The Credibility Report
Edition 19: Reinsurance Renewals, Cloud Risk Cat Bonds, and Record ILS Capital
April renewals soften further, Hannover Re launches first cloud risk cat bond, Stone Ridge ILS AUM hits $6.5bn, Japan cat rates down 16%, and a structural shift in the cat bond market.
The reinsurance market is sending mixed signals heading into mid-April: pricing continues to soften across most lines at the April renewals, yet investor appetite for catastrophe risk remains robust — Stone Ridge Asset Management reports ILS fund assets at their highest since late 2018, approaching $6.5 billion. Meanwhile, the cat bond market is branching into new territory: Hannover Re's $35 million cloud-risk bond marks the first dedicated expansion of cat bond capacity for data center exposures, a peril that is increasingly material as digital infrastructure becomes critical to economic resilience. On the primary insurance side, MGAs are turning to actuarial firepower to secure capacity, a trend that points to continued pressure on underwriting economics.
— Ron Richman, Founder, InsureAI
📰 Headlines
🏢 Arundo Re Posts 5% Revenue Growth, Combined Ratio Improves to 88.8% in 2025
Arundo Re delivered a solid 2025 result, growing its top line by 5% while improving its combined ratio to 88.8% — a signal that selective underwriting and expense discipline are translating into bottom-line gains even in a softening rate environment.
theinsurer.com →Decision Delta
- Signal: Selective reinsurers can still earn solid combined ratios despite softening rates — underwriting discipline pays.
- New vs last issue: New data point.
- Functions affected: Pricing, reserving, underwriting.
- Direction: 🟢 Bullish for disciplined underwriters.
- Horizon: Current year.
- Confidence: High — reported financials.
- Source quality: High — primary source (theinsurer.com).
- Actuary action: Benchmark your combined ratio against 88.8%; if above it, investigate expense ratio and attritional loss assumptions.
☁️ Hannover Re Launches $35m Cloud Risk Cat Bond, Expanding Peril Coverage for Data Centers
Hannover Re has structured a $35 million catastrophe bond specifically targeting cloud service disruption risk — a notable first for the traditional cat bond market, which has historically focused on natural catastrophes. The bond signals growing capital markets appetite for non-natural peril risk and could open a new capacity channel for data center exposures.
theinsurer.com →Decision Delta
- Signal: Capital markets are willing to take on cloud disruption risk via cat bond structures — new capacity channel emerging.
- New vs last issue: New peril expansion.
- Functions affected: Cat modeling, pricing, ERM.
- Direction: 🟢 New market infrastructure forming.
- Horizon: 12–24 months.
- Confidence: Medium — early-stage market development.
- Source quality: High — primary source.
- Actuary action: Begin quantifying cloud disruption exposure for data center accounts; watch pricing benchmarks from this bond.
🖥️ Insurers Turn to Catastrophe Bonds to Expand Data Center Risk Coverage
A broader industry trend: insurers are using cat bonds not just for traditional natural catastrophe protection, but to free up capacity for data center and technology-related risks. As digital infrastructure grows, traditional reinsurance capacity for these risks is proving insufficient, pushing carriers toward capital markets solutions.
programbusiness.com →Decision Delta
- Signal: Cat bonds are becoming a structural tool for capacity management, not just nat cat hedging.
- New vs last issue: New application of existing instrument.
- Functions affected: Cat modeling, ERM, capital management.
- Direction: ↗️ Expanding use case for ILS.
- Horizon: Ongoing.
- Confidence: Medium — market trend.
- Source quality: Good — trade press.
- Actuary action: Evaluate whether cat bond structures could reduce cost of capital for your data center book.
💰 Stone Ridge ILS Fund Assets Approach $6.5bn — Highest Since Late 2018
Stone Ridge Asset Management reports its mutual catastrophe bond and ILS fund strategies have grown to nearly $6.5 billion in assets under management, a level not seen since late 2018. The recovery reflects strong investor confidence in the ILS asset class and ample capital availability in the reinsurance market.
Artemis.bm →Decision Delta
- Signal: ILS capital is fully recovered from 2017–2018 loss years; record capital may compress cat bond spreads further.
- New vs last issue: New AUM milestone.
- Functions affected: Cat modeling, ERM, capital management.
- Direction: 📉 Bearish for cat bond pricing (more capital = tighter spreads).
- Horizon: Current.
- Confidence: High — asset manager disclosure.
- Source quality: Primary (direct from Stone Ridge via Artemis).
- Actuary action: Factor tighter cat bond spreads into alternative capital cost assumptions for cat modeling.
🌪️ Fewer Storms Expected in 2026 — But Insurers Still Face Losses: CSU
Colorado State University's early April forecast calls for a "somewhat below normal" 2026 Atlantic hurricane season. While welcome news, CSU cautions that loss potential remains material: a single landfalling major hurricane can still drive industry losses in the tens of billions, and models must remain calibrated for tail risk.
Triple-I →Decision Delta
- Signal: Below-normal hurricane season forecast — potential rate relief at January 2027 renewals.
- New vs last issue: New seasonal forecast.
- Functions affected: Cat modeling, pricing, reserving.
- Direction: 🟢 Cautiously bullish for hurricane-exposed books.
- Horizon: June–November 2026.
- Confidence: Medium — seasonal forecasts have wide error bands.
- Source quality: High — authoritative academic source.
- Actuary action: Update cat loading assumptions for H2 2026; communicate hurricane season outlook to underwriters.
📊 Aon: Record Reinsurance Capital Fueling Insurer Growth Ambitions
Aon's research indicates that record levels of reinsurance capital are enabling primary insurers to take on more risk and pursue growth strategies they might otherwise constrain. The ample capacity is keeping reinsurance pricing competitive while allowing carriers to expand into new lines and geographies.
Coverager →Decision Delta
- Signal: Abundant reinsurance capital is a tailwind for insurer growth — competitive pressure on reinsurance pricing continues.
- New vs last issue: New market intelligence.
- Functions affected: Pricing, underwriting, ERM.
- Direction: 📉 Bearish for reinsurance pricing; 🟢 Bullish for primary insurers seeking capacity.
- Horizon: Current.
- Confidence: Medium — broker research.
- Source quality: Good — major broker.
- Actuary action: Use Aon's capital market intelligence to benchmark your reinsurance program costs against industry trends.
🏦 Reinsurers Better Prepared for a Softening Market Than Ever Before: JP Morgan
JP Morgan's analysis finds that reinsurers have significantly strengthened their claims reserves, diversified their portfolios, and improved their capital structures since the last soft market cycle — positioning them to remain profitable even as pricing continues to soften. A meaningful shift from prior cycles where softening rates eroded returns sharply.
Reinsurance News →Decision Delta
- Signal: Reinsurers are structurally more resilient to soft market — no imminent correction likely.
- New vs last issue: New analyst view.
- Functions affected: Pricing, reserving, ERM.
- Direction: ↘️ Soft market may persist longer than historical norms.
- Horizon: 12–24 months.
- Confidence: Medium — sell-side research.
- Source quality: Good — major investment bank.
- Actuary action: Plan for extended soft market; focus on expense efficiency and attritional loss ratio discipline.
🇱🇺 Surprise Decline Hits Luxembourg's Reinsurance Market
Luxembourg's otherwise stable reinsurance hub experienced an unexpected contraction, with several market participants reporting fewer new mandates and tighter terms. The decline contrasts with broader European trends and may reflect specific geopolitical or regulatory pressures on Luxembourg-domiciled structures.
Insurance Business →Decision Delta
- Signal: Regional concentration risk is real — Luxembourg market showing unexpected weakness.
- New vs last issue: New regional data point.
- Functions affected: ERM, underwriting, capital management.
- Direction: ⚠️ Caution on Luxembourg concentration.
- Horizon: Current.
- Confidence: Medium — early warning signal.
- Source quality: Good — trade press.
- Actuary action: Review any Luxembourg-domiciled reinsurance structures in your program for counterparty risk.
Editor's note: Three cat bond headlines this week (Stone Ridge, Palomar, Kin) trace back to Artemis.bm as the originating source. The cat bond market's primary reporting ecosystem is narrow — these stories are included because they are genuinely new and materially important. No alternative primary sources were available for these items.
🔬 Research Spotlight
Transfer Learning in the Actuarial Domain: Foundations and Applications
North American Actuarial Journal | DOI: 10.1080/10920277.2025.2489637
This paper addresses one of the most practical questions in modern actuarial AI: when can a model trained on one portfolio or line of business be deployed successfully on another? Transfer learning — adapting a pre-trained model to a new target domain — has transformed computer vision and NLP, but its application to actuarial contexts has been underexplored and often misunderstood.
The authors lay out a rigorous framework for evaluating transferability in insurance settings, distinguishing between homogeneous transfer (same LOB, different geography) and heterogeneous transfer (different LOB, different exposure structure). They derive bounds on the performance penalty expected when transferring between domains, and demonstrate empirically that even modest amounts of target-domain data can dramatically improve transfer outcomes.
For practicing actuaries, the key insight is that transfer learning is not a magic wand — but it is a legitimate tool when the source and target domains share sufficient structure. The paper provides a principled methodology for deciding when transfer is appropriate, which could significantly reduce the data requirements for ML-based pricing models in commercial lines where some portfolios are too small to train from scratch.
💡 Why actuaries should care: This bridges the gap between ML research and actuarial practice — well-grounded in theory, clear about assumptions, and directly applicable to real underwriting decisions.
Additional Papers This Week
Compares offset vs. ratio approaches for Tweedie models when policies cancel mid-term. Shows the choice materially affects estimated relativities — essential for frequency-severity pricing modelers.
European Actuarial Journal | DOIComprehensive ML benchmark for auto fraud detection — ensemble methods consistently outperform single models. Highlights class imbalance as the key practical constraint.
Journal of Risk and Insurance | DOIProposes statistical distance metrics for validating synthetic data generators. Results sobering: many popular generators fail basic tail dependence and correlation structure tests.
Journal of Risk and Insurance | DOIQuantifies reputational penalties from privacy-related cyber events using public company event studies. Finds privacy breaches carry significantly higher reputational cost — critical for cyber insurance pricing.
NAAJ | DOI🔍 Deep Dive: Cat Bond Market Enters New Territory
The catastrophe bond market is evolving faster than many observers expected. Three developments this week illustrate the breadth of this evolution.
First, Hannover Re's $35 million cloud risk cat bond — the first dedicated capital markets instrument targeting cloud service disruption — signals that the industry has found a workable structure for transferring technology-peril risk to capital markets investors. Cloud disruption has long been identified as a material, non-natural exposure that traditional cat bonds did not address. The successful structuring of this bond suggests the market for non-natural peril ILS will continue to expand.
Second, the continued growth in cat bond sponsors — Palomar ($375m), Kin ($335m), Florida Peninsula ($250m), One Alliance ($125m), Mapfre Re ($200m) — reflects broadening adoption across geographies and perils. This diversifying is healthy for the market: concentration risk is reduced, and the investor base grows as new sponsors bring new risk profiles.
Third, Stone Ridge's $6.5 billion AUM milestone is a reminder that ILS capital is not just recovered from the 2017–2018 loss years — it has surpassed prior highs. More capital in the ILS space means more competition for reinsurance risk, which ultimately compresses spreads and challenges traditional reinsurer returns.
For actuaries: cat bond pricing is increasingly a relevant benchmark for reinsurance programs, cloud risk requires explicit modeling and capital allocation, and the expanding ILS market means the boundary between reinsurance and capital markets is becoming permanently more porous.
Decision Delta
- Signal: Cat bond market broadening in peril coverage, sponsor diversity, and total capacity — structural shift, not a cycle.
- New vs last issue: New instruments (cloud risk), new sponsors, new AUM high.
- Functions affected: Cat modeling, pricing, ERM, capital management.
- Direction: 📉 More competitive ILS market — tighter spreads, more capacity chasing risk.
- Horizon: Ongoing structural change; cloud risk cat bond pricing available in 3–6 months.
- Confidence: High — multiple independent data points corroborate the trend.
- Source quality: Multiple sources (primary + trade press).
- Actuary action: Add cloud disruption to cat modeling framework; use cat bond spread indices as pricing benchmark for reinsurance program cost-of-capital estimates.
💡 Practical Takeaways
- • Benchmark your combined ratio against Arundo's 88.8% — if above it, pressure-test expense assumptions.
- • Transfer learning can reduce data requirements for small-portfolio ML pricing models — legitimate tool when domains share structure.
- • Tweedie model choice (offset vs. ratio) materially affects mid-term cancellation relativities — audit your exposure methodology.
- • ML scenario generators often fail basic tail dependence tests — validate any synthetic scenarios before using for reserve estimates.
- • Luxembourg market showing unexpected weakness — review counterparty concentration in reinsurance recoverable.
- • Begin adding cloud disruption to cat modeling frameworks — Hannover Re's bond provides first pricing benchmark.
- • Below-normal CSU hurricane forecast is a starting point, not a tail-risk excuse — keep cat loads calibrated for tail events.
- • ILS capital at record highs means cat bond spreads likely stay compressed — update alternative capital cost assumptions.
- • Luxembourg counterparty concentration warrants review — unexpected weakness in a stable hub is a warning sign.
📚 From the Journals
Compares five biological age methods (MLR, KDM, PhenoAge, calibrated PhenoAge, Random Forest) for mortality and disease prediction using NHANES data — relevant for life and health underwriting.
European Actuarial Journal | DOIDevelops methodology for constructing and validating climate transition matrices using STOXX Europe 600 data; assesses denominator choice effects on corporate carbon scaling.
European Actuarial Journal | DOIApplies drawdown-optimisation techniques to reinsurance structure design — shows how optimal reinsurance can stabilise surplus fluctuations beyond traditional expected value criteria.
European Actuarial Journal | DOIModels embedded optionality in guaranteed lifetime withdrawal benefit annuities with conversion to LTC combo products — increasingly relevant as hybrid life/LTC products proliferate.
Scandinavian Actuarial Journal | DOIQuantifies reputational penalties from privacy-related cyber events; finds they carry significantly higher reputational cost than other breach types — critical for cyber insurance pricing.
NAAJ | DOI🔬 From arXiv
Why these arXiv papers matter this week: The wavelet-copula paper and the cardiovascular copula paper both address the same core actuarial problem — modeling dependence in high-dimensional, non-Gaussian systems. These methods are becoming practical for production actuarial use as variational inference tools mature. Actuaries working on cat bond pricing, enterprise risk modeling, or life insurance with competing risks should monitor this space.
Variational inference combining wavelet-based marginal representations with copulas for dependence modeling — directly applicable to high-dimensional insurance loss aggregation where standard parametric copulas fail.
arXiv:2604.02116 | Apr 2, 2026Novel Gibbs sampling algorithms (PL-Cox and GPL-Cox) for Cox proportional hazards models using Pólya-Gamma data augmentation — practical advance for survival analysis in life insurance and critical illness.
arXiv:2604.06034 | Apr 7, 2026Extends distributionally robust optimization to an online Bayesian framework where ambiguity sets update as new data arrives — applicable to dynamic economic scenario generation and solvency capital calculations.
arXiv:2604.06936 | Apr 8, 2026Nonparametric copula-based framework for modeling dependence between multiple correlated time-to-event outcomes — relevant for critical illness insurance where disease onset times are interdependent and subject to informative censoring.
arXiv:2604.03970 | Apr 10, 2026👀 What We're Watching
JP Morgan and Aon both signal the soft market has structural staying power this time around. Reinsurers are leaner, reserves are stronger, and capital is abundant. Don't expect a correction in 2026.
Hannover Re's cat bond is a proof-of-concept for cloud disruption risk transfer. If it prices well, expect rapid expansion. Actuarial models need to catch up.
April 1 Japan renewals showed sharper-than-expected cat rate decline. This follows a broader Asian market softening that contrasts with parts of the US market. Regional differentiation is increasing.
"Somewhat below normal" is not "safe." A single major hurricane landfall can still produce $50bn+ in industry losses. Don't let a benign forecast drive complacency in cat loading.
The Credibility Report is published weekly. Forward to a colleague who prices risk.
— The Credibility Report
Edition 19 | April 10, 2026