Hartford released first quarter 2026 results after the close on April 23, with the call the following morning. Consensus had Personal Insurance at a 94.3 combined ratio against 106.1 in Q1 2025, Business Insurance at 91.3, and Group Benefits core earnings margin between 7.5 and 8.0 percent.

Twelve points of personal lines improvement in a year is the headline. Whether it is rate finally earning through or a light catastrophe quarter is a different question, and the two look identical in the ratio.

94.3
Personal Insurance combined ratio consensus for Q1 2026
-6.9%
Auto policies in force year over year
91.3
Business Insurance combined ratio consensus

Key Takeaways

  • 12 points of Personal Insurance improvement on consensus, from 106.1 to 94.3, in a year where Q1 2025 carried 7 to 9 points of catastrophe load for California-exposed personal books and Q1 2026 did not.
  • Auto policies in force down 6.9 percent year over year while average premium per policy is up 15 to 20 percent, so earned premium grows on a shrinking book.
  • Earned rate at 18 to 22 percent into Q1 2026 against cumulative written rate of 25 to 30 percent taken across 2023 through the first half of 2025. The gap is the six-to-nine-month earning lag.
  • Progressive printed 86.4 consolidated with 9 percent PIF growth on April 15. The two books are not comparable: one is restoring rate adequacy by shedding policies, the other already has it.
  • $325 million of after-tax favorable development at Travelers came with an explicit uncertainty provision on AY 2025 IBNR. Hartford has historically released $50 million to $100 million without that framing.

What the 12 Points Contain

The catastrophe comparison does real work in the year-over-year swing. Q1 2025 carried roughly 7 to 9 points of catastrophe load in Personal Insurance for carriers with meaningful California exposure. Q1 2026 ran lighter and, where it was heavy, it was heavy in March severe convective storm activity rather than wildfire.

Allstate pre-announced $1.24 billion of Q1 catastrophe losses, tracking toward 9 to 10 points on its own book, but concentrated in that March SCS distribution. Hartford carries less California exposure than Allstate and a lighter SCS-exposed homeowners geography, so the comparison year is unusually clean for it, which is what makes the underlying loss ratio the number worth reading rather than the combined ratio.

The policies-in-force line is the counterweight. Auto PIF fell about 6.9 percent year over year, with homeowners declining at a similar pace. Part of that is targeted non-renewal in wildfire and SCS-exposed territories; part is shopping behavior as rate increases pushed policyholders to compare. For a book in rate-adequacy recovery that is the intended outcome, not a failure of retention.

Which line drove the improvement decides how much signal the print carries. Personal auto severity and frequency are observable close to real time, so a clean quarter of underlying auto improvement is evidence that earned rate now exceeds earned loss trend. Homeowners improvement on catastrophe timing is not, because Q2 and Q3 carry the normalized provision.

Earned Rate, Shrinking Book, and What Flatters the Expense Ratio

A rate filing applies at the effective date to new and renewing policies, so existing policies keep earning at the old rate until renewal. On a six- or twelve-month auto book the earned impact runs six to nine months behind written.

Hartford took cumulative written rate of 25 to 30 percent across 2023, 2024 and the first half of 2025, weighted toward California, Florida and several Midwestern states where loss trend ran hottest. Earned rate into Q1 2026 should therefore be running at 18 to 22 percent, and that is the figure the combined ratio actually reflects. The remaining 7 or so points of written rate are still to earn, which is the margin available to absorb a normalized catastrophe load in the back half of the year.

The same arithmetic complicates the expense ratio, which Hartford is framing as tech-led on the back of agent-facing generative AI tools rolled out through 2024 and 2025. Average premium per policy is up 15 to 20 percent across personal auto and homeowners while PIF is down 6.9 percent on auto, so earned premium is higher year over year on fewer policies. A larger denominator against flat or falling fixed cost improves the expense ratio whether or not a single workflow was automated.

Expense ratio driverCharacterDurability
Agent-facing generative AI rolloutStructural productivityDurable if retention holds
Internal workflow copilots (claims, UW)Structural productivityDurable, adoption-dependent
Average premium growth (rate action earning in)Mix effectTemporary, reverses when rate moderates
Headcount avoidance against growthLeverage effectDurable while growth resumes
Direct FTE reductionCost-outDurable, one-time reset

That is not an argument against the tech claim. It is an argument that the tech contribution has to be quantified separately to be readable, because the mix effect reverses when rate moderates and the productivity effect does not. Naming the contribution in basis points is the disclosure that makes the two separable, and it is the bar peers then either match or explain.

The peer set is not directly comparable on the headline. Progressive reported an 86.4 consolidated combined ratio on April 15 with 9 percent PIF growth and a personal property combined ratio of 78.3 carrying 12.5 points of net catastrophe load from March SCS. That is a growing book executing on achieved rate adequacy against Hartford's shrinking book restoring it. Allstate reports April 29 into the same question with heavier SCS geography.

The Booking Posture the Ratio Does Not Reveal

A 94.3 combined ratio is consistent with a conservative current-accident-year loss pick and with a central one. Nothing in the ratio distinguishes them, and the difference decides what the rest of 2026 looks like.

Travelers set the reference point this quarter, booking $325 million of after-tax favorable prior-year development alongside an explicit provision for uncertainty held against AY 2025 IBNR, as covered in the Travelers reserve release analysis. Hartford has historically released prior-year reserves more quietly, in the $50 million to $100 million after-tax range, without that framing. A materially larger release would say Hartford has built the same programmed cushion on AY 2022 through AY 2024; the historical figure would say it has not.

The 2023 precedent is why the distinction matters here rather than as general practice. Hartford's personal auto book appeared to stabilize around a 100 to 102 combined ratio in 2023 before bodily injury severity re-accelerated and the ratio widened again, the pattern management described as rate catching trend and then getting caught. The 2024 and 2025 rate actions were sized against that shock and a higher baseline trend. A conservative Q1 pick is the mechanism that would absorb a second re-acceleration; a central one is not.

Group Benefits carries the same shape of risk on a slower clock. The group disability loss ratio has run in the 70s through 2024 and 2025, incidence ticked up in 2024 on what management attributed to post-pandemic normalization, and claim duration is the slow-moving variable most likely to carry social inflation signal. Hartford is one of the largest group long-term disability writers, so movement in its duration trend is a reasonable read on whether social inflation is extending past property-casualty lines.

The commercial book has its own version. General liability severity on AY 2022 through AY 2024 is live in reserve reviews across the market, and NCCI has flagged early adverse development on AY 2021 through AY 2024 workers compensation in a handful of states. Workers compensation has carried the favorable half of every major commercial carrier's development for years. If that contribution thins while general liability strengthens, a 91.3 Business Insurance print becomes considerably harder to repeat.

Further Reading

Sources