From tracking vendor cat model updates across three major providers over the past two years, none have yet incorporated supershear effects into their standard earthquake modules, meaning carriers relying on off-the-shelf models are pricing with a known but unaddressed gap. A new paper published in the Journal of Catastrophe Risk and Resilience by MS Amlin researchers Luke Wedmore and William Sturgeon quantifies this gap for the first time: supershear earthquakes accounted for 66% of insured earthquake losses since 2016, roughly $13.2 billion, yet remain entirely absent from standard seismic hazard models, building design codes, and insurance catastrophe models. The finding lands at a moment when the California Earthquake Authority just priced a $425 million Sutter Re cat bond, Moody's RMS is updating its North America Earthquake HD Model, and property-cat reinsurance rates are falling at the fastest pace since 2014. For actuaries managing earthquake portfolios, structuring cat bonds, or setting capital requirements, the question is no longer whether supershear matters but how quickly vendor models and internal views of risk will catch up.

What Supershear Earthquakes Are

In a conventional earthquake, the rupture front propagates along the fault surface at speeds below the shear wave velocity of the surrounding rock, typically around 3 kilometers per second in the Earth's crust. A supershear earthquake breaks this speed barrier. The rupture front accelerates past the shear wave velocity and can reach compressional wave speeds of nearly 6 km/s, creating a Mach cone analogous to the sonic boom produced by a supersonic aircraft.

The physical consequences are severe and distinct from normal earthquakes in three ways that matter for loss modeling:

Amplified ground shaking. The Mach cone generated by a supershear rupture concentrates seismic energy into a coherent wavefront. Instead of spreading radially from the fault, the energy travels in a focused corridor with less geometric spreading and less attenuation than subshear ruptures. Ground motion intensities are higher not just near the fault but at considerable distances from it, meaning damage footprints extend well beyond what conventional attenuation models predict.

The "double punch" effect. Supershear ruptures generate successive seismic waves that arrive in rapid sequence, subjecting structures to repeated loading cycles before they can dissipate the energy from the initial pulse. Standard structural engineering design codes assume a single primary shaking event that attenuates over time. The double punch violates this assumption and causes disproportionate structural damage, particularly in mid-rise and high-rise buildings where resonance effects compound across floors.

Unusual torsional forces. MS Amlin's research highlights that supershear events produce rotational ground motions that twist structures around their vertical axis. Taller buildings are especially vulnerable because the torsional moment arm increases with height. This damage mechanism is entirely outside the scope of current building codes and, consequently, outside the vulnerability functions embedded in commercial cat models.

Supershear rupture is not a rare curiosity. Approximately 36% of major strike-slip earthquakes globally since 2010 have involved supershear propagation. The phenomenon occurs preferentially on long, straight, mature strike-slip faults, the exact geological features that characterize the San Andreas Fault in California, the North Anatolian Fault in Turkey, and the Sagaing Fault in Myanmar.

The $13.2 Billion Loss Gap

The central finding from Wedmore and Sturgeon's paper is stark. Of all insured earthquake losses worldwide since 2016, two-thirds came from events where supershear rupture was the dominant propagation mode. The aggregate: an estimated $13.2 billion in insured losses driven by a seismological phenomenon that no commercial catastrophe model accounts for.

This is not a theoretical exercise. The losses are real, the claims are paid, and the retrospective attribution to supershear mechanics is well supported by seismological analysis. Consider the recent track record:

Turkey-Syria earthquakes, February 2023. The Mw 7.8 Kahramanmaraş earthquake sequence devastated southeastern Turkey and northwestern Syria, killing over 62,000 people and generating insured losses estimated at $5.8 billion by CRESTA and TRY 116.9 billion by PERILS. Seismological analysis confirmed that the first earthquake exhibited supershear rupture characteristics, creating the focused shaking corridor and double-punch loading that contributed to the catastrophic collapse of thousands of structures. The Turkish government estimated total economic losses at approximately $105 billion.

Myanmar earthquake, March 2025. The Mw 7.7 earthquake along the Sagaing Fault produced a surface rupture extending 475 km, roughly 230 km longer than standard models predicted. The rupture initiated as bilateral subshear propagation but transitioned to supershear velocity of approximately 5.3 km/s about 100 km south of the epicenter, sustaining this velocity for more than 200 km. The event demonstrated that supershear propagation is not confined to a brief phase of the rupture; it can dominate the event for hundreds of kilometers.

San Francisco, 1906. Retrospective analysis has since identified the 1906 San Francisco earthquake as a supershear event. The rupture traveled approximately 470 km along the San Andreas Fault and produced shaking intensities that devastated the city and its surroundings. This reclassification is significant because the 1906 event anchors many of the historical loss scenarios used in California earthquake risk assessment.

Luke Wedmore, Senior Research Analyst at MS Amlin, stated it directly: "The sonic boom produced by these ruptures can cause more intense and widespread damage, yet the impact is significantly underestimated in models."

How Supershear Changes Cat Model Output

MS Amlin did not stop at identifying the blind spot. The researchers modeled the impacts of incorporating supershear effects on representative insurance and reinsurance portfolios, producing results that should concern anyone setting catastrophe loads or capital requirements.

5-10%
Loss increase at the 200-year return period when supershear effects are incorporated
30-60%
Loss increase at the 500-year return period when supershear effects are incorporated
66%
Share of global insured earthquake losses since 2016 attributable to supershear events

The asymmetry between the 200-year and 500-year impacts matters enormously. A 5-10% increase at the 200-year level is a meaningful adjustment for pricing and reserving, but carriers might absorb it through margin adjustments. The 30-60% increase at the 500-year return period is a different proposition entirely. This is the tail risk territory where solvency capital requirements, catastrophe excess-of-loss treaty structures, and cat bond attachment points are calibrated. A 30-60% understatement of loss at the tail means that:

  • Required capital is understated. Solvency II and similar frameworks derive the Solvency Capital Requirement (SCR) from the 200-year VaR. Even the 5-10% increase implies that earthquake-exposed carriers may be holding insufficient capital. For carriers writing earthquake business in California, Japan, or Turkey, the revision could trigger recalibration of internal models.
  • Reinsurance attachment points are too low. If the modeled 500-year PML increases by 30-60%, then the cat excess-of-loss treaty that was structured to attach above the "remote" loss scenario may now sit within the more probable loss range. The attachment itself becomes less remote than priced.
  • Cat bond expected losses are understated. Indemnity-triggered earthquake cat bonds rely on the sponsor's cat model to estimate expected losses. If the model does not account for supershear effects, the modeled expected loss is lower than the actual probability of attachment, meaning investors are receiving less spread per unit of risk than the model suggests.

California: Ground Zero for Supershear Risk

MS Amlin's research identifies California as the highest-consequence exposure zone for the supershear blind spot. Wedmore noted: "There is a significant chance that earthquake risk in California is markedly underestimated."

The geological setup is textbook. The San Andreas Fault is a long, relatively straight, mature strike-slip fault that has been locked for over a century in its southern section. Stress has been accumulating steadily. Research published in 2025 found that stresses along the San Andreas and San Jacinto faults are at their highest levels in the past 1,000 years. These are precisely the conditions under which supershear rupture initiates: high accumulated stress on a geometrically favorable fault that allows the rupture front to accelerate through the forbidden zone between the Rayleigh wave speed and shear wave speed.

The 1906 San Francisco earthquake, now classified as a supershear event, demonstrated that the San Andreas can sustain supershear propagation over hundreds of kilometers. The southern locked section, running from the Salton Sea through the Inland Empire and the western edge of Los Angeles County, represents an enormous concentration of insured exposure along a fault segment primed for the same type of rupture.

California also happens to be the world's largest earthquake insurance market. The California Earthquake Authority alone manages a risk transfer program that reached approximately $9.15 billion as of June 2026, combining traditional reinsurance with $2.875 billion in catastrophe bond coverage. Add in private carrier earthquake books, surplus lines markets, and commercial earthquake policies, and the total insured earthquake exposure in California runs into the hundreds of billions.

Every dollar of that exposure is being modeled, priced, and capitalized using tools that do not account for supershear rupture mechanics.

The Sutter Re 2026-1 Case Study

The timing of the MS Amlin research could not be more pointed. In May 2026, the California Earthquake Authority priced its latest catastrophe bond, Sutter Re Ltd. Series 2026-1, raising $425 million in fully collateralized earthquake reinsurance protection. The deal was structured across two tranches:

Tranche Size Expected Loss Spread (Final) Initial Guidance
Class C Notes $325M 2.30% 3.50% 4.25% - 5.00%
Class F Notes $100M 4.05% 5.50% 6.50% - 7.25%

The bond uses an indemnity trigger on an annual aggregate basis with a four-year coverage period. It replaces a $425 million tranche from Sutter Re 2023-1 that matured on June 13, 2026. Strong investor demand allowed the CEA to upsize twice, first from $300 million to $400 million, then to the final $425 million, while simultaneously tightening spreads well below initial guidance.

The key question is whether the expected loss estimates of 2.30% and 4.05% adequately reflect the supershear risk. These expected loss figures are generated by the CEA's catastrophe model, which, like all vendor models currently in the market, does not incorporate supershear rupture dynamics. If the MS Amlin research is correct that losses at the 200-year return period increase by 5-10% when supershear is modeled, then the expected loss on both tranches may be understated.

For the Class C notes at 2.30% expected loss, even a modest upward revision to 2.50-2.55% would change the risk-return calculus for ILS investors who calibrate their portfolios around precise expected loss multiples. For the Class F notes at 4.05%, a revision reflecting supershear effects could push the expected loss toward 4.25-4.45%, which at the final spread of 5.50% would represent a tighter multiple than investors priced in.

The broader ILS market context amplifies the concern. More than $6 billion in purely earthquake-exposed catastrophe bonds are outstanding across the 144A market, with an additional $17.6 billion in multi-peril bonds that include earthquake exposure. If the underlying models are systematically understating earthquake tail risk by 30-60% at the 500-year level, the entire earthquake ILS sector may be mispriced relative to the actual hazard.

What Cat Model Vendors Have Not Done

The gap between academic seismology and commercial catastrophe modeling is not new, but the supershear blind spot represents an unusually clear-cut case where the science is established, the loss impact is quantified, and the vendor response has been absent.

Supershear earthquake rupture is not a fringe theory. The seismological literature has documented supershear events dating back decades, and the phenomenon has been observed in laboratory fault experiments, numerical simulations, and field measurements from multiple earthquakes across different tectonic settings. The 2023 Turkey earthquakes, which generated one of the largest insured earthquake losses in recent history, provided direct observational evidence of supershear propagation with immediate insurance implications.

Yet as of June 2026, none of the three major commercial cat model vendors have incorporated supershear rupture dynamics into their standard earthquake modules:

Moody's RMS is updating its North America Earthquake HD Model, previewed at the Exceedance 2026 conference in Fort Lauderdale in early June. The update promises new science on earthquake occurrence and hazard assessment alongside reassessed exposure vulnerability. However, the publicly available conference materials do not specifically mention supershear rupture modeling. Whether the HD update will address this particular gap remains uncertain.

Verisk is in the midst of migrating its entire catastrophe modeling platform from Touchstone to Synergy Studio, a cloud-native environment supporting over 110 risk models. The platform transition has consumed significant development resources, and there is no public indication that the Next Generation earthquake models slated for Synergy Studio will incorporate supershear effects.

CoreLogic provides earthquake models approved for use by the NAIC and state regulators. The California DOI's approval of third-party cat models for rate filings creates a regulatory dependency on vendor model output. If the approved models do not capture supershear risk, rate filings built on those models may embed a systematic underestimate of earthquake loss potential.

MS Amlin, to its credit, has already updated its own internal catastrophe models and view of risk to incorporate supershear effects. The Lloyd's insurer is not waiting for vendor updates. But most carriers and reinsurers do not have the in-house seismological expertise or modeling infrastructure to develop proprietary overlays for this specific hazard characteristic.

What Actuaries Should Demand in the Next Vendor Update Cycle

The MS Amlin paper provides a clear framework for what carriers should be asking from their cat model vendors. The researchers recommend four specific actions that translate directly into vendor due diligence questions for the next model update cycle:

1. Identify which faults are supershear-capable. Not every fault can sustain supershear rupture. The phenomenon requires specific geological conditions: long, straight, mature strike-slip faults with sufficient accumulated stress. Vendors should classify every modeled fault segment by its supershear potential and flag the high-probability candidates in their event sets. For North American models, the San Andreas, Hayward, and Calaveras faults in California are obvious candidates, but the analysis should extend to every major strike-slip feature in the modeled domain.

2. Modify ground motion prediction equations (GMPEs). Current GMPEs in cat models assume subshear rupture propagation. For supershear-capable faults, the vendor should provide alternative or supplementary GMPEs that account for the Mach cone radiation pattern, extended damage corridors, and amplified ground motions at distance. This is not a small change; it affects the entire hazard module for every event on every flagged fault.

3. Update vulnerability functions for torsional loading. The unusual torsional forces generated by supershear rupture damage buildings through mechanisms that standard vulnerability curves do not capture. Building classes with significant height, irregular floor plans, or soft-story configurations are particularly susceptible. Vendors should develop supplementary vulnerability modifiers for torsional loading or, at minimum, provide sensitivity factors that carriers can apply to their portfolio-level results.

4. Integrate supershear scenarios into deterministic and stochastic event sets. The MS Amlin paper recommends testing alternative shaking patterns within existing models as an interim step before full integration. Carriers can adopt this approach immediately by requesting deterministic supershear scenarios from their vendors for key California earthquake events and comparing the modeled losses against their standard model output. The delta between the two provides a direct estimate of the model risk associated with the current blind spot.

Beyond these technical demands, actuaries have a governance obligation. ASOP No. 56 on Modeling requires actuaries to understand the limitations of models used in their work. The supershear blind spot is now a documented, quantified limitation with published research attributing $13.2 billion in losses to the gap. Actuaries relying on vendor earthquake models without acknowledging this limitation in their documentation may fall short of the standard's intent.

Implications for Capital Adequacy and Reinsurance Pricing

The practical consequences of the supershear blind spot extend across three dimensions of actuarial work: capital modeling, reinsurance purchasing, and rate-level adequacy.

Capital modeling. Carriers using vendor earthquake models to determine their SCR or internal economic capital are, per the MS Amlin research, understating their earthquake tail risk by potentially significant amounts. The 5-10% increase at the 200-year level and 30-60% at the 500-year level directly impacts the risk charge calculation. For a carrier with $500 million in earthquake PML at the 250-year return period, a 7.5% upward revision translates to $37.5 million in additional capital that should be held against earthquake risk. Across the California market, the aggregate capital shortfall implied by the supershear adjustment could run into the billions.

Reinsurance purchasing. Treaty structures calibrated to model output may need revision. If a carrier's cat excess-of-loss program attaches at the 250-year modeled PML, and that PML is 7.5% low due to the supershear gap, the attachment point effectively sits at a shorter return period than intended. The carrier is retaining more risk than its reinsurance program is designed to address. This matters especially at mid-year renewals where earthquake programs are being restructured in a softening market. Lower reinsurance rates create an opportunity to buy additional limit, but the sizing of that additional limit should reflect a corrected view of earthquake tail risk.

Rate-level adequacy. For carriers writing earthquake coverage in California, the catastrophe load in rate filings is derived from model output. If the model understates earthquake risk, the filed rate is inadequate relative to the actual hazard. The California DOI's approved cat models are the regulatory standard, but that approval does not insulate carriers from experience that eventually reveals the model's limitations. The pattern of cat model revisions following major loss events, where vendors update their models after the fact and carriers absorb the interim mispricing, is well documented. The supershear research provides an opportunity to get ahead of that cycle rather than waiting for the next California earthquake to force the revision.

The reinsurance market context adds urgency. Property cat reinsurance rates have been declining at double-digit rates through the first half of 2026, with record reinsurer capital compressing spreads across the market. If earthquake risk is systematically underpriced because models do not capture supershear effects, the softening market may be compounding the gap. Reinsurers are accepting lower rates for a risk that may be larger than the models suggest.

The Broader Pattern: Known Model Gaps and the Actuarial Response

The supershear blind spot fits a broader pattern that actuaries working in catastrophe risk should recognize. Swiss Re's sigma 1/2026 documented that secondary perils, once considered immaterial, drove 92% of $107 billion in 2025 insured nat cat losses. Secondary perils were undermodeled for years before the loss experience forced vendor recalibration. Severe convective storm models have undergone multiple revisions in recent years as actual losses consistently exceeded model predictions.

The difference with supershear is that the research community has now provided a clear, quantified framework for the gap before a major supershear event in a highly insured market like California forces the issue. The $13.2 billion figure is a backward-looking attribution; the forward-looking question is what happens when a supershear rupture on the San Andreas produces modeled losses 30-60% above what the industry expected.

For actuaries, the response should be proactive rather than reactive. The MS Amlin paper provides the basis for a stress test that any carrier or reinsurer with earthquake exposure can conduct immediately: take the current vendor model PML at key return periods, apply the published supershear adjustment factors (5-10% at the 200-year, 30-60% at the 500-year), and assess the impact on capital adequacy, treaty attachment points, and rate-level indications. This does not require a vendor model update. It requires actuarial judgment and a willingness to incorporate published research into the view of risk, consistent with the ASOP No. 56 expectation that actuaries understand and communicate model limitations.

MS Amlin has already moved. The rest of the market should not wait for a vendor release cycle that may still be years away.

Further Reading

Sources