New York DFS's Circular Letter No. 3 (2026) now requires every motor vehicle rate filing to carry a standalone Exhibit TR-1 quantifying the frequency and severity offset from the state's May 2026 tort reforms, with pending filings amended by August 31, 2026. Because no post-reform New York claims experience yet exists, filers must derive that percentage through claim-level re-triage and credibility-weighted cross-jurisdictional benchmarking, not trend extrapolation.

The Circular Letter's Mechanics

DFS issued Insurance Circular Letter No. 3 (2026) on July 1, 2026, directing every addressee to complete a new "Exhibit TR-1 Automobile Tort Reform Calculation" in all pending and future motor vehicle rate filings. The exhibit demands two things: the percentage decrease in anticipated claims and loss adjustment expenses attributable to the reforms, and a complete explanation of how the addressee derived that percentage, including the specific calculations and all actuarial processes, procedures, methodologies, and assumptions applied. Filers with submissions already in SERFF must amend them by August 31, 2026, and the department has updated its Rate Filing Sequence Checklist to enforce the exhibit going forward, as summarized in a July 6, 2026 client alert from InsureReinsure.

The underlying law is Part EE of Chapter 58 of the Laws of 2026, enacted through S9008-C/A10008-C and signed by Governor Hochul on May 27, 2026, applicable to actions commenced on or after May 26, 2026. It eliminates the 90/180-day non-permanent-injury category from the serious-injury definition under Insurance Law §5102(d), imposes fault-first jury sequencing under §5104(a) so that liability for non-economic loss is not established until a fault finding is made, adds a new comparative-fault recovery bar under CPLR §1411(b) for claimants found 50% or more at fault, and caps non-economic damages at $100,000 under new Insurance Law §5104(d) for a narrow set of at-fault claimant categories. The Governor's office projected the combined reform package could save drivers an average of 10% on auto insurance, roughly $200 per vehicle per year, but that figure is a policy estimate, not a filed actuarial indication, and DOI reviewers will now test every carrier's own derivation against it line by line.

Step One: The Claims Re-Triage Frequency Offset

The 90/180 repeal is a pure elimination effect: some previously compensable bodily-injury claims will no longer meet any surviving Insurance Law §5102(d) category (fracture, permanent loss of use, significant limitation, permanent consequential limitation, disfigurement, loss of a fetus, death, or the 90/180 category itself, now deleted). Building the offset starts with pulling a representative sample of closed New York BI claims and re-mapping each one against the eight surviving categories. Any claim that qualified for compensation solely on 90/180 grounds, meaning it failed every remaining serious-injury test, becomes a claim that would not have been compensable under the reformed statute. Expressing that count as a share of total historically compensable BI claims produces the pure claim-elimination frequency offset.

This step requires claim-file-level review, not aggregate coding, because most claim systems record a serious-injury threshold flag only at the point of settlement or verdict, and that flag frequently reflects the strongest available category rather than every category the claim could plausibly have satisfied. A claim tagged "90/180" in the system of record may also have had a colorable significant-limitation argument. Under-crediting the surviving categories inflates the frequency offset; over-crediting them understates it. The re-triage sample should be stratified by injury type and settlement value, since soft-tissue and strain claims cluster disproportionately in the 90/180 category while fracture and permanent-injury claims rarely relied on it.

Step Two: The Comparative-Fault Frequency Offset, and a Jury-Behavior Caveat

CPLR §1411(b) bars recovery outright once a claimant's fault exceeds the defendant's, at 50% or greater. Modeling this offset means building a histogram of plaintiff-fault percentages from the same closed-claim sample, using adjuster and defense-counsel fault assessments as a proxy for what a jury would find, then calculating the share of claims where assessed fault meets or exceeds the 50% threshold. That share is a second, additive frequency offset, but it is not independent of Step One: some claims eliminated by the 90/180 repeal would also have been barred under the new comparative-fault rule, and summing both offsets without checking for overlap double-counts them. The reconciliation requires flagging the intersection set (claims that fail both the surviving-category test and the fault-bar test) and subtracting it once before combining the two offsets.

A second complication is more subtle and worth disclosing explicitly to DOI: adjuster fault assessments are not a clean proxy for jury verdicts in modified comparative negligence regimes. Eli K. Best and John J. Donohue III's study in the University of Chicago Law Review found that juries in modified comparative negligence jurisdictions are significantly less likely to assign a plaintiff more than 50% fault, and significantly more likely to cluster their findings at or just under the threshold, because jurors resist a finding that eliminates a sympathetic plaintiff's recovery entirely. A naive histogram built from pre-reform adjuster assessments, which had no reason to avoid the 50% line, will therefore overstate the share of claims that actually cross the new bar once juries start adjusting their own findings to the stakes. The prudent treatment is a downward haircut on the raw histogram-derived offset, disclosed as a judgmental adjustment rather than folded silently into the base calculation.

Step Three: The Narrow Severity Cap, Not a Blanket Trend Load

The $100,000 non-economic damages cap under §5104(d) applies only to claimants who are themselves at fault and who fall into one of three categories: operating an uninsured vehicle, driving with a license lapsed more than 30 days, or convicted of driving while impaired or of felony flight from a police officer, with fatal injuries excluded from the cap. That population is small and identifiable in most claim systems through existing fields (insurance status, license status, criminal disposition), so the correct treatment is to isolate that subpopulation's historical share of large non-economic awards, cap the modeled severity for those claims at $100,000, and blend the capped result back into the book's overall severity trend using exposure-weighted averaging. Applying a flat percentage severity haircut across the entire book overstates the reform's effect, since the vast majority of claimants are not uninsured, unlicensed, or criminally at fault, and understates it for the narrow subpopulation where the cap actually binds against historical awards well above six figures.

The Judgmental Piece: Fault-First Sequencing and ALAE

Insurance Law §5104(a)'s fault-first jury sequencing has no direct frequency or severity formula. Its effect runs through settlement leverage: when a jury must resolve fault before non-economic damages attach, plaintiffs with weak liability positions lose negotiating power earlier in the claim lifecycle, which should compress defense litigation costs and settlement timelines. There is no closed-claim field that isolates this effect from the threshold and fault-bar changes already captured in Steps One and Two, so the ALAE trend component of the offset is necessarily judgmental. DOI's demand for full disclosure of "actuarial processes, procedures, methodologies, and/or assumptions" means this judgmental piece cannot simply be absorbed into the frequency offsets; it should be presented as its own labeled line item with an explicit rationale, consistent with the disclosure standard in ASOP No. 41 (Actuarial Communications).

Claim-Filing Date, Not Loss Date

Part EE applies to actions commenced on or after May 26, 2026, not to accidents occurring on or after that date. A 2024 accident where suit is filed in 2027 falls under the reformed statute; a 2026 accident settled without litigation before a suit is ever filed never triggers the reform at all. That means the offset cannot be blended into a conventional annual trend selection keyed to accident date. It behaves as a step function keyed to expected claim-filing date, which for litigated claims can run one to three years behind the loss date. Building the accident-year loss development triangle for the affected years therefore requires overlaying a filing-date distribution on top of the standard development pattern, since two claims from the same accident quarter can fall on opposite sides of the reform depending on when suit is filed.

Blending New York and Out-of-State Evidence: A Worked Credibility Example

With zero New York post-reform claims experience, the re-triage estimate from Steps One through Three carries limited credibility on its own. The standard treatment is a classical credibility blend: let the New York re-triage estimate be the direct indication, let an out-of-state analog, drawn from a prior enactment of a comparable 50%-or-greater comparative fault bar, serve as the complement of credibility, and combine them under a limited-fluctuation or Buhlmann credibility formula keyed to the size of the New York closed-claim re-triage sample.

Illustratively, suppose a carrier's commercial auto book produces a New York re-triage sample of 1,400 closed BI claims, indicating a combined frequency-and-severity offset of 14% after the double-count reconciliation in Step Two. Limited-fluctuation standards commonly require roughly 1,082 claims for full credibility at a 90%/5% standard, so 1,400 claims approaches Z near 1.0 under that benchmark, though a Buhlmann approach keyed to the variance of the carrier's own fault-assessment histogram may derive a materially lower Z. If the out-of-state analog, drawn from a prior 50%-bar adopter's closed-claim experience, indicates an 18% offset, a Z of 0.75 (a level a smaller or more volatile sample would plausibly support) produces a blended indication of 0.75(14%) plus 0.25(18%), or 15%. Commercial and personal auto should not share one blended figure: commercial's claim-fault mix skews toward contested third-party liability, while personal auto carries more single-vehicle claims where the 90/180 category rarely stood alone. Build and disclose the sample, histogram, and credibility weight separately by book.

What Exhibit TR-1 Should Look Like on the Page

DOI reviewers will expect each input as a discrete, auditable line: the re-triage sample and elimination rate for the 90/180 repeal, the fault histogram and jury-nullification haircut for the comparative-fault bar, the double-count reconciliation between the two, the capped-severity blend for the narrow non-economic damages population, the judgmental ALAE load for fault-first sequencing, the filing-date-versus-loss-date triangle treatment, and the credibility formula, its inputs, and the resulting blended percentage, by book. ASOP No. 23 (Data Quality) requires disclosing the reliability of the closed-claim sample; ASOP No. 41 requires the assumptions, not merely the resulting rate, to be reconstructable by a DOI actuary. Under Circular Letter No. 3, the assumptions are the filing exhibit.

Why This Matters

New York has converted a policy question, how much will tort reform actually save drivers, into a line-by-line actuarial audit point embedded in every pending motor vehicle rate filing. Carriers that treat Exhibit TR-1 as a single blended percentage backed into from the Governor's 10% target will draw DOI objections; the circular letter's language, demanding "specific calculations" and "all actuarial processes, procedures, methodologies, and/or assumptions," reads as an explicit rejection of that shortcut. The filers who move fastest through the August 31, 2026 amendment deadline will be the ones who already maintain claim-level serious-injury and fault-assessment coding rich enough to support a defensible re-triage sample, since that data infrastructure, not the credibility formula itself, is the binding constraint on how quickly a carrier can produce an auditable Exhibit TR-1.

Sources

  1. New York Department of Financial Services, Insurance Circular Letter No. 3 (2026): Motor Vehicle Insurance Reforms (July 1, 2026)
  2. InsureReinsure, "New York Department of Financial Services Informs Insurers of Motor Vehicle Insurance Reforms" (July 6, 2026)
  3. JTNY Law, "NY Tort Reform 2026 Passed: 90/180 Rule + 50% Fault Bar Analysis"
  4. Hurwitz Fine, "A Tune-up on Automobile Claims in New York"
  5. National Law Review, "New York's 2026 Tort Reform: Key Changes to Civil Procedure Laws"
  6. Monaco Cooper Lamme & Carr, "Shifting Gears: Understanding New York's Tort Reform"
  7. Wilson Elser, "New York's 2026 Tort Reform: Key Changes to Civil Procedure Laws"
  8. Office of Governor Kathy Hochul, "Governor Hochul Secures Reforms to Lower Auto Insurance Premiums for New Yorkers" (May 2026)
  9. Eli K. Best and John J. Donohue III, "Jury Nullification in Modified Comparative Negligence Regimes," University of Chicago Law Review, Vol. 79, Issue 3

Further Reading