertainty, fees, illiquidity, and tail concentration. A bond paying 9% with 5% expected loss offers a different margin than one paying 8% with 1.5% expected loss, even before considering correlation with the rest of the portfolio.
Portfolio analysis should aggregate positions by peril, region, season, sponsor, trigger type, and event—not merely security count. Several layers can all attach after the same Florida hurricane or Japanese earthquake and produce correlated losses.
Post-event monitoring and collateral release
After a potential event, monitor preliminary notices, sponsor reserves, catastrophe-model updates, reporting-agency estimates, calculation-agent determinations, and changes in trapped collateral. Initial loss estimates can develop substantially as claims, inflation, litigation, and business interruption become clearer.
Confirm when collateral can be released, how disputed losses are reserved, whether the risk period can extend, and how investment income and expenses are allocated during an extension. Legal maturity is not always the date investors receive all principal back.
Useful reporting bridges opening collateral to premium, investment income, fees, paid loss, reserves, principal reduction, trapped amount, and distributable proceeds. Vague statements that a layer remains under review are insufficient without quantified attachment headroom.