Basis risk in the catastrophe bond market has changed address. Artemis data through August put indemnity triggers on almost 78% of 2026 issuance, a record, while aggregate structures fell to 36.1% of outstanding limit (Artemis, September 2026). The first figure closes the gap between an index and a sponsor's loss. The second reopens a different gap, between a single event above attachment and a year of storms below it.
The September 10 analysis draws on a Deal Directory covering roughly $224 billion of issuance, and it sets the two series side by side without joining them. Joined, they describe a sponsor who has traded one mismatch for another: the per-occurrence indemnity note that now dominates the market responds to the cedant's own incurred loss, but only once, and only when a single designated event clears the attachment.
Key Takeaways
- 77.8% of 2026 cat bond limit issued through August carries an indemnity trigger, up from 67.5% in 2022, 72.5% in 2023, 73% in 2024 and 75.6% in 2025 (Artemis, September 2026).
- 36.1% of outstanding risk capital is aggregate, against 58% in March 2019 and 39.4% a year ago, so per-occurrence structures now hold 63.9% of an outstanding market last counted at $65.6 billion.
- 81% of second-quarter issuance, $9.2 billion of $11.3 billion across 55 of 80 tranches, was indemnity-triggered, which means issuance since June has leaned less on indemnity than the record quarter did.
- Over $40 billion a year of US severe thunderstorm insured loss in 2021 to 2025, against about $4 billion in 1998 to 2005, is the frequency pattern a per-occurrence note leaves on the sponsor's net (Verisk PCS, July 2026).
- 6.63% average spread in Q2 2026, down from 7.12% the prior quarter and 7.23% a year earlier, shows investors taking more indemnity and more per-occurrence exposure at tighter pricing.
Two Series, Two Denominators
The 78% and the 36.1% measure different things. The indemnity share is a flow figure: the proportion of limit issued in a calendar year whose payout is calculated from the sponsor's own ultimate net loss. Artemis reports it at 67.5% for 2022, 72.5% for 2023, 73% for 2024, 75.6% for 2025 and 77.8% for 2026 through August, and calls the current reading "the highest figure for any year in the market's history" (Artemis, September 2026).
The aggregate share is a stock figure: the proportion of all outstanding risk capital whose limit erodes across multiple qualifying events within a risk period. It stood at 58% in March 2019, dipped below 50% for the first time in November 2021, and has fallen through 47.4% at mid-2023, 41.7% at the end of 2023 and 39.4% a year ago to 36.1% now. Per-occurrence structures therefore hold 63.9% of the $65.6 billion outstanding market the Q2 2026 report counted at the end of June.
Against the second quarter, the year-to-date indemnity figure also needs reconciling. In Q2 2026 indemnity took 81%, or $9.2 billion of $11.3 billion, across 55 of 80 tranches; industry-loss index triggers took $1.7 billion, or 15%, across 17 tranches, and three parametric tranches placed $360 million (Artemis, July 2026). A year-to-date share of 77.8% after a quarter at 81% means the rest of the year's issuance, the first quarter plus July and August, ran several points below the record quarter, so the trend toward indemnity is real but the second quarter overstated its pace.
| Artemis series | Denominator | Earliest reading | Latest reading |
|---|---|---|---|
| Indemnity trigger share | Limit issued in the calendar year | 67.5% (2022) | 77.8% (2026 through August) |
| Indemnity trigger share, quarterly | Limit issued in the quarter | 81% (Q2 2026, $9.2bn of $11.3bn) | |
| Aggregate structure share | Outstanding risk capital | 58% (March 2019) | 36.1% (September 2026) |
| Per-occurrence structure share | Outstanding risk capital | About 42% (March 2019) | 63.9% (September 2026) |
Artemis's own explanation for the trigger trend is the sponsor base. It counts 14 first-time sponsors in 2026 against 15 in all of 2025, most of them primary insurers, and observes that such insurers "prefer an indemnity structure, to more closely align cat bond cover with their traditional reinsurance arrangements" (Artemis, September 2026). This site's first-half review read the 81% second-quarter share as loss-estimate risk migrating onto debut sponsors' own reserving; the September data extend that migration by another year of issuance.
Where the Residual Mismatch Now Sits
An indemnity, per-occurrence note is a collateralized excess-of-loss reinsurance layer with the index basis risk removed. What it leaves in place is the event definition. An occurrence layer in a traditional program also responds to one event, and the cedant carries everything below it. The difference in 2026 is that more of the market's cover is built that way, so the frequency pattern below occurrence attachments falls on more sponsors' nets than at any point since Artemis began the aggregate series.
Verisk's own essay inside the Artemis report supplies the size of that pattern. Harry White, Head of Commercial Strategy at PCS, writes that "average annual ST insured losses in the United States" rose "from approximately $4 billion (1998-2005) to over $40 billion (2021-2025)" (Verisk PCS, July 2026). Annual claims grew from 1.3 million to 3.5 million as more storms cleared the $25 million threshold for a PCS catastrophe designation; Texas alone went from six designated events a year to 32, and its share of national severe thunderstorm loss rose from 13% to 23%.
A $40 billion annual mean spread across dozens of designated events is a peril of many losses in the $100 million to $2 billion range, and a per-occurrence note attaching at a one-in-50 or one-in-100 modeled return period almost never sees one of them. Verisk's 2026 modeled benchmark puts severe thunderstorm at 40% of the global insured average annual loss, more than any other peril, and the whole of that share sits in the working layers. A sponsor who replaces annual aggregate limit with occurrence limit therefore keeps roughly the same expected loss on the frequency peril while its modeled tail protection improves, a trade that is invisible in a one-in-100 PML and fully visible in a five-year average of net catastrophe loss.
Investors are pricing the other side of the trade cheaply. The Q2 report puts the average spread at 6.63%, down from 7.12% in Q1 and 7.23% in Q2 2025, with average multiples between 2.33 and 2.57 in every spread band except the highest, where the multiple fell to 1.97, and an average spread above expected loss of 3.74%, the lowest since Q1 2023 (Artemis, July 2026). Deals priced 10.2% below the midpoint of initial guidance on average. Capital that settles on the sponsor's own loss adjustment, in the structure least exposed to frequency, has never been cheaper.
Aggregate Comes Back Inside the Occurrence Note
The complication sits inside Artemis's own publications. The Verisk essay opens with the observation that "as the market softens and aggregate and multi-peril features return to many catastrophe bond structures, losses from severe thunderstorms (ST) could once again have a significant role to play" (Verisk PCS, July 2026). Two months later the September data show the aggregate share of outstanding limit still falling, to 36.1%. Both statements hold together only if the aggregate features arriving in 2026 take a different form from the annual aggregate limits of 2019.
Artemis's June 2023 analysis of the same series attributed the flight from aggregate to investor losses "due to aggregation across smaller catastrophe and severe weather events from 2017 on," and to structures with "lower deductibles and retentions, as well as more expansive peril classifications" (Artemis, June 2023). The market's response was to move from franchise deductibles to event deductibles and to raise aggregate attachments, and that is the form in which aggregate is returning: multi-peril indemnity notes with per-event deductibles, attaching high enough that a $40 billion frequency year does not reach them.
Traditional reinsurance has made the same move. Gallagher Re's Josh Knapp conditioned reinsurer appetite on an attachment point that is "practical for everyone," and the broker warned that a well-structured aggregate may require cedants to "carry a greater share of losses within their net retentions" (Gallagher Re, September 2025). This site's coverage of the July 2026 renewal found aggregate and multi-year covers priced at usable attachments for the first time in the cycle, and its tracking of 2026 severe convective storm losses above $35 billion through mid-August is the first test of those attachments.
That leaves the sponsor's net where the two Artemis series put it. The 78% indemnity share means the bond pays on the cedant's own number, and the 63.9% occurrence share means most of the market pays on that number once, above an attachment a year of severe thunderstorm claims never reaches. The event deductible that repaired investor confidence in aggregate is a per-event retention by another name, and every dollar of it accumulates on the ceding company's net across the 32 Texas designations a year that Verisk now counts. The market has priced the index out of the trade and priced the calendar in.
Further Reading
- Cat Bonds Hit $18B in H1 2026: What the Records Actually Mean – the first-half review this piece follows, which read the 81% second-quarter indemnity share as loss-estimate risk moving onto debut sponsors.
- Mexico's M7.3 Quake Missed Its Cat Bond Trigger by One Tenth of a Magnitude – the trigger-definition basis risk that indemnity structures remove, priced cell by cell.
- Parametric Reinsurance for Secondary Perils: Basis Risk, RBC Credit, and the Actuarial Certification Gap – why the frequency perils that sit below occurrence attachments are the hardest to cover with an index.
- US Severe Convective Storm Losses Top $35B as the August Derecho Outruns 2026 Cat Budgets – the 2026 version of the frequency pattern a per-occurrence book leaves on the cedant's net.
- OSFI Opens MCT Capital Credit to Indemnity Cat Bonds With Collateral Held in Canada – a regulator ratifying the indemnity trigger as the only structure that earns full reinsurance credit.
Sources
- Artemis: Catastrophe bond market shifts further towards indemnity triggers and per-occurrence coverage (September 10, 2026)
- Artemis: Q2 2026 Catastrophe Bond & ILS Market Report (July 2026), including the Verisk PCS essay "Exposure Concentration and Claim Severity" by Harry White
- Artemis: As severe thunderstorm plays larger role in ILS, exposure growth & claim severity trends are key: PCS' White (July 20, 2026)
- Artemis: Catastrophe bond market records that were broken in H1 2026 (July 6, 2026)
- Artemis: Visualising the catastrophe bond market shift from aggregate to occurrence (June 15, 2023)
- Reinsurance News: Catastrophe bond issuance exceeds $11.3bn in record second quarter (July 7, 2026)
- Verisk PCS: Consolidated catastrophe loss methodology paper ($25 million US designation threshold)
- Artemis: Property aggregate reinsurance re-emerges amid expanding market capacity: Gallagher Re (September 16, 2025)