Tariff-Driven Severity Has Entered P&C Triangles: How to Find It

Auto parts prices rose 6.0% year-over-year in 2025, with June running at 5.0% and July at 6.6% as pre-tariff inventory supplies were exhausted (CCC Crash Course 2026). That increase entered loss triangles not as a gradual trend ramp that averages smoothly across accident years, but as a hard break in the 2025 accident year diagonal -- a step-change produced by tariff effective dates that are known, datable, and structurally different from the underlying severity trend. The diagnostic is in the 12-to-24 month column of the accident year 2025 diagonal. The treatment is not to weight it away.

The Hard Event Dates

A 25% tariff on assembled vehicles took effect April 3, 2025, under Section 232 of the Trade Expansion Act. A second tranche covering 150 auto parts categories followed on May 3. Before either date, suppliers and distributors ran down pre-tariff inventory, temporarily suppressing cost signals in Q1 2025 data. By Q3, the buffer was gone. CCC's Crash Course 2026 documents the inflection precisely: average part prices increased 5.0% in June and 6.6% in July 2025, with the acceleration tied explicitly to post-depletion purchasing at tariff-inclusive prices.

This is not the kind of severity shift that averages quietly into a multi-year LDF selection. The average tariff rate on U.S. imports jumped from 2.6% before 2025 to approximately 13% to 14% by mid-year (CCC Crash Course 2026). A change of that magnitude in input cost, concentrated in a specific calendar window rather than distributed evenly across accident years, produces exactly the kind of structural break that Berquist-Sherman adjustments were designed to address when changes in operational processes alter the development pattern. The analogy holds: just as a shift in case reserve adequacy distorts the link ratios in prior accident years, a tariff effective date distorts link ratios by creating a step in severity that the averaging of historical data will dilute rather than reflect.

The American Property Casualty Insurance Association quantifies the claims cost consequence: tariffs on parts from China, Mexico, and Canada could increase personal auto claims costs by $26 to $52 billion annually (APCIA). Approximately 60% of U.S. replacement parts are sourced from those three countries. The range in the APCIA estimate reflects policy uncertainty about which tariff rates hold through the accident year development window -- a variable actuaries cannot resolve with historical data alone, but can treat through scenario-weighted LDF selections.

Physical Damage: The Clearest Diagnostic

Auto physical damage is the right line to examine first. APD claims close fast -- most within 12 to 18 months -- so the 2025 accident year is substantially developed by year-end 2025 valuations. The exposure is direct: repair estimates reference parts prices at the time the invoice is written, not at policy inception, so tariff-elevated costs appear immediately in paid amounts without a case reserve lag.

The development factor to scrutinize is the 12-to-24 month link ratio in the accident year 2025 diagonal. On a calendar-year paid triangle, the 12-month evaluation of the 2025 accident year captures claims closed through December 2025. Because the heaviest tariff cost entered paid invoices beginning in July 2025, claims with accident dates in Q3 and Q4 carry tariff-elevated severity in the 12-month evaluation. Comparing that factor against the 12-to-24 link ratios for accident years 2022, 2023, and 2024 should reveal a step up that is not explained by claim count mix, LAE allocation, or case reserve development. It reflects what claims actually cost to repair in a post-tariff parts environment.

The headline severity metric from CCC requires careful reading here. The average total cost of repair reached $4,818 in 2025, up only 1.7% year-over-year -- the lowest annual increase since 2017 (CCC Crash Course 2026). That subdued headline understates the actual severity pressure on the book, and understanding why is central to the triangle diagnostic.

Why the Average Total Cost of Repair Understates the Step

Parts per repair fell from 13.6 in 2024 to 13.0 in 2025 (CCC Crash Course 2026), a 4.4% reduction. Individual repair estimates used fewer parts on average, which suppresses total repair cost arithmetically even as the price per part rose 6.0%. The explanation sits in the total-loss rate.

Total-loss frequency hit a record 23.1% in 2025 (CCC Crash Course 2026). When tariff-elevated parts prices push a repair estimate past the economic total-loss threshold -- typically 70% to 80% of actual cash value -- the carrier pays ACV rather than repair cost and removes that claim from the repairable pool. The cheapest repairable claims nearest the threshold become total losses. The surviving repairable pool shifts toward complex or low-ACV vehicles where the economics still favor repair, suppressing the average TCOR while the per-part cost that triggered the shift continues running at 6.0%.

The actuarial implication: the industry average TCOR is not the right severity metric for triangle diagnostic work because it blends the mix shift. The correct input is severity per paid repairable estimate at each development point, isolated from the total-loss component. An LDF selection anchored to average paid amounts per claim -- blending repairable and total-loss payments across a distribution that shifted materially in 2025 -- will underweight the actual cost environment for the repairable segment that drives development at 24 and 36 months.

Homeowners and Commercial Property: The Slower Signal

The tariff transmission into property lines is real but develops across a longer window. Steel mill products rose 20.7% year-over-year and aluminum mill shapes increased 33% by 2026 (Associated General Contractors). Canadian softwood lumber carries a separate 35.2% duty. These are not minor adjustments; they affect the per-unit reconstruction cost of every roofing, framing, and exterior repair job that closes in 2025 and 2026.

For homeowners and commercial property, the triangle columns to examine are different. Property claims develop over 24 to 48 months -- longer for large commercial losses or subrogation pursuits. Tariff costs enter the triangle in the first paid development column for 2025 accident year claims, but the full signal spreads across the 12-to-24 and 24-to-36 development periods. That makes the diagnostic noisier than APD: catastrophe losses, construction labor shortages, and localized material supply disruptions compound the tariff effect in the same columns. Separating the tariff component requires an external cost index approach rather than pure factor analysis.

The cleaner read for property lines is replacement cost valuation at time of loss versus insured value at time of policy inception. Policies written in 2024 on pre-tariff reconstruction cost indices are closing on tariff-elevated material costs. When reconstruction costs run 20% above the values on which the policy was written, the carrier faces higher-severity claims before development factors are even applied. That coverage adequacy gap shows up as a loss ratio drift that development-only triangle methods attribute to adverse development when it is more accurately described as underinsurance at original valuation.

Where Chain-Ladder and Bornhuetter-Ferguson Break

The chain-ladder development method selects age-to-age factors by averaging historical link ratios, typically over three to five accident years. A three-year weighted average of the 12-to-24 factor for accident years 2023, 2024, and 2025 assigns roughly one-third weight to the 2023 pre-tariff environment, one-third to 2024, and one-third to the 2025 tariff-regime diagonal. If the 2025 link ratio runs 5% above its predecessors because of parts cost step-change, the three-year weighted average understates the expected 12-to-24 development for current and future accident years by approximately 1.6 points -- not because the method is wrong, but because it was designed for stationary processes and this one is not stationary.

The arithmetic scales quickly. For a carrier with $200 million of APD estimated ultimate at 12 months, a 1.6-point LDF understatement translates to roughly $3.2 million of understated IBNR for the 2025 accident year before any consideration of future accident years that will also close at tariff-elevated costs. Across a multi-line book where personal auto APD, commercial auto physical damage, and homeowners property all carry tariff exposure, the cumulative understatement is material enough to warrant explicit commentary rather than a note-and-carry-forward.

The Bornhuetter-Ferguson method has a structural advantage here, but only conditionally. BF is more resilient to an outlier diagonal because the a priori loss ratio anchors the IBNR estimate independently of the emerging development pattern. That advantage evaporates if the a priori loss ratio itself was developed from 2023 or 2024 experience, before the tariff effective dates. An a priori anchored to pre-tariff on-level loss ratios carries exactly the same understatement as the chain-ladder through a different mechanism.

Progressive's President and CEO Susan Griffith named the directional risk on the company's Q4 2024 earnings call, before the April 2025 effective dates: tariffs represent "one-directional risk to loss costs." The asymmetry she identified is the actuarial point. For policies priced before April 2025 that are still in development, the expected severity direction is up. A symmetric LDF selection that weights pre-tariff and post-tariff experience equally does not reflect that asymmetry and will produce IBNR estimates that are systematically low until the 2025 accident year is fully developed.

Separating Step from Trend in the Factor Plot

The diagnostic test is to plot age-to-age link ratios by accident year rather than averaging them. A genuine trend -- social inflation, ADAS calibration adoption, medical cost growth -- produces a roughly linear or geometric increase across accident years: the link ratio at accident year 2020 is lower than at 2021, lower than 2022, and so on, with no sharp discontinuity. A step-change produced by a discrete tariff effective date produces a different picture: stable link ratios for 2022, 2023, and 2024, then a discrete jump at 2025 that is disproportionate to the year-over-year change in adjacent periods.

If the factor plot shows a step, the correct treatment is to weight the LDF selection toward the 2025 diagonal for the most recent accident years rather than using a multi-year average that dilutes the tariff signal with pre-tariff data. For reserving, that means estimating the 2025 accident year with an LDF that reflects 2025 experience only, or with an explicit external index adjustment. CCC's monthly parts price data provides exactly the kind of objective external benchmark that credibility blending envisions: a 6.0% annual parts cost increase applied to the pre-tariff LDF produces a mechanically defensible adjusted factor without requiring the actuary to invent a trend assumption.

The step-versus-trend distinction also has different pricing implications. If tariff severity is a new trend with ongoing velocity, carriers that filed rate increases in 2024 are partially but not fully behind. If it is a level step that persists at the 2025 parts cost level without further acceleration, those same carriers may be closer to adequacy than the emerging data suggests, because the step is already captured in the most recent accident year at full magnitude. Running both interpretations as scenarios and reporting the range to management is more defensible than committing to a single assumption while the post-tariff cost data is still limited to one accident year of observations.

Why This Matters for Pricing, Reserving, and the Appointed Actuary

For reserving actuaries, the 2025 accident year deserves explicit attention at year-end evaluations. ASOP No. 36 requires appointed actuaries to disclose in their opinions events or conditions that are likely to cause material adverse development on the reserves they are certifying. Tariff-elevated parts costs in personal and commercial auto are documented in third-party vendor data as of Q3 2025, are already present in the 12-month diagonal, and meet the threshold for disclosure. The appropriate response is not necessarily to increase carried reserves above the chain-ladder indication, but to document the assessment of whether the LDF selection adequately reflects the current cost environment, and to note the uncertainty range associated with the step-versus-trend interpretation.

For pricing actuaries, the step interpretation has an immediate rate filing consequence. Severity trend selections based on calendar years running through 2024 understate the current period trend. The appropriate response is to break the trend period at the April to May 2025 tariff effective dates and test whether the pre-date and post-date severity trend rates are statistically distinguishable. When they are, the prospective rate level should reflect the post-date trend, not a composite that averages in the lower pre-tariff period as though the two regimes are draws from the same distribution.

For carriers with longer regulatory approval cycles -- states where filing-to-use lags six months or more -- the total pricing adjustment cycle of 12 to 18 months means 2025 and early 2026 accident years are fully exposed before corrected rates earn through. The reserve development on those accident years will be visible by year-end 2026. Carriers that isolated the step early, built it into LDF selections, and adjusted a priori loss ratios will show reserve adequacy or modest redundancy; those that carried five-year weighted-average LDFs without explicit adjustment for the structural break will be explaining adverse development to reserve committees while simultaneously filing catch-up rate increases in a softening market. The two problems are connected. The triangle is where the diagnosis starts.