Bumper cover prices at U.S. collision shops rose 6.7% year-over-year in early 2026, nearly double the 3.2% rate recorded a year earlier, according to Ryan Mandell's April 10 analysis published by Mitchell International. The total loss rate hit 23.1% of all U.S. auto claims in 2025, the highest proportion on record. North American light vehicle production is on track to fall approximately 944,000 units in 2026 as supply chains absorb a 25% tariff on imported vehicles and replacement parts that took effect in early April under IEEPA authority. Three measurements, all from the same six-month window, converge on the same methodological problem for pricing actuaries: the historical APD severity trend line does not describe the cost environment you are projecting into for 2026 and 2027 commercial auto rate filings.

The Tariff Shock in APD Claims Data

Roughly 60% of U.S. auto replacement parts by volume originate from tariff-exposed countries, primarily China, Mexico, and Canada. Tariffs on these supply chains do not transmit uniformly across parts categories. OEM parts subject to direct import duties, and often without a domestic alternative on late-model vehicles, are inflating fastest. Aftermarket alternatives, where domestic production options exist, face less immediate pressure. Recycled and salvage parts carry the least direct tariff exposure, but supply is increasingly constrained: as total-loss rates rise, more vehicles are processed as total losses, removing them from the recycled-parts stream at the same time demand for those parts increases.

The American Property Casualty Insurance Association estimates that tariff-driven parts and vehicle price inflation will add $26 to $52 billion annually to claims costs across the U.S. P&C market. Repairs on vehicles under three years old now run approximately 10% more than repairs on four-to-six-year-old vehicles, a partial reversal of the historical pattern driven by OEM parts costs indexing more directly to new vehicle tariff pricing. Average first-party deductibles rose 3.25% in 2025. U.S. repairability, after a decade of decline, actually improved approximately 1% in 2025: claims adjusters shifted decisions toward repair and away from total losses to manage catastrophic replacement costs, a rational short-term response that nonetheless means the repair pool now contains claims that historically would have been written off.

Metric Value Source
Bumper cover inflation, 2026 YoY 6.7% Mitchell International (April 2026)
Bumper cover inflation, prior year 3.2% Mitchell International
Total loss rate, 2025 (record) 23.1% Solera / Mitchell
Est. annual P&C claims cost increase from tariffs $26–52B APCIA
NA light vehicle production decline, 2026 ~944,000 units S&P Global Mobility
Avg. first-party deductible increase, 2025 3.25% Mitchell International

How Total Loss Rate Changes Distort Repairable Severity

The interaction between total loss frequency and repairable severity is counterintuitive, and most APD severity trend analyses miss it. When parts costs rise and push more claims across the total loss threshold, the highest-severity repairable claims exit the repairable dataset. Those claims, the ones nearest the total loss threshold before the tariff shock, were the most expensive repairs in the distribution. Removing them truncates the repairable severity distribution from the high end, suppressing average reported repairable severity even as the underlying cost environment has worsened. A pricing actuary who evaluates only average repairable severity may see moderate trend acceleration, while the cost signal is concentrated in the rising total loss frequency and in the severity of total loss claims themselves.

The total loss threshold in most states sits between 70% and 80% of the vehicle's actual cash value, under either state total loss formulas or insurer policy. When repair cost indices, measured by Mitchell, CCC Intelligent Solutions, or the BLS Producer Price Index for motor vehicle parts (NAICS 3363), rise faster than ACV indices from J.D. Power or Black Book, the effective threshold fraction falls. A repair that cost 68% of ACV before the tariff shock may now cost 74%, crossing into total loss territory. The frequency shift is a direct mechanical consequence of the cost structure change, not a behavioral anomaly. Both effects must be modeled separately: repairable severity on the truncated distribution, and total loss frequency calibrated to the new cost-to-ACV ratio.

The Chow Test for APD Structural Breaks

The classical APD severity trend model assumes a stable geometric growth rate and fits a log-linear regression over the full historical period. A tariff-induced step change in parts costs violates that assumption, but the pricing actuary cannot simply discard pre-tariff data without testing whether the break is statistically significant. The Chow test provides that test in a form directly translatable to actuarial rate filing exhibits.

The procedure runs three regressions. First, fit a log-linear severity regression across the full historical period (accident quarters 2020Q1 through 2026Q1, or the most recent available). Second, fit separate regressions on the pre-tariff sub-period (through 2025Q1) and the post-tariff sub-period (2025Q2 onward), noting that the tariff's effects on claims data appear as early as 2025Q4 because anticipatory price inflation from announced and expected tariffs preceded the April 3, 2026 IEEPA implementation date by several months. Third, compute the F-statistic: F = [(RSS_full minus RSS_pre minus RSS_post) / k] divided by [(RSS_pre plus RSS_post) / (n minus 2k)], where RSS is the residual sum of squares for each regression and k is the number of parameters per sub-regression. If the F-statistic exceeds the 95th percentile critical value for an F(k, n minus 2k) distribution, the structural break is confirmed at the 95% confidence level, and the actuary selects a prospective trend using only the post-break period.

When the break is confirmed, six quarters of post-tariff data is thin for a reliable log-linear slope estimate. The actuary supplements it with external parts price indices to extend the effective estimation window: the BLS PPI for motor vehicle parts and accessories (series PCU336300336300), the Mitchell Collision Parts Price Index, or the CCC parts benchmark series are all well-established and consistently accepted by state insurance regulators in commercial auto rate filings. ASOP No. 13, which governs trending procedures in P&C ratemaking, explicitly permits the use of external data sources when internal experience data is limited, and requires disclosure of those sources and the basis for weighting them.

Credibility-Weighted Trend After a Structural Break

With the structural break confirmed and external indices sourced, the selected trend formula is: T_selected = w × T_post_break + (1 − w) × T_index, where T_post_break is the trend estimated from the post-tariff sub-period, T_index is the trend observed in the external BLS PPI or blended Mitchell/CCC index, and w is a credibility weight determined by data volume analogously to the Buhlmann credibility formula. In the Buhlmann framework, w = n / (n + k), where n is the number of post-break observations and k is the ratio of between-period variance to within-period variance. With only six post-break quarters of data, a credibility weight of 35% to 45% on plan data against 55% to 65% on the external index is reasonable for most commercial APD books; carriers with large, high-quality claim datasets may justify weights closer to 50-50.

The worked numbers make the stakes concrete. If the pre-tariff historical trend was +4.0% per year, six quarters of post-tariff data show +9.5%, and the BLS parts PPI is tracking +6.8%, a credibility weight of 40% on plan data and 60% on the external index gives T_selected = (0.40 × 9.5%) + (0.60 × 6.8%) = 3.80% + 4.08% = 7.88%, rounded to 7.9%. The naive straight-line extrapolation from the full historical series would produce 4.0%. Over a standard two-year trend period from the midpoint of the experience period to the midpoint of the effective period, the ratio of projected severity factors is 1.079 squared divided by 1.040 squared, or 1.164 divided by 1.082, a 7.6% difference in projected severity. On a $10 million commercial APD book, that gap is $760,000 in additional projected ultimate losses before expense loading, a material rate indication difference for any fleet account with multi-year terms or guaranteed-cost pricing.

The 3.9 percentage point spread between a naive historical selection and a credibility-weighted post-break selection is also a disclosure obligation. ASOP No. 13 requires the actuary to describe departures from mechanical trending and the rationale for them. A filing memo that simply states "trend selected was 7.9% based on recent experience" does not satisfy that obligation; one that presents the Chow test result, the external index series, the credibility formula, and the worked calculation does. Regulators who receive a material upward trend departure are predictable in their response: they ask for the data behind it. Presenting it proactively shortens the review cycle.

Modeling the Total Loss Threshold Shift

The APD severity trend selection addresses the repairable claims pool. The total loss frequency assumption requires a separate update. The standard approach treats total loss frequency as stable at the historical average, but that assumption fails when tariff inflation has structurally shifted the repair cost distribution relative to vehicle ACV. The modeling procedure starts with the distribution of estimated repair costs for the insured fleet, shifted rightward by the tariff inflation factor from the external index. The ACV distribution is updated using current J.D. Power or Black Book valuations adjusted for the fleet's vehicle age and model year composition, not the historical ACVs embedded in the experience period data. Comparing the shifted repair cost distribution against the updated ACV distribution yields the proportion of claims where repair costs exceed the state total loss threshold, producing a prospective total loss frequency assumption calibrated to the current cost environment.

For total loss claims, the severity pick equals ACV minus salvage recovery. Both components have moved since the historical period. New vehicle values have risen on tariff-inflated replacement costs, increasing gross ACV. Salvage recovery rates have also increased: as imported vehicle and parts prices rise, the domestic recycled-parts market assigns higher value to salvage, partially offsetting the ACV increase. The net effect on total loss severity depends on the balance between the ACV increase and the salvage improvement, and it varies by vehicle age, model, and regional salvage market conditions. Carriers with commercial fleet accounts should obtain current salvage rate data from their salvage auction partners rather than applying a fixed historical recovery percentage to updated ACVs.

In-Flight Claims and LDF Revisions for 2024–2025 Accident Years

The structural cost shift also affects loss development patterns for claims already in the triangle. Claims opened on 2024 and 2025 accident years were initially estimated under pre-tariff parts costs, but they are now developing in a higher-cost repair environment. Parts orders placed after the tariff effective date carry the full import duty, and supplemental invoices from ADAS calibration and specialty parts sources with longer lead times surface at the 24-to-36 month development stage. The expected pattern is a compressed 12-to-24 month link ratio, as initial estimates close before the full tariff-period supplements arrive, followed by adverse development at 24-to-36 months as those supplements post. Actuaries who assume that the 12-to-24 month link ratio on recent accident years is tracking historical norms are likely underestimating the ultimate.

The corrective in the LDF selection is to compare the most recent diagonal's link ratios at each development maturity against historical weighted averages. Where recent diagonals at the 12-24 and 24-36 development points exceed historical averages by more than one standard deviation, a credibility-adjusted selection weighting recent experience at 0.60 to 0.70 against the all-year weighted average is appropriate. The Bornhuetter-Ferguson method provides a useful crosscheck: compare the adjusted chain-ladder ultimate against the BF ultimate using an independently selected expected loss ratio. If the adjusted chain-ladder exceeds the BF estimate, the gap measures adverse development not captured by the BF prior, suggesting the initial BF expected loss ratio itself needs recalibration to reflect the tariff-period cost environment.

OEM Parts Exposure, ADAS Calibration, and Geographic Variation

Commercial fleet contracts compound the OEM parts inflation problem. Personal auto and small commercial policies permit adjusters to substitute aftermarket or recycled parts where available, limiting effective tariff exposure. Commercial fleet service agreements and large fleet accounts commonly specify OEM-only repairs for warranty compliance and fleet standardization. A carrier with concentrated commercial fleet exposure faces a higher effective severity trend than the aggregate Mitchell or CCC data, which blends OEM, aftermarket, and recycled parts across the full personal and commercial claims population, would suggest. The fleet-specific parts mix should inform the external index selection: an OEM-only fleet should be benchmarked against an OEM-specific parts inflation sub-index rather than the blended market composite.

ADAS calibration adds a technology cost layer that operates independently of tariff effects. After most front-end collisions on vehicles with forward-facing cameras, radar sensors, or lane-departure systems, calibration is required before return to service, adding $300 to $1,200 per claim depending on vehicle configuration. Enlyte's 2026 Trends Report documents calibration cost growth as a persistent driver of supplemental severity, distinct from parts inflation. For commercial fleets with high concentrations of late-model ADAS-equipped vehicles, isolating the ADAS calibration trend as a separate component and loading it additively onto the APD severity trend produces a more accurate projection than blending the two signals into a single regression that dilutes both. Geographic variation compounds all of these effects: West Coast markets source a higher proportion of parts from Pacific Rim suppliers and are showing earlier and steeper APD severity acceleration than Midwest or Southeast markets. Carriers with concentrated West Coast commercial fleet exposure should assign higher credibility weight to post-tariff territorial data in their geographic trend selections.

Filing Memo Documentation Under ASOP No. 13

ASOP No. 13 requires the actuary to disclose the methodology used to select trends, any departures from standard procedures, and the key assumptions underlying the selection. A structural break methodology with credibility-weighted external indices touches each of these requirements directly. The filing memo should present, as discrete actuarial exhibits: the Chow test results, including the F-statistic, degrees of freedom, and critical value, documenting whether the break was confirmed at the stated confidence level; the external index series used with the full time history through the most recent available period; the credibility weight selection, including the parametric basis or the judgment rationale if weights are experience-based; the total loss threshold shift calculation, showing pre-tariff and post-tariff cost distributions and the resulting change in projected total loss frequency; and the ADAS calibration load, if applied, with supporting industry data from Enlyte, CCC, or Mitchell.

Carriers moving from a 4.0% historical trend selection to a credibility-weighted 7.9% should anticipate information requests from regulators who receive that departure without context. The Mitchell and CCC index histories and the BLS PPI series for motor vehicle parts (NAICS 3363) are publicly available, consistently cited in commercial auto filings across multiple states, and generally well-received by rate analysts who understand the tariff exposure. An organized set of exhibits that documents the structural break test, shows the index data, and traces the credibility formula from inputs to selected trend converts a large departure into a supported actuarial judgment. That is the distinction between a filing that moves through review and one that stalls. Carriers building 2026 and 2027 commercial auto rate filings now, while Q3 renewals and mid-year filings are in motion, should be incorporating this framework into their trend selection exhibits rather than waiting for the adverse development to appear in the diagonal before adjusting.

Further Reading

Sources

Stay ahead with daily actuarial intelligence - news, analysis, and career insights delivered free.

Subscribe to Actuary Brew Browse All Insights