The Credibility Report - Edition 37

July 03, 2026

AI-curated actuarial intelligence, designed by actuaries, for actuaries.

Opening Bell

This week's signal is that softening catastrophe capacity is no longer just a Florida renewal story. Mid-year market evidence now points to broader property-cat rate reductions, more active alternative capital, and a familiar actuarial trap: lower ceded cost can improve the economics of risk transfer, but it should not automatically become a lower gross loss-cost assumption. The work is in the bridge: exposure, attachment, limit, reinstatement, basis risk, capital relief, and retained volatility.

This Week's Headlines

1. July renewals extend the soft property-cat signal

Guy Carpenter's July 2026 renewal resource center reports continued competitive reinsurance conditions, including a 16% reduction in the global property catastrophe rate-on-line index at mid-year renewals. For actuaries, the useful question is not whether the market is softer; it is whether the renewal movement reflects pure price, changed attachment, additional limit, different peril mix, or better-performing cedant data.

2. Data centers are becoming an accumulation and interruption-risk problem

First Street's global data-center climate-risk report finds that 79% of global data-center capacity is exposed to elevated acute climate hazards, with chronic heat and drought also affecting a large share of capacity. That matters for property, engineering, cyber, technology E&O, and business interruption models because cloud infrastructure concentration can turn local physical hazards into wider service-continuity losses.

3. North Carolina bans third-party litigation funding

Triple-I reports that North Carolina has become the first U.S. state to ban third-party litigation funding. The actuarial read-through is social inflation: if litigation finance materially changes claim severity, settlement timing, and defense strategy, then legal reform is not just a claims department issue; it belongs in trend selections, reserve diagnostics, and scenario testing.

4. Retirement shocks remain a product-design and assumption-governance issue

The SOA Research Institute's 2026 retirement risk survey highlights persistent financial shocks, caregiving gaps, and inflation pressure. Life, annuity, and pension teams should treat this as assumption evidence rather than consumer-color commentary: liquidity strain, care needs, and inflation sensitivity all affect lapse, utilization, drawdown, and guarantee behavior.

5. EIOPA keeps pressure on private-equity-owned insurers

EIOPA's consultation on supervision of insurers and reinsurers related to private equity sets expectations around acquisitions, portfolio transfers, mergers, and ongoing supervision. The actuarial governance issue is asset-liability alignment: shorter investment horizons, illiquid assets, complex ownership structures, and intra-group reinsurance can all move risk without making it disappear.

Research Spotlight

Paper of the Week: lag relevance for forecasting models

Autorelevance function and other feature relevance measures for univariate time series forecasting proposes model-agnostic ways to measure which lags matter in forecasting models. The actuarial translation is practical: mortality, lapse, claims frequency, payment-speed, and inflation forecasts often fail governance review because the model's time dependence is hard to explain.

Structured shrinkage for continuous-time disease networks

Continuous-Time Bayesian Networks with Structured Shrinkage studies time-dependent interactions among multiple long-term conditions. Health and life actuaries should care because morbidity pathways, comorbidity clustering, and transition timing are exactly where simple incidence assumptions lose information.

Linear baselines still matter in time-series forecasting

How Good Can Linear Models Be for Time-Series Forecasting? is a useful antidote to model-fashion risk. Before introducing larger sequence models into pricing, reserving, or capital workflows, teams should know which gains survive against well-tuned linear baselines and which are mostly benchmark theatre.

Uncertainty-aware medical forecasting

Beyond Point Estimates for Glaucoma Visual Field Forecasting is not an insurance paper, but its uncertainty framing is relevant to health analytics: monitoring and intervention decisions need predictive distributions, not only expected trajectories.

Practical Takeaways

• Reinsurance: require a ceded-cost bridge that separates exposure, attachment, limit, reinstatement, terms, model change, loss trend, and pure price before using July renewal savings in planning.
• Property and cyber accumulation: add data-center climate exposure to accumulation views where cloud dependency can create correlated business interruption or technology service losses.
• Casualty reserving: treat litigation-funding reform as a scenario variable for severity trend, settlement lag, and defense cost rather than a binary legal headline.
• Life and retirement: test lapse, utilization, drawdown, and guarantee assumptions against inflation, caregiving, and liquidity-shock evidence.
• Model governance: make every time-series model beat a transparent baseline and explain its lag dependence before promoting it into production.

What We're Watching

• Whether July reinsurance softening changes retained catastrophe appetite or mainly reduces ceded budget pressure.
• Whether infrastructure climate risk becomes a standard input to property, cyber, and business-interruption accumulation models.
• Whether litigation-funding restrictions spread beyond North Carolina and show up in observable casualty severity trends.
• Whether actuarial ML teams become more disciplined about baseline comparisons and time-dependence explanations.

Edition 37 - July 03, 2026