Skip to content
Hamilton, the capital of Bermuda, seen from Hamilton Harbour, with waterfront office buildings and moored boats

VII Reinsurance & Specialty Risk·Analysis·Americas

Retro refuses to soften as quake tail risk drifts to state pools

On 18 November a primary insurer's cat bond priced at the bottom of its guidance, while reinsurers' retro bonds held in the upper half. State pools are lining up to absorb the peak earthquake tail that retro does not reach, but all three lack legislation.

Talay Insight Editorial · 19 November 2026 · 8 min read · 10 sources

Talay Insight Analysis DeskAI-assisted analysis, reviewed by an editorPublisher of record: Lu Apps LLCTrack recordRequest a correction

Hamilton, Bermuda, a hub for catastrophe bond and reinsurance vehicles, 27 June 2015 — archive photo, illustrativePhoto: Captain-tucker / Wikimedia Commons · CC BY-SA 4.0 · resized · Source

Why it matters

Noise: cat bond issuance has passed $20.37bn and the market has topped $66bn, and the volume is being read as proof of softening. Signal: on 18 November pricing split in two. Fish Pond Re closed at 2%, the bottom of lowered guidance. Retro, the reinsurer's own protection, did not get cheaper. Kanata Re's guidance tightened upwards to 4.75–5%, and S6 Re is seeking five-year retro. Where retro ends, in peak quake risk, Canada, India and four regional pools are building state vehicles. What all three lack is not capital but law.

Implications

  • According to Artemis, Fish Pond Re upsized to $400m on 18 November and its Class A notes priced at 2%, the bottom of lowered guidance. Retro seller Kanata Re saw its guidance tighten from 4.5–5% to 4.75–5%.
  • Lancashire's debut cat bond, S6 Re, aims to lock in $175m of retro for five years, from January 2027 to December 2031. Guidance is 4.75–5.5% against an expected loss of 2.41%.
  • Canada's C$50bn earthquake facility awaits legislation, and India wants an exemption in its 1938 insurance law to allow cat bonds. No amount has been disclosed for the four regional pools' Global Reinsurance Facility.

Noise

Cat bond issuance is running at a record pace, so every layer of reinsurance is getting cheaper

Signal

On the same day, primary bond guidance moved down while retro bond guidance moved up

Signal vs Noise ›

Map: Retro refuses to soften as quake tail risk drifts to state pools

The noise is in volume, the signal in where guidance moves

On 17 November the cat bond market took year-to-date issuance to $20.37bn. Artemis says the threshold was crossed a week earlier than in 2025. Risk capital in the market topped $66bn for the first time and has grown roughly 8% since January. Headlines read this volume as proof that pricing is softening.

Volume is not news. The $3.286bn pipeline and 2025's record $25.6bn were already known. The news came on 18 November, in the direction of price guidance. The primary insurer's bond cut its guidance and priced at the bottom. Two retro bonds, which sell reinsurers their own protection, held in the upper half of their ranges. Retro is the cover a reinsurer buys to pass its own risk on to another pool of capital.

Talay's reading is this. Softening is reaching the primary layer, not the reinsurer's own tail protection. Governments, meanwhile, are preparing to take on the peak earthquake risk that retro does not cover. Canada, India and four regional pools all moved in the same week. What all three lack is not capital but legislation and design.

1 · Two prices on the same day

Texas Farm Bureau's Fish Pond Re bond launched with a $250m target. Artemis reports it grew 60% to $400m. Initial guidance for the Class A notes was 2.25–2.75%; they priced at 2%. Class B opened at 3–3.5% and priced at 2.75%. Both classes landed at the bottom of a range that had already been lowered.

MS Amlin's Kanata Re bond sells retro protection. It grew 50% to $150m, but the price did not fall. Initial guidance was 4.5–5%; sources say the range tightened to 4.75–5%. The lower end moved up by 0.25 points. The bond covers US windstorm, US and Canadian earthquake and European windstorm over four years.

On the same day Lancashire brought its first cat bond, S6 Re 2027-1, to market. It seeks $175m of aggregate retro covering five years, from January 2027 to December 2031. Guidance is 4.75–5.5% and the expected loss is 2.41%. Artemis reports these terms on the basis of sources; the issuer has not confirmed them.

2 · Watch the direction of guidance, not the multiple

There is an easy trap in this comparison. On Fish Pond's Class A notes the spread is about 7.7 times the expected loss. At Kanata the ratio is about 1.9, at S6 Re 2.0–2.3. That gap alone does not show retro is cheap. A remote layer with an expected loss of 0.26% commands a high multiple in any market.

The comparable measure is how guidance moves. The primary sponsor's guidance was cut by 0.25–0.5 points. The retro bond's guidance tightened upwards from its lower end by 0.25 points. Above expected loss, investors are demanding a risk margin of 2.2–2.5 points at Kanata and 2.3–3.1 points at S6 Re. On Fish Pond A that margin is 1.74 points.

Tenor also tells a story. Kanata covers four calendar years, S6 Re five. The reinsurer is fixing the cost of its own protection at today's price out to 2030–2031. A buyer expecting prices to fall quickly would renew annually. Talay reads this choice as a sign that expectations of retro price declines are weak.

3 · Where retro ends, the state pool begins

Retro bonds work at high industry-loss thresholds. For an event to count under S6 Re, industry losses must exceed $10bn. Florida windstorm and California earthquake carry most of the expected loss. In Canada the problem is not the price of capacity but the size of the loss relative to industry capital.

The Insurance Bureau of Canada (IBC) estimates insured losses from a severe earthquake at C$52.6bn. The industry's capital across all lines is about C$50bn. In its 5 November budget the government announced an earthquake facility providing up to C$50bn of emergency funding. On 18 November Morningstar DBRS wrote that the federal mechanism has still not been set up. The agency warns that an extreme quake could bankrupt several insurers. In Quebec only 4–7% of households carry earthquake cover.

In India the constraint is the Insurance Act of 1938. On 18 November an IFSCA working group proposed a targeted exemption in the law. It would allow special-purpose insurers that fully collateralise their risk to be set up in GIFT City. The plan's first product is the cat bond. Gross premiums written by insurance offices in GIFT City were $162m in fiscal 2025 and $649m in 2026.

The third step is global. On 16 November at COP31, four disaster risk pools covering Africa, the Caribbean, the Pacific and Southeast Asia advanced a joint proposal for a Global Reinsurance Facility. The pools began examining the idea in March 2026. Germany's development ministry pledged additional funding. No amount, structure or timetable has been disclosed.

4 · The constraint is the channel, not the capital

The common thread across the three initiatives is clear. The capital markets hold $66bn of risk capital, up about 8% in 2026. That money goes into risk that can be modelled, collateralised and placed on a legal footing. Canada's facility lacks its statute, India's cat bond lacks its legal vehicle and the global pool lacks a number. Until the channel exists, abundant capital cannot reach this tail.

Sticky retro pricing reinforces the picture. Reinsurers are locking in their top-layer protection at around twice expected loss, for five years. They will struggle to sell the same tail to primary insurers below that cost. Talay's reading: in peak earthquake risk, price will be set by retro costs and governments' legislative calendars, not by primary-market softening.

5 · What would prove this wrong

A single data point would refute this reading. If S6 Re upsizes beyond $175m and prices below 4.75%, softening has reached retro too. The 18 November divergence would then be no more than a one-to-two-week lag. The scenario table gives this path a 30% probability.

A second refutation could come from Canada. If the facility legislation passes Parliament before 1 January 2027, the quake tail exits reinsurance negotiations early. The 5 November announcement carried no bill number; Talay sees this path as low probability.

Probabilities

Scenarios

ScenarioProbabilityTriggerMarket impact
H1Retro holds firm, pools move slowly50%S6 Re prices within its 4.75–5.5% guidance; Canada's facility bill slips to 2027 and India's exemption awaits a legislative step.At 1 January primary catastrophe pricing falls faster than retro; the peak quake tail stays in reinsurance costs and on insurers' balance sheets.
H2Softening reaches retro30%S6 Re upsizes beyond $175m and prices below 4.75%; new retro bonds come to market in December.The divergence closes, reinsurers' own protection costs fall and discounts pass through to top layers.
H3Retro hardens again20%S6 Re downsizes or prices above guidance; a North American catastrophe loss above $10bn touches the retro layer.The top layer becomes more expensive and state facilities gain urgency; Canada and India bring their legislative timetables forward.

Module A

Constraints Matrix

STRUCTURAL AVG 3.5 · TACTICAL AVG 2.5Structural constraints dominate: the outcome is set more by these limits than by the actors' preferences.

Hard structural constraintspersistent · beyond the actors' will

  • Thin retro supply · Lloyd's of London

    4/5

    The two retro bonds of 18 November total $325m, about 0.5% of a $66bn market. Kanata upsized but its guidance did not come down.

  • No law in Canada

    4/5

    The C$50bn facility was announced in the 5 November budget, but no bill is before Parliament yet. A C$52.6bn scenario loss would fall on industry capital of about C$50bn.

  • 1938 Insurance Act · India

    3/5

    India needs a statutory exemption for fully collateralised special-purpose insurers; the IFSCA proposal has awaited central government action since February 2026.

  • Industry index threshold

    3/5

    Under S6 Re an event counts only if industry losses exceed $10bn; the index trigger does not match the reinsurer's own losses one for one.

Tactical frictiontemporary · eases over time

  • Renewal calendar weeks

    3/5

    The 1 January 2027 renewal is about six weeks away; as of 5 November Canada's facility did not even have a bill number to meet that date.

  • Low take-up months

    3/5

    Earthquake cover take-up is 4–7% of households in Quebec and 50–65% in British Columbia; the facility protects insurers, not uninsured households.

  • Unverified terms days

    2/5

    S6 Re's terms are preliminary details Artemis attributed to sources on 18 November; the issuer has not confirmed them and final price and size are undisclosed.

  • Pool of unknown size months

    2/5

    The four regional pools' Global Reinsurance Facility advanced on 16 November; Germany's pledge, the facility's capital and its timetable are undisclosed.

Module B

Signal vs Noise

SIGNAL 57% · NOISE 43%

Module C

Asset-Class and Positioning Implications

Asset classExposureTransmission channelH1H2H3ExpectedConvictionHorizonWhat to watch
CreditRetro cat bond spreadsWhile retro supply stays scarce, spreads tighten more slowly than on primary bonds+−−+++0.30●●●0–3 monthsWhere S6 Re's final price lands within its 4.75–5.5% guidance
CreditPrimary cat bond spreadsAbundant capital is pushing primary sponsors' guidance lower−−+−0.60●●●0–3 monthsFinal pricing of December issues relative to initial guidance
EquitiesGlobal reinsurance sectorPrimary pricing falls while retro costs do not; net margins are squeezed−+−−0.40●●●3–12 monthsThe gap between retro and primary catastrophe pricing in January 2027 broker reports
EquitiesCanadian property insurersThe quake tail stays on balance sheets and in reinsurance costs until the law passes−+−−−0.60●●●3–12 monthsParliamentary timetable for Canada's earthquake facility bill
Sovereign debtCanadian federal budgetA C$50bn contingent earthquake liability enters the public balance sheet by statute00−−0.20●●●12+ monthsThe facility's funding structure and the contingent liability line in budget documents

How to read: ++ strong structural support · + support · 0 neutral · − pressure · −− strong pressure. “Expected” is the direction weighted by scenario probabilities. H1: Retro holds firm, pools move slowly · H2: Softening reaches retro · H3: Retro hardens again.

General, scenario-conditional analysis at asset-class level. It contains no specific security, price target or trade timing and is not personalised investment advice (Turkish Capital Markets Law No. 6362).

Second-order effects

And then what?

Starting point

On 18 November the primary sponsor's Fish Pond Re bond priced at 2%, the bottom of lowered guidance. Retro seller Kanata Re tightened its guidance to 4.75–5% and stayed in the upper half; S6 Re is seeking five-year retro.

  1. 1

    Retro costswithin weeks

    Reinsurers fix their retro protection for four to five years, locking in 2027–2031 tail costs at today's prices. At 1 January retro pricing falls less than primary catastrophe pricing.

    Watch: S6 Re's final pricing and size; the retro price line in January 2027 broker renewal reports

  2. 2

    Reinsurance pricingwithin weeks

    With retro costs holding, reinsurers limit discounts on top-layer earthquake cover. In Canada, where the scenario loss is C$52.6bn against industry capital of about C$50bn, insurers keep protecting this tail through reinsurance at that cost.

    Watch: Pricing of Canadian property insurers' 1 January 2027 earthquake programmes and DBRS's next sector note

  3. 3

    Public budgets and legislationwithin months

    While the peak tail stays expensive, governments are pushed to build vehicles faster. Canada's facility, India's special-purpose insurer exemption and the four pools' joint facility would open a legal channel to $66bn of cat bond capital.

    Watch: Introduction of Canada's earthquake facility bill in Parliament; the 1938 Act exemption; the size of the Global Reinsurance Facility

What breaks the chain

If S6 Re upsizes beyond $175m and prices below 4.75%, softening reaches retro and the chain stops at the first step. If Canada's legislation passes before 1 January 2027, the second step weakens.

Triggers

Thresholds to watch

IndicatorThresholdTodayWhat it means
US 10-year yield< 4.50%5.10If the risk-free yield drops below this level, the return on cat bond collateral falls; more of the coupon has to come from the risk margin and scarce retro supply becomes even more expensive

Sources

  1. Artemis — Cat bond issuance hits $20bn at fastest pace in 2026, market size exceeds $66bn for first time (17 November 2026)
  2. Artemis — Texas Farm Bureau secures $400m Fish Pond Re 2026-1 cat bond at reduced pricing (18 November 2026)
  3. Artemis — MS Amlin syndicate increases Kanata Re 2026-1 cat bond target 50% to $150m (18 November 2026)
  4. Artemis — Lancashire enters cat bond market seeking $175m S6 Re aggregate peak peril retro (18 November 2026)
  5. Reinsurance News — Morningstar DBRS calls for federal insurance backstop to strengthen Canada's quake resilience (18 November 2026)
  6. Insurance Business — Canada creates earthquake insurance backstop as Carney budget addresses decade-long gap (6 November 2026)
  7. Newswire.ca — Insurance Bureau of Canada statement on Canada's 2026 federal budget (5 November 2026)
  8. Artemis — IFSCA working group finalises ILS and SPI regulatory recommendations for India (18 November 2026)
  9. Setindiabiz — GIFT City Draws Global Reinsurers as 15 Eye Financial Year 2027 Entry (24 September 2026)
  10. Reinsurance News — Global Reinsurance Facility proposal moves forward with German backing at COP31 (16 November 2026)
Share

Sourcing and verification rules: methodology · Report an error: contact

Related reports