P&C Claims Severity Faces a Four-Factor Compounding Problem in 2026

Four severity drivers are each well documented on their own for the 2026 accident year: tariff-inflated parts, social inflation, ADAS repair complexity and construction cost escalation. Each carries its own published figure and its own trend adjustment.

The number that does not exist anywhere is what they do together on one claim. Standard severity trend selections treat them as additive, and the total loss threshold is where that assumption fails.

Key Takeaways

• 2.7% collision repair cost increase from the 25% imported parts tariff, per APCIA modeling, translating to $3.4 billion of additional personal auto premium and roughly $80 to $100 per repairable estimate.
• 23.1% total loss frequency, a record, alongside repairable appraisals of $2,000 or less falling from 41.5% of claims in 2019 to 25.5% by mid-2025.
• 57% inflation in US liability claims over the past decade, with over $71 billion of nuclear verdicts awarded between 2023 and 2025.
• 10.7% additive versus 12% to 14% compounded is the gap the interaction terms produce, a 1.3 to 3.3 point difference before it is carried across two or three accident years.
• 61% of vehicles need calibration and 35.6% of estimates capture it, so a large share arrives as late supplements at maturities where development factors assume stability.

Four Drivers, Four Published Numbers

Each factor is measurable in isolation, which is why each has its own trend component.

• Tariffs. A 25% duty on imported auto parts, with 25% on steel and 50% on aluminum against 80% of automotive-grade aluminum sourced from Canada. APCIA models a 2.7% collision repair cost increase, roughly $80 to $100 per repairable estimate on a baseline of 13.5 parts per repair order, with about 44% of collision OEM parts produced outside the US.
• Social inflation. Swiss Re's index puts US liability claim inflation at 57% over the decade, averaging 5.4% annually from 2017 through 2022 and peaking near 7% in 2023, with over $71 billion of nuclear verdicts from 2023 through 2025.
• ADAS. CCC's 2026 Crash Course records total loss frequency at 23.1%, bodily injury severity up 10.3% year over year, and 28.3% of repairable estimates carrying at least one calibration line at roughly $500 each.
• Construction. Construction Analytics shows steel mill products up 10%, aluminum 14%, copper 8% and lumber 7% between December 2025 and April 2026, against construction unemployment below 4% through four consecutive peak seasons.

Each of those is a defensible trend component. Applied separately, they are also mutually consistent, which is what makes the additive treatment look reasonable.

The Threshold Is Where Addition Breaks

The drivers do not sit side by side on a claim. They interact through the total loss decision, and that interaction changes the population being averaged.

Take an eight-year-old ADAS-equipped vehicle needing a bumper, windshield and front radar module. Tariff-inflated parts add $80 to $100, calibration adds roughly $500, and the vehicle is already near the economic repair boundary because the fleet has aged: CCC counts 12 million fewer vehicles six years old or newer in Q3 2025 than in 2020, with vehicles aged 7 to 12 years now 41% of valuations against 33.4%. A repair that would have run $4,800 before tariffs and calibration reaches $5,400, crossing a 75% threshold on a $7,000 actual cash value and becoming a total loss.

That is not an additive cost increase. It removes a claim from the repairable pool, and it removes it from the cheap end. Repairable appraisals of $2,000 or less have already fallen from 41.5% of claims in 2019 to 25.5% by mid-2025, which is the same mechanism observed at scale: average repairable severity rises partly because the inexpensive repairs have stopped being repairs.

Run the arithmetic both ways. Tariffs at 2.7%, ADAS mix shift at 3% and social inflation at 5% on liability components give a 10.7% additive severity trend. But if tariff-inflated parts push a further 2% of vehicles into total loss, the surviving repairable claims carry a higher average before the other two effects apply at all, and the compounded result lands nearer 12% to 14%. The 1.3 to 3.3 point gap is modest in one year and is not a one-year quantity: carried across two or three accident years in a development selection, it is a reserve shortfall for any carrier trending additively.

Property runs the same structure through a different threshold. Steel up 10%, aluminum up 14% and skilled trades up 4% to 11% raise the direct rebuild cost, while labor scarcity extends the timeline, which extends additional living expense on homeowners and business income on commercial property. The second effect is a function of the first, not an addition to it.

The Severity Arrives Where the Factors Assume Stability

The compounding would be tractable if it emerged at first report. Much of it does not.

Revv's benchmarking finds 61% of vehicles arriving for collision repair require some form of ADAS calibration, while 35.6% of estimates capture one. The difference is calibration that is missed at initial estimating and added later through supplements, which means a material share of ADAS severity enters the triangle at development points where loss development factor selections assume the case estimate has settled. Diagnostic scans now appear on 87.7% of estimates and calibration capture on DRP estimates rose 8.7 percentage points year over year, so the capture rate is improving, which makes the historical relationship between early and late development itself non-stationary.

Liability carries a longer version of the same problem. ADAS-related lawsuits went from 3 cases in 2018 to 61 in 2024 with settlements from $200,000 to over $1 million, a tail that most pricing models do not yet contain, and third-party litigation funding is forecast to exceed $30 billion in the US. Swiss Re's behavioural work found only 56% of respondents now think there are too many lawsuits, down from 90% in 2016, while 76% say damages are too low or about right against 58% eight years earlier. Those are jury pool inputs, and they move slowly in one direction.

The timing is what makes it consequential. AM Best projects the P&C combined ratio rising 1.9 points to 96.9 in 2026, citing materials-driven claims cost and casualty reserve development together. The trend gap and the tail arriving late are landing in the same accident years that a softening rate environment is pricing, which means the correction, when it appears, appears against premium already set.