ALL — Q2 2026 Financial Report Analysis

Lighter storms and a $692 million reserve release, not a better book, drove Allstate's Q2

Allstate earned $3.24 billion for common shareholders in the second quarter of 2026, up 55.9% from $2.08 billion a year earlier, on a combined ratio of 86.6% — the strongest second-quarter underwriting result the company has posted in years. Almost all of the improvement came from two items that are, by their nature, not repeatable on demand: catastrophe losses ran $268 million lighter than last year and well below Allstate's own ten-year seasonal norm, and the company released $692 million of reserves it had set aside for accidents in prior years, nearly double the $376 million released a year ago. Strip both out and the underlying book performed about the same as it did in Q2 2025.

For readers new to insurance accounting: the combined ratio is the industry's core profitability measure — claims plus expenses expressed as a percentage of the premiums earned. Below 100% means the insurer made money on insurance itself, before counting anything it earns on its investment portfolio. Catastrophe losses are claims from events the company designates as catastrophes (hail, tornadoes, wildfires, hurricanes); they are lumpy quarter to quarter and are what makes a home insurer's earnings volatile.

Q2 2026 at a glance

Takeaway: Adjusting the combined ratio for both catastrophes and prior-year reserve movements, Allstate's underwriting result was roughly 79.7% this quarter against 79.8% a year ago — effectively flat. The entire 4.5-point headline improvement is explained by lighter weather and a larger reserve release, while the company simultaneously cut auto prices (Allstate brand average auto premium fell 3.6% to $819) and raised advertising spend 18.6% to buy policy growth. That is a deliberate trade of current margin for future volume, and it means the reported 86.6% combined ratio overstates the run-rate earning power of the book.

Decomposing the $723 million underwriting gain

Allstate Protection's underwriting income rose $723 million year over year, to $2.01 billion. Management's own explanation is "an increase in premiums earned, lower catastrophe losses and the benefit of prior year reserve releases, partially offset by higher expenses." The arithmetic bears that out, and shows how concentrated the gain is:

• Lower catastrophes: ~$268 million. Total catastrophe losses fell to $1.72 billion from $1.99 billion. The filing discloses that the ten-year average effect of catastrophes on the second-quarter combined ratio is 13.5 points; this quarter it was 11.5 points. Weather was better than normal, not just better than last year.
• Larger reserve releases: ~$271 million. Total prior-year reserve reestimates released $641 million versus $370 million in Q2 2025. Excluding catastrophe-related reestimates, releases were $692 million against $376 million.
• The remainder — roughly $184 million — comes from $572 million of additional earned premium net of $263 million of higher underwriting expenses.

A reserve release happens when claims from earlier years settle for less than the insurer originally estimated, and the surplus flows straight into current-quarter profit. Allstate's release was overwhelmingly auto injury: $597 million of the quarter's $648 million auto release related to injury coverages, with approximately 51% traced to accident year 2025 and 33% to accident years 2023 and 2024. Management attributes it to "improved prior period loss development and better than expected claim outcomes." This is genuine good news about how the 2023–2025 accident years are settling — but it is earnings recognized from past underwriting, not evidence that policies written today are priced better.

Auto: growing units by giving back price, while severity still rises

Auto is the tension in this quarter. Premiums written were essentially flat at $9.57 billion, up just $39 million (+0.4%), because two forces offset: policies in force grew 2.8% (up 708 thousand to 25.95 million) and new issued applications rose 8.8% to 2.35 million — with the direct channel up 12.6% — while the Allstate brand average auto premium fell 3.6% to $819. The filing attributes the lower average premium to "a shift in product mix towards affordable, simple and connected protection," i.e. cheaper products, not purely rate cuts.

Meanwhile costs are not falling. The MD&A states that "estimated report year 2026 incurred claim severity for Allstate brand increased compared to report year 2025 for major coverages, reflecting ongoing inflationary pressures, including rising medical costs and continued increases in attorney representation." Severity is the average cost per claim.

The reported auto combined ratio improved to 83.3% from 86.0%. But 4.3 points of that 83.3% is prior-year reserve release (against 3.0 points last year). Neutralize the releases and auto sits near 87.6% versus about 89.0% — a real but far more modest improvement, achieved while unit prices fall and claim costs rise. Management's stated posture is to "pursue rate adjustments in states where we are achieving acceptable returns, while implementing rates where needed to keep pace with increasing costs."

Homeowners: a swing to profit that weather paid for

Homeowners flipped to $226 million of underwriting income from a $76 million loss, with the combined ratio at 94.6% versus 102.0%. Premiums written grew 8.1% to $4.75 billion on a 5.8% higher average premium ($2,399) and 2.9% policy growth to 7.82 million.

The caveat is in the catastrophe line. Catastrophes contributed 33.5 points to the homeowners loss ratio this quarter, down from 42.8 points — but the filing discloses a ten-year average of 43.5 points for the second quarter. Homeowners catastrophe load ran roughly ten points below its own decade norm. On a normalized weather assumption, homeowners would still be running near or above breakeven on underwriting. The non-catastrophe picture is also not uniformly improving: gross claim frequency excluding catastrophes rose, and paid severity excluding catastrophes rose "primarily due to fire perils."

Catastrophe activity itself was ordinary in composition — 41 wind/hail events costing $1.66 billion, with wildfires immaterial at $30 million. The year-over-year six-month comparison is distorted by the January 2025 California wildfires, which cost $1.11 billion and account for most of the $1.23 billion decline in first-half catastrophe losses. Allstate notes it is pursuing subrogation recoveries on those wildfires, has recognized nothing for them, and expects any recovery "to primarily benefit the Company's reinsurers."

The expense ratio is going the wrong way — on purpose

The expense ratio rose 1.0 point to 21.8%, which management attributes to "higher advertising and legal expenses, partially offset by higher earned premium growth relative to costs." Advertising alone was $524 million, up $82 million or 18.6%, adding 0.4 points to the ratio. Read alongside the 8.8% rise in auto applications and the 16.4% rise in homeowners applications, this is spending to convert a soft-pricing posture into unit growth. It is a coherent strategy; it is also why the underlying margin is flat rather than expanding.

Investments, and the quarter's non-operating tailwind

Net investment income rose 33.8% to $1.01 billion. The split matters: market-based income rose to $837 million from $733 million on higher average invested balances, while performance-based income (private equity, real estate, infrastructure) more than doubled to $255 million from $90 million on "higher real estate and private equity valuation increases." The filing adds a concentration warning worth heeding — "the top 10 investments contributing approximately 88% of performance-based income, while the broader portfolio generated modest returns" — and notes these valuations are recorded on a lag, primarily reflecting the quarter ended March 31, 2026.

Separately, and outside insurance operations, Allstate booked $1.06 billion of pre-tax net gains on investments and derivatives ($829 million after tax), against a $144 million loss last year, "primarily related to valuation gains on equity investments." That single mark-to-market swing is roughly a quarter of net income to common and will reverse if equity markets do. Working the other way, the prior-year quarter was itself flattered by an $890 million gain on disposition of operations from the 2025 sale of the employer voluntary benefits and group health businesses — so the 55.9% earnings growth is measured against an already-inflated base, which makes the underlying comparison better than the headline, not worse.

Capital returns

Allstate repurchased 8 million shares — 3.0% of shares outstanding at year-end 2025 — for $1.66 billion in the first half, leaving $2.60 billion on the $4.00 billion authorization approved on February 4, 2026 and running to February 2028. Diluted share count fell to 259.1 million from 267.9 million, which is why EPS grew 61.2% against 55.9% net income growth. Common and preferred dividends absorbed $541 million and $59 million respectively in the half. Book value per diluted share reached $123.38, up 49.7% year over year, and trailing-twelve-month return on average common equity was 49.1%.

What to watch from here

Pricing and growth. Management gave no formal guidance in the 10-Q, but the stated approach is explicit: rate adjustments where returns are acceptable, rate increases where costs demand them. In homeowners, 4.9% of rate increases were implemented in the first half for a 1.3% estimated premium impact excluding replacement-cost inflation. In auto the direction is the opposite — average premium down 3.6% — so the near-term question is whether 2.8% auto unit growth and 8.8% application growth are being bought at prices that hold up against the rising injury severity the filing itself flags.

Geographic constraints. Allstate is not writing new homeowners business in Florida and completed non-renewal of certain policies there during the quarter; California new business flows only through its excess-and-surplus carrier, North Light. The company states plainly that it "may not be able to grow in certain states without regulatory or legislative reforms." Two of the largest US property markets are therefore closed or near-closed to growth, which caps how far homeowners premium can scale even with pricing power elsewhere.

Catastrophe exposure and reinsurance economics. The modeled 1-in-100 annual aggregate probable maximum loss for hurricane, earthquake and wildfire was approximately $3.2 billion net of reinsurance at June 30. The cost of that protection rose to $378 million in the quarter from $305 million, and to $686 million from $562 million for the half — a 22% increase in the half-year cost of ceding catastrophe risk, roughly 83% of which reduces homeowners premium. Q2 is also pre-hurricane-season; the third quarter carries the heavier named-storm exposure.

Reserve adequacy. Two straight quarters of large auto injury releases — $1.27 billion for injury coverages in the first half — indicate reserves set in 2023–2025 were conservative. That is a favorable signal on balance-sheet quality. It is also a finite source: the more the 2023–2025 years are released, the less cushion remains, and the more future quarters must be carried by current-accident-year margin.

My read. The balance sheet and capital position are in good shape: book value per share up nearly 50%, buybacks running at a 3%-per-half pace, and reserves developing favorably. The earnings quality is the softer part of the story. Of the 4.5-point combined-ratio improvement, essentially all of it traces to weather that ran two points below the ten-year norm and a reserve release that nearly doubled. Underwriting margin on business written today is flat, deliberately so, while Allstate spends on advertising and trims auto price to add units. That trade should look good if unit growth persists and auto severity inflation moderates; it will look expensive if severity keeps climbing — and the MD&A says it currently is. A normal-weather third quarter with a smaller reserve release would likely show a combined ratio several points higher than 86.6%, and that would be the more honest picture of where this book stands.