OSFI amended Section 4.3 of its Minimum Capital Test guideline on July 20, 2026, to recognize natural catastrophe bonds as reinsurance for insurance-risk capital credit, effective immediately (OSFI, July 2026). Qualifying bonds need an indemnity trigger, collateral in high-quality assets held in Canada, full funding under a reinsurance security agreement, and prior approval. OSFI classifies them as unregistered reinsurance with no margin requirement, and that last clause is where the economics sit.

Key Takeaways

  • No margin requirement is the load-bearing clause. Unregistered reinsurance normally needs roughly 120 cents of security per dollar of ceded exposure before the MCT ratio reflects the whole cession.
  • 81% of second-quarter 2026 issuance carried indemnity triggers, $9.2 billion across 55 of 80 tranches, so the indemnity condition ratifies the dominant structure and refuses credit to the rest.
  • C$1.25 of required capital falls per dollar of qualifying earthquake protection, roughly C$1.88 of held capital at the 150% supervisory target. C$100 million of limit carries about C$188 million of relief.
  • 2.9% and 6.75% are the spreads TD paid on its two MMIFS Re series, against expected losses of 0.42% and 1.96%. Both were placed with no MCT credit attached.
  • C$8.5 billion of insured catastrophe losses in 2024, against a C$701 million annual average from 2001 through 2010. Sponsor demand ran two years ahead of the regulatory reward.

What Changes Inside Section 4.3

Section 4.3 is where the MCT decides whether ceded risk actually leaves the capital base. Registered reinsurers pass credit through subject only to counterparty credit factors. Every other counterparty is unregistered, and the guideline deducts recoverables from unregistered reinsurers out of capital available unless the cedent holds acceptable security: non-owned deposits under a reinsurance security agreement, funds held, or letters of credit capped at 30% of reinsurance contract assets.

Dollar-for-dollar coverage alone does not finish the job. The guideline layers a margin on top equal to 20% of the unexpired-coverage premiums and incurred-claims recoverables due from the unregistered counterparty, and that margin falls to zero only where premiums payable and acceptable collateral exceed the exposure itself. A cedent taking full credit for a collateralized cover from an unregistered Bermuda reinsurer therefore needs roughly 120 cents of security per dollar of ceded exposure.

A catastrophe bond cannot post 120 cents, which is why the no-margin classification matters more than the eligibility list. Investors fund the collateral trust at exactly the principal they put at risk: the limit is the trust and the trust is the limit. The 20% margin exists to buffer the gap between what a troubled reinsurer owes and what it happens to have pledged, and a fully funded trust has no such gap. Waiving it acknowledges that a limit-equals-collateral structure has nothing left to secure.

Scope is otherwise narrow. Every federally regulated P&C insurer except mortgage insurers can apply, and eligible perils are natural catastrophes only: earthquake, flood and wildfire qualify, while terrorism, industrial accidents and cyber stay outside. The boundary has bite, given that the wider market priced its first US terrorism catastrophe bond only weeks earlier (Insurance Business, July 2026).

The Capital Term in the Sponsor's Cost Comparison

Credit already existed in one corner of the MCT. The guideline computes earthquake reserves as the earthquake premium reserve plus the earthquake reserve component, all multiplied by 1.25, where the component equals earthquake risk exposure less financial resources, floored at zero. Eligible resources already included reinsurance coverage and capital market financing. What a bond lacked was standing as reinsurance for the rest of the insurance-risk calculation and for every peril outside earthquake.

The earthquake line shows the leverage. A dollar of qualifying protection that closes the earthquake reserve component gap removes C$1.25 of required capital; at the industry-wide supervisory target of 150%, that is roughly C$1.88 of actual capital held. In round numbers, C$100 million of limit that closes an equivalent component gap carries about C$188 million of held-capital relief, and at a 10% hurdle rate the capital term alone is worth nearly C$19 million a year.

Set that against the only Canadian price history anyone has. TD Insurance placed MMIFS Re Ltd. Series 2025-1 in January 2025, C$150 million of per-occurrence earthquake and severe convective storm cover attaching at C$2.35 billion, at a 2.9% spread on a 0.42% expected loss. The January 2026 return trip bought C$115 million of annual aggregate cover across five perils, attaching at C$350 million with exhaustion at C$500 million, settling at 6.75% on a 1.96% expected loss after guidance opened at 5% to 5.5%.

SeriesSizePerilsStructureSpreadInitial expected loss
2025-1 (January 2025)C$150mEarthquake, severe convective stormPer occurrence, attaches at C$2.35bn, term to end 20272.9%0.42%
2026-1 (January 2026)C$115mNamed storm, earthquake, SCS, winter storm, wildfireAnnual aggregate 2026-2028, C$350m attachment, C$500m exhaustion, C$25m event deductible6.75%1.96%

The remote earthquake layer cleared at roughly 6.9 times expected loss, in line with low-probability diversifying perils. The working-layer aggregate cleared at about 3.4 times, and only after investors pushed the spread roughly 29% above the original midpoint, in a year when spread compression has been the story everywhere else. TD paid both spreads with no MCT relief attached, and the notice reprices the next series before a dollar of spread moves.

The demand behind it was already built. CatIQ put 2024 insured catastrophe losses at C$8.5 billion, past the C$6.2 billion record set in 2016, with four billion-dollar events inside a 27-day window: C$3 billion for the August Calgary hailstorm, C$2.7 billion for the remnants of Hurricane Debby and C$1.1 billion for the Jasper wildfire.

Where the Conditions Cost Something

The indemnity condition tracks where issuance already sits, but it is a basis-risk decision with a price. Index and parametric recoveries settle on modelled or industry-wide quantities, so the cedent keeps the residual between what the instrument pays and what its gross loss turns out to be, and a formula-driven capital test has no cell for that residual. Grant an index bond full credit and the MCT would overstate protection in exactly the tail scenarios it exists to capitalize.

The cost lands on the investor side of the trade and comes back as spread. Buyers of an indemnity deal absorb the sponsor's underwriting quality, claims practices and loss development, and they price the structural controls that manage it: TD's 2026 aggregate carries a C$25 million event deductible and a C$175 million per-event cap. Sponsors also pay in disclosure and time, because an indemnity bond needs offering-document detail on the subject business and a loss-development mechanism running through an extension period.

A Canadian carrier wanting capital credit for index-based protection stays outside the tent, and index triggers took 15% of second-quarter issuance while parametric deals took $360 million (Artemis, July 2026).

Situs is the second cost. Nothing forces the issuing vehicle onshore, so the Bermuda incorporation, the 144A distribution and the existing ILS investor base survive; the trust assets move to Canadian custody where the cedent's security interest can be perfected.

Both MMIFS Re series parked proceeds in European Bank for Reconstruction and Development notes paying roughly CORRA less a small margin. Whether supranational paper can be held through a Canadian account that satisfies OSFI, or whether qualifying collateral drifts toward Government of Canada treasury assets, is an open structuring question with a basis-point answer.

The collateral also stays inside the capital net. Qualifying assets remain subject to the MCT's capital requirements for unregistered reinsurance collateral, so the trust portfolio attracts asset-risk factors by type and credit quality like any other security the cedent looks through to. That is the argument for boring collateral: a trust of Government of Canada bills minimizes the look-through charge along with the legal work, while supranational notes buy a few basis points of yield at the price of a longer approval file.

Against a waived 20% margin the budget for those basis points is generous, which is precisely why the first sponsor to price a bond with the credit attached will reveal how much of the relief the market keeps rather than pays away.

Further Reading

Sources

  1. OSFI: Regulatory notice on using natural catastrophe bonds in the Minimum Capital Test guideline (July 2026)
  2. OSFI: Minimum Capital Test Guideline (2026)
  3. Artemis: Canada's regulator adds catastrophe bonds as a form of reinsurance for capital credit (July 2026)
  4. Artemis Deal Directory: MMIFS Re Ltd. (Series 2025-1)
  5. Artemis Deal Directory: MMIFS Re Ltd. (Series 2026-1)
  6. Artemis: Q2 2026 Catastrophe Bond & ILS Market Report (July 2026)
  7. Artemis: MMIFS Re 2026-1 cat bond to help manage rising multi-peril costs in Canada, TD Insurance CEO (January 2026)
  8. CatIQ: Canadian Insured Losses from Catastrophic Events Total CAN $8.5 Billion in 2024 (January 2025)
  9. Insurance Business Canada: OSFI clears natural catastrophe bonds as reinsurance for capital relief (July 2026)
  10. Canadian Underwriter (Insurance Institute): 2024 catastrophic insured losses smash records (January 2025)