The Credibility Report
Edition 20: The Market Correction Deepens
US commercial rates correct to +2.9%, cyber and D&O now price-declining, AI production era arrives (80% in production, 38% generating value), steepest reinsurance price decline in over a decade, Solvency II 2026 review advancing.
The global P&C market is in a deepening correction phase: US commercial rates softened to +2.9% in Q4 2025, down sharply from +5.6% a year prior, with cyber (-1.5%) and D&O (-2.5%) now in real price decline. Meanwhile, the AI production era has arrived — 80% of carriers now run predictive models in production, though only 38% generate measurable value at scale. This edition focuses on the sharpest reinsurance price correction in over a decade and the research implications of the AI production-to-value gap.
— Ron Richman, Founder, InsureAI
Headlines
US Commercial Rates Correct to +2.9% — Cyber and D&O Now Price-Declining
The Q4 2025 reading marks a decisive turn: down from +5.6% a year ago and +3.8% in Q2/Q3 2025. Cyber (-1.5%) and D&O (-2.5%) are now in real price decline — a buyers' market driven by intensifying competition and loosening reinsurance pass-through.
WTWDecision Delta
- Signal: Correction phase with real declines in some lines.
- New vs last issue: Yes — Q1 2026 confirming and accelerating softening.
- Functions affected: Pricing, underwriting.
- Direction: Softening across most lines.
- Horizon: Through 2026-2027 minimum.
- Confidence: High — multiple broker and rating agency sources.
- Source quality: High — broker data.
- Actuary action: Review rating adequacy models; cat load assumptions need adjustment.
Sixth Year Above $100B: Nat Cat Losses at $107-127B — New Baseline $148B
2025 confirmed at $107B (Swiss Re) / $127B (Aon) — the 6th consecutive year above $100B. No major US hurricane made landfall for the first time in 10 years, yet severe convective storms drove losses above the long-term average. The 2026 planning baseline is now $148B, with peak potential of $320B.
Aon Catastrophe ReportDecision Delta
- Signal: Six consecutive years above $100B = structural baseline, not cyclical exception.
- New vs last issue: 2025 confirmation plus 2026 projections are new.
- Functions affected: Cat modeling, pricing, reinsurance purchasing.
- Direction: Elevated — new baseline established.
- Horizon: Multi-year structural shift.
- Confidence: High — Swiss Re, Munich Re, Aon.
- Source quality: Highest — Aon, Swiss Re.
- Actuary action: Recalibrate cat load factors to $148B baseline; peak at $320B.
April Reinsurance Renewals: 10-30% Price Reduction — Steepest in Over a Decade
Property cat pricing reductions of 10-30% across geographies — the steepest decline in more than 10 years. Record capital is the structural driver. Reinsurer returns on capital remain strong at 20%+ despite cuts. Terms and conditions are broadening. The market expects continued decline through 2027 absent a materially larger loss event.
BeInsureDecision Delta
- Signal: Record capital + disciplined reinsurers = structurally loose; buyers' market.
- New vs last issue: Yes — April renewals confirm January trend at steeper discounts.
- Functions affected: Reinsurance purchasing, capital management.
- Direction: Sharp decline (10-30%), steepest in over a decade.
- Horizon: 2026-2027 unless major loss event.
- Confidence: High — Fitch, Gallagher Re, Aon, Moody's.
- Source quality: High — broker, rating agency, trade press.
- Actuary action: Re-evaluate cession rates; favorable window for multi-year placements.
SOA Emerging Risks: AI + Cyber Ranked #1 Combined Risk for Insurers
January 2026 SOA survey of 1,000+ C-suite leaders finds AI adverse outcomes and cyber events as the top combined threat for the first time, overtaking climate. For P&C specifically, tech risks (AI, cyber) now rank above climate for near-term threats. SOA also published Climate Risk Essay Collection (March 2026) on integrating climate risks into actuarial processes.
SOA Emerging Risks SurveyDecision Delta
- Signal: AI + cyber is the #1 combined emerging risk for insurers.
- New vs last issue: Yes — January 2026 survey is new data.
- Functions affected: Pricing, ERM, underwriting.
- Direction: Emerging risk priority increasing.
- Horizon: 2026-2027.
- Confidence: High — C-suite survey across industry.
- Source quality: Highest — SOA official research.
- Actuary action: Build explicit AI-risk modules in pricing and ERM frameworks.
AI Production Era: 80% Deploy Predictive Models — But Only 38% Generate Value at Scale
2026 is the "AI production era." Underwriting cycles are 70-75% faster in AI-enabled workflows; operating costs reduced 20-40% for adopters. The gap between deployment and value generation is the new competitive fault line. Data infrastructure and governance are the binding constraints.
Insurance Thought LeadershipDecision Delta
- Signal: AI production is real but execution variance is the differentiator.
- New vs last issue: Yes — production numbers new vs. pilot-stage reporting.
- Functions affected: Pricing, underwriting, claims, distribution.
- Direction: Scaling rapidly with large execution variance.
- Horizon: 2026-2027.
- Confidence: High — Everest Group, BCG, Insurance Thought Leadership.
- Source quality: High — consulting + research.
- Actuary action: Assess own AI production maturity; data quality investment is the prerequisite.
Solvency II 2026 Review: ORSA Now Requires Explicit Cyber and Macro Stress Scenarios
Pre-2027 effective date confirmed. ORSA must now explicitly cover macroeconomic developments, cyber risks, and solvency under stressed market conditions. Board-level accountability is being embedded. DORA alignment is a parallel focus. EIOPA is driving Delegated Acts updates on risk-free curve construction and long-term guarantees.
AdvisenseDecision Delta
- Signal: ORSA is becoming more Board-accountable and forward-looking.
- New vs last issue: Partial — Delegated Acts progression and 2027 date are new.
- Functions affected: ERM, capital management, actuarial reporting.
- Direction: Strengthening governance; supervisory convergence.
- Horizon: Pre-2027 effective date.
- Confidence: High — regulator + consulting.
- Source quality: High — EIOPA official + consulting.
- Actuary action: Begin ORSA readiness assessment; ensure cyber and macro stress scenarios are included.
CAS Updates GLM Pricing Monograph with AI Integration; EVT Work Underway
"From GLMs to Comprehensive Insurance Pricing" (Chalk et al.) is updated for October 2026 Exam 8, incorporating predictive analytics for modern rating models. A new Extreme Value Theory monograph for modeling rare events is in development. The $10,000 CAS Monograph Prize was awarded to Goldburd et al. Customer Lifetime Value RFP offers up to $75,000.
CAS MonographsDecision Delta
- Signal: AI/ML integration is now mainstream in actuarial pricing education.
- New vs last issue: Yes — GLM update and EVT development are new.
- Functions affected: Pricing, cat modeling.
- Direction: Practice evolution accelerating.
- Horizon: October 2026 exam cycle.
- Confidence: High — CAS official.
- Source quality: Highest — CAS official publications.
- Actuary action: Update pricing workflows to reflect GLM + AI integration standards.
IFoA "Parasol Lost": Global Warming Sensitivity Underestimated — Transitions Need Supercharging
The January 2026 climate paper warns that current warming sensitivity estimates are underestimated and energy transition investments need to be "supercharged" — with material financial risk implications for insurers. The IFoA Climate Risk and Sustainability Course positions climate alongside traditional actuarial risks (interest rates, mortality) in modeling frameworks.
IFoA Sustainability HubDecision Delta
- Signal: Climate risk is now a core actuarial risk, not a special topic.
- New vs last issue: Ongoing — IFoA publications continuing.
- Functions affected: Cat modeling, ERM, strategic planning.
- Direction: Integrating into standard actuarial practice.
- Horizon: Multi-year.
- Confidence: High — IFoA official.
- Source quality: High — IFoA official.
- Actuary action: Incorporate underestimated warming scenarios into cat load models.
Deep Dive — Market: Reinsurance April Renewals
April 2026 reinsurance renewals have confirmed what brokers and rating agencies had been signaling since January: the property cat reinsurance market is experiencing its steepest price decline in more than ten years. Property cat pricing reductions of 10-30% were the norm across the market — not isolated to best-performing programs, but broadly distributed across geographies and peril groups.
The structural driver is clear: record capital is deployed in the market. Reinsurer balance sheets are strong, and returns on capital remain above 20% even at these lower price levels — meaning the price cuts are not driven by financial stress but by intense competition for premium volume. This is a buyers' market by design.
Terms and conditions are broadening simultaneously: attaching points are moving lower in layers, coverage is stretching further up tower structures, and multi-year deals are being negotiated at levels that would have been unthinkable 18 months ago. Japan's April 1 renewals saw cat pricing down approximately 16% — consistent with the broader Asia pattern of 15-20% declines.
The market expectation — universally expressed across brokers, reinsurers, and rating agencies — is that this softening continues through 2027 absent a major cat event. A single large loss event could interrupt the cycle, but the structural capital overhang is large enough that even a meaningful event may only pause, not reverse, the trend.
For ceding companies, this is a window. Programs that are well-documented, with clean data and transparent exposure management, are in a strong negotiating position. Actuaries responsible for reinsurance program design should be revisiting cession rates and considering multi-year structures to lock in current conditions.
Decision Delta
- Signal: Record capital + disciplined reinsurers = structurally loose; buyers' market.
- New vs last issue: Yes — April renewals confirm January trend at steeper discounts.
- Functions affected: Reinsurance purchasing, capital management.
- Direction: Sharp decline (10-30%), steepest in over a decade.
- Horizon: Through 2027 unless major loss event.
- Confidence: High — Fitch, Gallagher Re, Aon, Moody's corroborate.
- Source quality: High — broker, rating agency, trade press.
- Actuary action: Re-evaluate cession rates; favorable window for multi-year structural placements.
Deep Dive — Research: Weighted Tweedie Framework
Varying Risk Exposure in Auto Insurance: A Weighted Tweedie Framework for Experience Rating and Cancellation Penalties
arXiv | arXiv:2604.02400v1 | Published April 2, 2026
Pricing actuaries have long struggled with mid-term cancellations: the standard assumption that risk exposure is proportional to coverage duration breaks down when policyholders cancel before maturity. This paper addresses it directly.
Using an automobile insurance dataset, the authors build on the classical Tweedie GLM framework to introduce flexible weighting functions and a premium penalty structure that depend on the level of exposure. Mid-term cancelers show different claims profiles — ignoring this heterogeneity leads to mispriced premiums.
The methodological contribution includes formalizing the weighting structure within the Tweedie framework, comparing several weighting approaches using deviance-based criteria, and introducing an area-between-curves criterion from concentration and Lorenz curves plus Murphy diagrams grounded in Bregman dominance — under-used tools in actuarial model comparison that deserve wider attention.
Why actuaries should care: Mid-term cancellations are near-universal across auto books and carry different claims profiles. The weighted Tweedie approach is implementable today. The Bregman/Murphy validation tools are a genuinely novel addition to the model comparison toolkit.
Decision Delta
- Signal: Mid-term cancellations are a source of pricing heterogeneity that standard Tweedie GLMs miss.
- New vs last issue: Yes — most recent arXiv submission (April 2, 2026); directly implementable.
- Functions affected: Pricing, underwriting.
- Direction: New methodology for experience rating.
- Horizon: Applicable now.
- Confidence: High — Canadian auto data; rigorous validation.
- Source quality: High — arXiv, April 2 2026.
- Actuary action: Review own Tweedie pricing models for mid-term cancellation treatment; consider weighting functions and Bregman-based validation criteria.
Practical Takeaways
- Correction phase is real: cyber (-1.5%) and D&O (-2.5%) are price-declining. Review rating adequacy and reinsurance pass-through assumptions.
- Weighted Tweedie paper: mid-term cancellations carry different claims profiles — audit your GLM cancellation treatment.
- Chubb (81.2%) and Travelers (80.2%) set new combined ratio benchmarks. Use 81-85% as "well-managed" reference.
- Industry strengthened reserves +$8.3B in 2025 — workers' comp remains a pressure point requiring separate segmentation analysis.
- ML scenario generators often fail basic tail dependence tests — validate any synthetic scenarios before using for reserve estimates.
- $148B is the new planning baseline for nat cat losses — six consecutive years above $100B means cat load factors should be recalibrated upward permanently.
- ORSA must now explicitly include cyber risk and macro stress scenarios — update cat model scenario libraries accordingly.
- AI + cyber is the #1 combined emerging risk per SOA January 2026 C-suite survey — build explicit modules into enterprise risk frameworks.
- Only 38% of AI deployments generate value at scale — data quality is the binding constraint, not technology sophistication.
From the Actuarial Societies
January 2026 survey of 1,000+ C-suite leaders: AI adverse outcomes and cyber events are the top combined threat for the first time, overtaking climate. For P&C specifically, tech risks rank above climate for near-term threats. SOA also published Climate Risk Essay Collection (March 2026) on integrating climate risks into actuarial processes.
SOA Emerging Risks Survey SOA Climate Essay Collection"From GLMs to Comprehensive Insurance Pricing" (Chalk et al.) updated for October 2026 Exam 8 with full AI and predictive analytics integration. New Extreme Value Theory monograph for rare-event modeling in development. $10,000 CAS Monograph Prize awarded to Goldburd et al. Customer Lifetime Value RFP: up to $75,000 for peer-reviewed research.
CAS Monographs"Parasol Lost" (January 2026) warns that current warming sensitivity estimates are underestimated and energy transition investments need to be "supercharged." The IFoA Climate Risk and Sustainability Course positions climate alongside traditional actuarial risks (interest rates, mortality) in modeling frameworks.
IFoA Sustainability HubFrom the Journals
North American Actuarial Journal. Rigorous framework for evaluating when a model trained on one portfolio can be deployed on another. Directly applicable to commercial lines pricing with small portfolios.
NAAJNAAJ. Privacy breaches carry significantly higher reputational cost than other breach types. Critical for cyber insurance pricing.
NAAJEuropean Actuarial Journal. Constructs and validates climate transition matrices using STOXX Europe 600 data.
EAJJournal of Risk and Insurance. Provides rigorous validation framework for ML-based scenario generators. Applicable to Solvency II ORSA and internal model validation.
JRIFrom arXiv
Why these arXiv papers matter this week: The weighted Tweedie paper (most recent arXiv submission), the climate-copula paper for ORSA scenario libraries, and the Dirichlet ensemble paper for uncertainty quantification in AI deployments are the key picks.
Weighted Tweedie GLM framework accounting for mid-term policy cancellations in auto insurance pricing. Implementable today.
arXiv:2604.02400v1Copula-based framework for sovereign credit rating migration with climate risk extensions. Relevant for ORSA scenario libraries.
arXiv:2604.07567v1Dirichlet parameter estimation using ensembles of softmax outputs for uncertainty quantification in neural network classifiers. Relevant for AI deployment trust.
arXiv:2604.06032v1Robust utility maximization with known marginal but unspecified dependence structure. Relevant for reinsurance pricing under model ambiguity.
arXiv:2604.04649v1What We Are Watching
April renewals confirmed 10-30% reductions. We are watching whether the anticipated continued softening materializes, or whether the ~$148B nat cat baseline triggers a pause. The difference between a buyers window and a structural shift is a single large event.
80% of carriers in production is the headline; the real story is the 42-percentage-point gap between deployment and value generation. We track whether this gap narrows as data infrastructure investment catches up.
Congress must act. A failure to reauthorize would have profound implications for flood insurance markets and reinsurance capacity in the US. Watch for legislative movement in Q2-Q3 2026.
"Parasol Lost" is positioning climate risk alongside mortality and interest rates in actuarial modeling. We are watching for how quickly this translates into ORSA scenario libraries and pricing assumptions.
The Credibility Report is published weekly. Forward to a colleague who prices risk.
— The Credibility Report
Edition 20 | April 11, 2026