Insurance Desk

Cat Bond Desk Soren Vaeth

The Artemis dashboard hands us a clean quantitative frame this week: $65.8 billion in outstanding risk capital, $18.9 billion in YTD issuance across 92 deals, market yield sitting at 9.46% — 5.71% insurance risk spread over a 3.75% collateral return — against a market-level expected loss of 2.44%. That puts the spread-over-EL at roughly 3.27 percentage points at the market level. The 10Y-2Y curve at 0.46 percentage points and effective fed funds at 3.63% tell you the collateral return is not going to compress meaningfully from here in the near term, so the all-in yield stays attractive to allocators who would otherwise be staring at HY OAS of 2.71% in a tight credit market. The cat bond is still the better carry trade on a risk-adjusted basis relative to high-yield credit this week.

The J.P. Morgan report flagged by Artemis makes a claim worth examining carefully: that ILS and alt-capital are exhibiting 'few of the undisciplined behaviors witnessed in the last soft market.' This is the right question to ask at $18.9 billion of YTD issuance. The recent deal flow supports it — Matterhorn Re at $345 million (Swiss Re cedent, US and Canada named storm and earthquake), 3264 Re at $200 million (Hannover Re, same perils), Harbor Crest Re at $100 million (Porch Group, a broad multi-peril book including wildfire and winter storm), and the LADWP's 123 Lights Re at $100 million specifically covering California wildfire. These are not marginal perils being pushed into the market at distressed structures — they are core sponsors accessing the market at deal sizes averaging $138 million.

Twelve Securis consolidating its cat-bond and private ILS teams is worth noting as an operational signal, not a distress signal. When a specialist ILS manager integrates its public and private strategies, it is optimizing for blended-portfolio efficiency, not retreating. That is what a disciplined market does with scale. The departure of Etienne Schwartz is a personnel event; the structural move is the tell.

I want to flag one thing to my colleague Dr. Chandrasekar's territory: Harbor Crest Re's $100 million note covers wildfire among its multi-peril basket. The Washington arson event — a suspect who explicitly timed ignition for a high-wind day — is a non-modeled ignition source that sits outside the standard fire-weather hazard curve. That is not a spread problem today; it becomes one at renewal if loss experience forces the market to reprice human-ignition probability upward.

At 5.71% insurance risk spread over 2.44% market-level expected loss, cat bonds remain more attractive carry than HY credit at current levels, and J.P. Morgan's discipline read is consistent with the recent deal flow — but wildfire multi-peril structures carry unpriced arson/human-ignition tail risk.

Bias flag — Treats cat risk as a tradeable spread; the comfort at 5.71% insurance risk spread over 2.44% market EL underweights model error in wildfire EL estimation and the tail scenario of trapped collateral if a major hurricane sequence hits in rapid succession.

The Cycle Margaret Ennis

Ninety-two deals and $18.9 billion of YTD cat-bond issuance is not a soft market — it is a hard market attracting capital so efficiently that the capital itself becomes the next soft market's seed corn. That is the cycle's iron logic, and J.P. Morgan's 'discipline' read is the part of this story that deserves the most skepticism from a renewal-cycle perspective. Discipline in ILS is always loudest at the top of issuance records. The question is not whether managers are behaving better than 2017–2018; the question is whether the pricing buffer between 9.46% market yield and 2.44% expected loss is wide enough to survive a sequence of active loss years before spreads compress under the weight of this capital.

Berkshire Hathaway's Q2 numbers are the reinsurance cycle's most telling data point this week. Overall re/insurance underwriting earnings fell year-over-year, and property volumes were declining — the Tokio Marine whole-account quota share deal was needed to prop the headline. Berkshire is not a distressed player; it is the most disciplined capital allocator in the business. When Berkshire's organic property volume is shrinking, it is telling you that the risk-adjusted return at the margin no longer meets its hurdle. That is a cycle signal, not a one-quarter anomaly.

The ALIRT findings on residual markets deserve to be read as a cycle-within-a-cycle story. The broader U.S. property market has returned to profitability — that is the headline. But the residual markets in California, Florida, Louisiana, and Texas are diverging. What ALIRT is describing is the classic hard-market sorting: the voluntary market has repriced and retreated, leaving the residual market holding the concentration risk the private market refused. When the broad market is profitable and the residual market is stressed, you are not in a clean hard market — you are in a market where the hard pricing has worked for the voluntary carriers but has pushed the most exposed policyholders into pools that may not be adequately capitalized for the next event sequence.

Soren's read on deal flow discipline is fair, but I would add a timing caveat: we are in August, mid-year issuance is running hot, and the Atlantic season has not yet delivered a major named-storm loss. The discipline the market is showing today is discipline in the absence of a real test. Mid-year cat bond structures that include named-storm peril — Matterhorn Re and 3264 Re both cover US and Canada named storm — will be the proof of the pudding. If we exit the 2026 hurricane season without a major loss, the capital that entered disciplined will exit undisciplined.

Berkshire's declining organic property volumes signal that risk-adjusted returns at the margin no longer clear its hurdle — a more reliable cycle tell than any ILS discipline narrative, which has yet to be tested by a major 2026 named-storm event.

Bias flag — Mean-reversion lens may miss structural regime shift: if climate non-stationarity has permanently shifted the EP curve for named storm and wildfire, the 'hard market sows the next soft market' logic breaks down — the capital returning may be insufficient for the new loss environment.

Modeled Loss Dr. Ravi Chandrasekar

The Washington State arson case reported by Insurance Journal is a peril-modeling problem masquerading as a criminal justice story. The suspect, a 37-year-old man, explicitly told a detective he planned the ignition to coincide with a 'high winds' day. This is human-actuated ignition timed to maximize fire-weather synergy — the very scenario that sits in the tail of the fire-weather hazard curve but is typically assigned a low ignition probability by standard wildfire models, which treat ignition as a stochastic process correlated with human activity density rather than deliberate optimization of fire-weather conditions.

The cat-bond market's California wildfire exposure is acutely relevant here. The LADWP's 123 Lights Re issuance — $100 million covering California wildfire, priced in July 2026 — and Harbor Crest Re's multi-peril basket both carry wildfire exposure. The expected-loss assumptions embedded in these structures are derived from event catalogs and fire-weather models that capture accidental ignition probability and lightning-driven fire starts. They do not capture the frequency distribution of arsonists who read forecast data and optimize timing. This is a model gap, not a catastrophic one, but it is the kind of secondary-peril complexity that causes actual losses to outrun modeled losses in a bad fire year.

I want to engage Soren's observation about Harbor Crest Re's wildfire exposure directly. He is right to flag it, but I would sharpen the concern: the issue is not just that arson is unmodeled — it is that in an environment where wildfire is already a structurally non-stationary peril (the historical event catalog understates future frequency and severity because of vegetation accumulation, urban-wildland interface expansion, and now confirmed arson optimization), the EL estimate for wildfire tranches is subject to upward revision from multiple directions simultaneously. The spread-over-EL looks comfortable until the EL moves.

The ALIRT report's finding that conditions are diverging across California, Florida, Louisiana, and Texas residual markets is consistent with what modeled secondary-peril exposure tells us: these four states are not exposed to the same peril mix. California is wildfire-dominant; Florida is named-storm and inland flooding; Louisiana sits at the intersection of hurricane, storm surge, and riverine flood; Texas blends named storm, severe convective storm, and hail. A single 'residual market stress' narrative obscures peril-specific model uncertainty. The EP curves for these four states have very different tails, and a stress event in one does not necessarily propagate to another — unless the capital pools are interconnected, which FAIR Plan and Citizens structures sometimes are.

The Washington arson event exposes a structural gap in wildfire cat models: human-optimized ignition timing is not captured in standard ignition-probability distributions, creating upward pressure on actual-vs-modeled loss ratios for wildfire-exposed cat-bond tranches precisely when the LADWP and multi-peril structures have just been issued.

Bias flag — Over-trusts the EP curve as the primary analytical frame; the Washington arson case is an apt self-critique, but social inflation and litigation-driven loss development in California and Florida are equally significant model gaps that receive less attention here.

Solvency Watch Eleanor Pryce

ALIRT's finding — that the broader U.S. property market has returned to profitability while residual markets in California, Florida, Louisiana, and Texas are diverging — is the week's most important regulatory signal, and it is not getting nearly enough attention in the context of insurer-of-last-resort adequacy. The pattern ALIRT is describing is structurally dangerous: when the voluntary market hardens and returns to profit, the political pressure to hold residual market rates flat intensifies, because regulators and legislators can point to voluntary-market profitability as evidence that the crisis is 'over.' It is precisely when that narrative takes hold that the residual markets become most vulnerable to the next major event.

Berkshire Hathaway's Q2 earnings are instructive from a balance-sheet perspective. Overall re/insurance underwriting earnings fell year-over-year, and the company explicitly cited the Tokio Marine whole-account quota share agreement as the mechanism that prevented a steeper headline decline in premium volumes amid lower property underwriting. This tells you something about where even the most capitalized reinsurer sees risk-adjusted value in the current market: not in direct property risk, but in structured quota-share arrangements that spread exposure. The implications for smaller carriers who cannot access that kind of counterparty or structure are significant.

The SEC filing data for the Insurance sector shows TRV (Travelers) at 47.2% Item 1A novelty and BRK-B at 45.4% — these are the two largest risk-language rewrites in the insurance cohort this cycle, behind only PRU at 66.8%. When Travelers is substantially rewriting its risk-factor disclosures, that is a signal worth tracking: Travelers is a bellwether for commercial and personal-lines property exposure. The direction of the rewrite (additions vs. deletions) is not specified in the corpus, but the volume — 246 sentences added, 251 deleted, net roughly flat — suggests a substantive repositioning of risk language rather than simple expansion. That warrants scrutiny by state regulators.

The four diverging residual markets deserve individual attention. California's FAIR Plan is exposed to wildfire — and the LADWP arson case is a reminder that the event catalog underpins every rate-adequacy analysis. Florida Citizens is exposed to named-storm concentration in a post-Ian, post-Idalia rate environment where legislative reform is still working its way through the system. Louisiana Citizens faces the cumulative overhang of multiple storm seasons and ongoing population and insurer-of-choice attrition. Texas TWIA sits in a hail and named-storm corridor with persistent funding questions. These are four different solvency problems, not one.

ALIRT's diverging residual-market findings across four states represent four distinct solvency risk profiles — not a single crisis — and the political incentive to declare the property insurance crisis 'over' because the voluntary market is profitable creates exactly the conditions for residual-market undercapitalization to go unaddressed.

Bias flag — Reads every residual-market divergence as impending insolvency risk; underweights the possibility that ALIRT's contrasting findings include some states where residual-market conditions are actually improving, and that recent legislative reforms in Florida may be working.

Protection Gap Daniela Owusu-Reyes

ALIRT's report says the broad U.S. property insurance market has returned to profitability. That is the voluntary market talking about itself. The people who live in California, Florida, Louisiana, and Texas — the four states where ALIRT flags contrasting and deteriorating residual-market conditions — are not in the voluntary market. They are in the residual market because the voluntary market left. Profitability in the voluntary market and stress in the residual market are not two separate stories; they are one story with a sharp distributional edge. The voluntary market's profit is partly a function of having shed its highest-risk, lowest-income policyholders into pools that are now under stress.

The $18.9 billion in YTD cat-bond issuance and the record $65.8 billion in outstanding ILS capital are abstractions from the perspective of a homeowner in Metairie, Louisiana or Compton, California who received a non-renewal notice. The capital is there — it just does not flow to the people who need it most, because the cat-bond market is a reinsurance-layer instrument, not a primary-coverage instrument. The protection gap is not a function of capital scarcity; it is a function of capital routing. The ILS market's discipline, which J.P. Morgan and Soren both find reassuring, is precisely the discipline of not taking primary-layer risk in the most exposed ZIP codes.

I want to engage Eleanor's framing directly: she is right that these are four different solvency problems. But I would add that they are four different protection-gap problems with a common driver — the voluntary market's rational risk-based withdrawal has been faster than state residual-market institutions can absorb the volume, recapitalize, and price adequately. The Washington arson case is not just a modeling problem for Dr. Chandrasekar; it is a coverage problem for the homeowners downwind of a fire that a standard peril map would never have predicted at that location. The protection gap is widest exactly where the models are most uncertain.

The Berkshire-Tokio Marine quota-share arrangement is worth noting from a consumer perspective: this is reinsurance capital flowing to a structured deal between two of the world's most sophisticated institutions, not to the primary market where coverage availability problems live. The reinsurance market's health does not automatically translate to coverage availability in high-risk U.S. zip codes. That transmission mechanism is broken, and ALIRT's findings are the evidence.

The voluntary property market's return to profitability is inseparable from its shedding of high-risk policyholders into residual pools that ALIRT now flags as diverging and stressed — the profit and the protection gap are two sides of the same balance sheet.

Bias flag — Frames every voluntary-market withdrawal as market failure; underweights that risk-based pricing and residual-market stress are partially the consequence of decades of subsidized below-risk-cost coverage that created moral hazard and concentration in high-risk zones.

Carrier Books Theo Marchetti

The equity-relevant read this week starts with AIG and Berkshire. AIG reported Q2 2026 net income of $948 million attributable to shareholders, down from $1.1 billion in the prior-year quarter — a year-over-year decline driven by changes in what the summary describes as unspecified items, but the headline is a miss on net income despite what the company characterized as higher underwriting profits. This is the reserve-development and investment-income tension that shows up in P&C earnings: you can post better underwriting while the net income line moves in the wrong direction if your investment portfolio, mark-to-market positions, or prior-year development is working against you. With the 10-year/2-year curve at 0.46 percentage points and effective fed funds at 3.63%, the fixed-income reinvestment tailwind for P&C carriers is not expanding meaningfully — it is plateauing.

Berkshire's story is the more structurally interesting one. The Tokio Marine quota-share propped the Q2 premium volume in what would otherwise have been a down quarter on property. Berkshire is choosing structured quota-share capital efficiency over organic property growth — that is a combined-ratio management decision as much as a capital-allocation one. The question for Berkshire's P&C book is whether the Tokio Marine arrangement creates reserve co-mingling complexity that shows up in future quarters as development noise. Whole-account quota shares are clean at inception; they get complicated when losses develop unevenly across lines.

The SEC 10-K filing novelty data for the insurance sector is the week's most underappreciated signal for equity analysts. PRU leads at 66.8% Item 1A novelty — that is life insurance risk language being substantially rewritten, not P&C. TRV at 47.2% and BRK-B at 45.4% are the ones with direct P&C implications. Travelers adding 246 sentences and deleting 251 in its risk factors section is a near-zero net change in volume but a near-50% novelty score, which means the content is being substantially reworked, not merely expanded. That kind of rewrite at a P&C bellwether, in a year when property cat exposure is the dominant risk topic, suggests legal and actuarial teams are repositioning disclosures around new peril exposures, litigation environment, or reserve adequacy — any of which could surface in future earnings. The VIX at 15.15 and HY OAS at 2.71% mean the equity market is not pricing tail risk for insurers right now. That is fine until it isn't.

AIG's Q2 net income declining to $948 million despite higher underwriting profits, combined with Travelers' near-50% risk-factor novelty score rewrite, suggests that the P&C earnings picture is more complex below the underwriting headline than current market pricing implies.

Bias flag — Over-indexes on the quarterly combined ratio and net income headline; the AIG analysis is limited by the corpus summary's lack of specificity on what drove the net income decline — the 'changes in unspecified items' framing is a genuine data gap that limits confidence.