Mercury General's second-quarter 2026 combined ratio improved to 89.9% from 92.5% a year earlier and net income rose 58.3% to $263.5 million, but the release also disclosed that net incurred losses on the January 2025 Palisades and Eaton wildfires had climbed roughly $80 million over the first half of 2026 (Mercury General, August 2026). Eighteen months is a long tail for a single-peril reserve item still to be moving, and the movement here ran adverse, not favorable, in the same quarter the company's personal auto book posted about $35 million of favorable prior-year development.

Two reserve stories sit inside one earnings release, and only one of them made the headline. Mercury's stock-friendly numbers, the 89.9% combined ratio, the 58.3% net income gain, the book value climbing to $51.20 a share from $43.64 at year-end, are real and were widely reported (Mercury General, August 2026). Less widely reported is what the Form 10-Q filed the same week shows about the wildfire reserve specifically: a line item that grew rather than shrank across the first six months of 2026, financed by subrogation recoveries that are themselves still developing against a utility whose equipment regulators had, six days earlier, formally named as the cause of one of the two fires.

What the Loss-Reserve Rollforward Shows

Mercury's earnings release attributes the figure loosely to "2026 catastrophe losses" without separating the three months ended June 30 from the six months ended June 30 (Mercury General, August 2026). The Form 10-Q is more precise, and it is a six-month figure, not a single-quarter one. Note 10's loss-reserve rollforward for the two fires shows gross losses and loss adjustment expenses rising from $2.19 billion at December 31, 2025 to $2.29 billion at June 30, 2026, a $102.0 million increase (Mercury General, Form 10-Q, August 2026). Subrogation recoverable grew alongside it, up $24.4 million on the Eaton fire and $2.3 million on the Palisades fire, offsetting roughly a quarter of the gross deterioration. Reinsurance recoverable did not move at all: the catastrophe treaty covering both fires was already fully exhausted at $1.29 billion before the six-month period began, so every additional dollar of gross loss now flows straight through to the net line rather than being absorbed by a reinsurer (Mercury General, Form 10-Q, August 2026).

Net losses before Mercury's FAIR Plan share rose $75.2 million, from $312.7 million to $388.0 million. Add a $4.6 million increase in the net FAIR Plan charge and the combined net figure for both fires moved from $380.4 million to $460.2 million, a $79.8 million increase that the earnings release rounds to $80 million (Mercury General, Form 10-Q, August 2026). That reconciliation matters for anyone reading the press release literally: the $80 million did not happen inside the $75 million of catastrophe losses the company booked in the second quarter alone. It happened over the half, with an exhausted reinsurance program ensuring the increase landed entirely on Mercury's own balance sheet.

Component (Eaton + Palisades)Dec 31, 2025Jun 30, 2026Six-Month Change
Gross losses and LAE$2,191.8M$2,293.7M+$102.0M
Eaton fire subrogation recoverable$(537.5M)$(561.9M)$(24.4M)
Palisades fire subrogation recoverable$(48.0M)$(50.4M)$(2.3M)
Reinsurance recoverable$(1,293.5M)$(1,293.5M)$0
Net losses before FAIR Plan$312.7M$388.0M+$75.2M
Net FAIR Plan losses$67.7M$72.3M+$4.6M
Total net losses and LAE$380.4M$460.2M+$79.8M

The subrogation side of that ledger is large and mostly one counterparty. Mercury has booked $561.9 million of estimated subrogation recoverable against Southern California Edison for the Eaton fire, which the company states is approximately 55% of its own estimated ultimate loss on that fire (Mercury General, Form 10-Q, August 2026). On the Palisades fire, Mercury sold its subrogation rights outright in June 2025 for a guaranteed recovery of about $50 million plus an upside share, of which $34 million had been received by June 30, 2026, with the remaining $16 million settling in installments (Mercury General, Form 10-Q, August 2026).

The Cause Finding Landed the Same Week

Six days before Mercury's earnings release, the Los Angeles County Fire Department's arson investigation unit and Cal Fire concluded an 18-month investigation into the Eaton fire's origin. Electrical arcing on an out-of-service SCE transmission tower above Eaton Wash ejected hot metal particles into dry vegetation below, igniting the blaze that killed 19 people and destroyed more than 9,000 homes across Altadena and Pasadena (Los Angeles County Fire Department and Cal Fire, reported by CBS News Los Angeles, August 4, 2026). "While the cause has been determined, our focus remains on assisting our residents to rebuild their homes and livelihoods," LA County Fire Chief Anthony Marrone said the day the report was released (Anthony Marrone, CBS News Los Angeles, August 4, 2026).

Edison had already conceded the substance of that finding before the official report landed. Edison International disclosed in February 2026 that SCE had settled two subrogation claims from insurance companies at an average of 55% of losses incurred, following a September 2025 negotiated agreement with one insurer to pay 52% of losses without litigation (Mercury General, Form 10-Q, citing Edison International, August 2026). Those settlement percentages are the direct reference point for the roughly 55% recovery rate Mercury has booked against its own $561.9 million Eaton subrogation asset. The official cause determination does not by itself change Mercury's balance sheet, since the company had already recorded the recoverable based on the weight of prior evidence and SCE's own admissions, but it removes one of the larger residual uncertainties, whether a final liability determination would support the recovery rate already embedded in the reserve.

Why a Wildfire Reserve Still Moves After 18 Months

Property catastrophe claims from wind and hail typically close within a year. A mega-fire in a dense, litigious California market does not, and four mechanisms explain why the Palisades and Eaton reserve keeps developing well past the point most catastrophe losses have settled. Litigation timing is the largest: more than 10,000 plaintiffs have sued SCE alleging inadequate infrastructure maintenance, and the subrogation recovery an insurer can book depends on how those cases and Edison's parallel compensation program resolve, not on when the fire itself went out. Smoke and contents claims develop more slowly than total-loss structure claims, because policyholders continue discovering smoke damage to structures that survived and personal property inventories take months to itemize completely. Subrogation timing risk compounds the two: Mercury's own Eaton recoverable grew $24.4 million in six months, evidence the recovery estimate is still being refined against SCE's claims-payment pace and the case-by-case settlement percentages it is negotiating. Rebuild-cost inflation is the fourth mechanism, distinct from claim counts or coverage disputes: California's post-fire labor and materials market has run hot enough that per-claim rebuild-cost estimates set in early 2025 have needed repeated updating as actual contractor bids come in above the original case reserves.

A Split-Screen Quarter

Two reserve stories ran through Mercury's second quarter at once. Personal auto, the company's largest line, produced approximately $35 million of favorable development, primarily lower-than-estimated loss and loss adjustment expenses (Mercury General, August 2026), extending a run of auto releases that has offset property-line volatility across much of the sector this year, a pattern actuary.info has also tracked at Allstate's own Q2 2026 auto reserve release. Homeowners moved the other direction, with the release stating that year-to-date favorable development was "partially offset by adverse development on the homeowners line of insurance business" (Mercury General, August 2026), a category that captures both the wildfire tail and ordinary homeowners severity trend.

The net effect on Mercury's headline ratios was still favorable. The loss ratio improved to 65.0% from 68.8%, and the expense ratio rose to 24.9% from 23.7%, for a combined ratio of 89.9%, down 2.6 points from 92.5% a year earlier (Mercury General, August 2026). Net premiums earned grew 9.6% to $1.50 billion, statutory surplus stood at $2.77 billion, and Mercury secured a 6.9% California homeowners rate increase effective July 2026 covering more than 650,000 policyholders (Insurance Business America, August 2026). None of that growth was in question this quarter. What the results show is that a carrier can post genuinely strong headline metrics while a specific mega-cat reserve line inside those results keeps moving the wrong way, a distinction a reserve reviewer is trained to isolate and a headline combined ratio is not built to surface.

What the Adverse Move Implies About the Original Picks

Mercury is not an outlier in still refining its wildfire numbers a year and a half out; it is a data point in a pattern visible at the industry level too. Verisk's initial modeled estimate, published within weeks of the fires, put combined industry insured losses for Palisades and Eaton at $28 billion to $35 billion, with CoreLogic's independent estimate running higher, at $35 billion to $45 billion (Verisk, January 2025). Those were catastrophe-model point estimates produced before a single claim had been fully adjusted. Fourteen months later, the California Department of Insurance's claims data as of March 2026 showed $21.97 billion in reported insured losses across both fires, Palisades at $14.58 billion and Eaton at $7.39 billion (California Department of Insurance, March 2026), still climbing as claims continued to develop and still below even the low end of Verisk's original range. Mercury's own $80 million half-year swing is a small piece of the same story: initial loss picks on a fire of this scale get revised as litigation, subrogation and rebuild costs resolve in ways a catastrophe model cannot fully anticipate at the moment of loss.

The FAIR Plan mechanism compounds the reserving problem for every California carrier with exposure, not only Mercury. Mercury's own share of FAIR Plan assessed losses from the two fires has reached approximately $97 million, of which California regulators approved recoupment of $25 million through a policyholder surcharge, leaving a net $72 million charge still running through the company's results (Mercury General, Form 10-Q, August 2026). Actuary.info has previously covered how the FAIR Plan's assessment and recoupment surcharge mechanism pushes a share of catastrophic loss cost back onto admitted-market carriers regardless of their own direct exposure to the triggering event, and Mercury's Note 10 disclosure is a concrete illustration of how that mechanism keeps moving through a single carrier's reserve line eighteen months after a fire most of the market has stopped covering as breaking news.

The Reserving Read-Through

For an actuary setting loss-development-factor selections on a book with meaningful California wildfire exposure, the Mercury disclosure argues against treating an eighteen-month-old mega-cat reserve as settled simply because it has aged past a typical property catastrophe development window. Subrogation-heavy claims carry a distinct development pattern: the net reserve depends not only on gross loss estimation but on a counterparty's ability and willingness to pay, which in this case is tied to SCE's own claims-payment pace, litigation posture and the liability determinations working through more than 10,000 individual plaintiff suits. A subrogation recovery assumption embedded in a reserve is, in effect, a credit-risk and litigation-timing assumption wearing a loss-reserve label, and it deserves the same scrutiny a reserving actuary would apply to a large reinsurance recoverable from a counterparty whose payment obligation is still being litigated.

Carriers with wildfire exposure reporting results over the next several quarters are worth watching for the same signal Mercury's own homeowners line already shows: favorable development on frequency-driven lines like personal auto funding, or masking, continued adverse development on severity-driven property catastrophe lines tied to the same 2025 events. Whether the $80 million six-month swing represents the last major correction to the Palisades and Eaton reserve or another interim data point in a longer development pattern is not something the current disclosure answers. That question, not the improved combined ratio, is the one a reserve opinion on this book would need to resolve before calling the January 2025 wildfire reserve adequate.

Further Reading

Sources