Moody's RMS HD Models Redraw the Severe Convective Storm Cat Budget

Moody's launched its North America Severe Convective Storm HD Models on December 11, 2025, replacing tools the company describes as built on coarse hazard resolution, small event sets, simplified vulnerability curves, and constrained financial modules.

US insured SCS losses exceeded $45 billion in 2025 for a third consecutive year, with a single March outbreak running $8 to $10 billion across 26 states. Per-event costs sat 31 percent above the prior decade's average.

Key Takeaways

• Per-event SCS costs run 31 percent above the prior decade average, on losses exceeding $45 billion for a third consecutive year.

• The HD catalog holds more than 12 million events across 50,000 simulated years, with 16 distinct hail size categories separating cosmetic damage from structural failure.

• Vulnerability is calibrated on $55 billion of location- and policy-level claims, producing more than 2,700 damage curves that carry the full loss distribution rather than a mean damage ratio.

• The revision lands hardest in the 1-in-10 to 1-in-50 range, not the 1-in-250 or 1-in-500 tail, which is exactly where aggregate cover attachment points sit.

• SCS remains informational in the NAIC Rcat framework, while the wildfire charge was adopted at Spring 2026 and the July 9, 2026 meeting is expected to advance the SCS analysis.

What the Architecture Change Fixes

Three legacy limitations account for most of the difference, and each maps to a specific pricing artifact.

Hail, tornado, and straight-line wind hazards vary over tens to hundreds of meters. A model running at one-kilometer or coarser resolution cannot represent the block-by-block gradient that decides whether a property takes cosmetic hail damage or total roof failure. The HD hazard layer simulates atmospheric dynamics rather than resampling history, at grid resolutions of hundreds of meters, across more than 12 million events over 50,000 simulated years with 16 hail size categories.

The event set problem is separate. The meaningful US record runs roughly 70 years, so a generator calibrated on the 1950-2010 period underweights derechos, which can produce losses comparable to small hurricanes on entirely different tracks, and underrepresents multi-day outbreak clusters whose damage accumulates across hours-clause boundaries. Derechos now get explicit simulation treatment.

Vulnerability is the change with the most direct pricing consequence. Legacy SCS models estimate a mean damage ratio at a given intensity and discard the variance, but variance at attachment is what shapes the loss distribution where a per-occurrence layer sits. A model that gets the mean right and the dispersion wrong misprices occurrence layers against aggregate covers. The HD curves, calibrated on $55 billion of location- and policy-level claims across more than 2,700 curves, carry the full distribution, with roof age, roof geometry, and construction quality as explicit inputs rather than post-processing corrections.

The financial module brings deductible structures, cosmetic damage exclusions, hours clauses, and post-event loss amplification inside the stochastic engine. Amplification has been a consistent source of miss: the July 2023 North Texas hail outbreak reached roughly $8 billion in part because Dallas-Fort Worth roofing contractor capacity stayed saturated for months. Embedded in the engine, that function scales with event size and geographic concentration instead of being applied as a flat percentage afterward.

The Middle of the Distribution Is Where Attachment Lives

The migration does not move the numbers a cat program is usually stress-tested on. It moves the ones the program is priced on.

Legacy models understate the frequency of moderate-severity events, because coarse resolution and mean-only vulnerability curves miss spatial clustering of moderate hail over dense urban exposure. They do not systematically misestimate the extreme tail, where the physics of convective storms imposes its own ceiling. So HD output raises the 1-in-10 to 1-in-50 return period range more than the 1-in-250 or 1-in-500 estimates.

That asymmetry matters because SCS is a frequency peril. A carrier with material Central Plains exposure can expect 10 to 20 events a year generating loss notifications, of which three to five clear a routine-loss threshold. The structures that fit are aggregate stop-loss, low-attaching per-occurrence covers, and quota shares, not a tower built for an infrequent high-severity peril. Aggregate attachment points sit in precisely the return period band the HD revision moves.

The consequence is arithmetic. If the HD estimate of the 1-in-10 aggregate return period is materially higher than the legacy estimate, an unchanged attachment point now attaches more often than the program was priced for. The cover is thinner than modeled and the premium load is understated, and the carrier discovers both mid-cycle rather than at renewal.

Timing works in the buyer's favor here. Property cat rates fell 15 to 20 percent at June 2026 renewals on record dedicated capital and a below-normal hurricane forecast. A carrier that has run HD output before negotiating can ask for attachment adjustments supported by the revised distribution and extend aggregate limits into soft pricing. One that has not is making two errors at once: a misstated PML, and a reinsurance cost allocation into the primary cat load computed off that misstatement.

Adopting the Better Model Can Cost Capital First

The regulatory calendar and the model calendar are not aligned, and the gap creates a genuinely awkward incentive.

Severe convective storm currently sits in the NAIC's hurricane and earthquake disclosure framework as informational, so it does not feed the Rcat component of the P&C RBC formula. That is changing. The Catastrophe Risk Subgroup adopted a formal wildfire Rcat charge, Proposal 2025-20-CR, at Spring 2026, and the July 9, 2026 meeting is expected to advance the parallel SCS analysis. The consolidation of three cat oversight bodies into a single task force in 2026 puts SCS on the wildfire trajectory rather than in deferral.

The wildfire charge requires carriers to estimate a net 1-in-100 PML using one of four NAIC-approved vendor models, with the charge a factor on that estimate. If SCS follows the same structure and the approved list includes the HD models, a carrier that has already revised its SCS PML upward carries a higher charge than one still reporting legacy output, for holding the better estimate of the same exposure.

Deferring does not escape it. The regulatory deadline arrives whenever it arrives, and a carrier that waits compresses its migration into that window and gives up the informational advantage in reinsurance negotiation at the same time.

The disclosure requirement removes the option of migrating quietly. AM Best's P&C RBC cat model guidelines require PML changes above a threshold against the prior year to be disclosed and explained in the actuarial opinion on risk-based capital. A 20 percent increase in the 1-in-100 SCS PML has to be explained as an architecture change, tested against recent event experience, and paired with the retention and treaty decisions it implies. A carrier that adopts HD output and leaves its reinsurance structure alone has the harder conversation of the two.

Further Reading

• Severe Convective Storms Overtake Hurricanes as the Costliest Insured Peril
• Swiss Re Sigma 1/2026: Secondary Perils Hit 92% of the $107B Nat Cat Bill
• Verisk Synergy Studio Rewrites the Cat Modeling Playbook
• NAIC Merges Three Oversight Bodies Into a Unified Catastrophe Risk Task Force
• Property Cat Reinsurance Down 14%: How to Recalculate Your Cat Load
• AI Replaces the Weekly PML Run: Cat Accumulation Goes Real-Time

Sources

• Moody's: Introducing the North America Severe Convective Storm HD Models (December 2025)
• Moody's: Why HD Models Are the Required Solution for Managing SCS Risk
• Moody's 2025 Catastrophe Review: Wildfires and SCS Dominate Global Losses
• Artemis: Moody's Unveils New North America Severe Convective Storm Models
• Reinsurance News: Moody's RMS Launches New North America SCS Models
• Mondaq: NAIC Spring 2026 Meeting Update, P&C RBC Working Group and Catastrophe Risk Subgroup
• NAIC: Catastrophe Risk (E) Subgroup
• BeInsure: U.S. SCS Losses Hit $42B in First Nine Months of 2025