RESEARCH DOSSIER
Motor Insurance Rate Filing and Supervisory Review
Abstract. Motor insurance rate filing is the documentary and regulatory process through which an insurer presents, supports, changes, and, where required, obtains authorization to use premium rates and associated risk classifications. Supervisory review assesses the legal admissibility, actuarial justification, data integrity, conduct consequences, and temporal applicability of the proposed rate structure. This monograph examines the evidentiary content of filings, the relationship between prospective loss-cost estimates and proposed premiums, the supervisory decision process, and the contrasting legal architectures of California, New York, Ontario, the European Union, and the United Kingdom. It develops an original comparative analysis of what supervisors can infer from filings, why aggregate indicated rate changes cannot substitute for analysis of individual impacts, and how supervisory oversight continues where price schedules are not subject to prior approval. The method combines primary legal and regulatory analysis, actuarial reconstruction, examination of published supervisory statistics, and transparent numerical illustrations. The findings establish that the effective object of oversight is a versioned, coverage-specific, legally constrained transformation of evidence into prices, with traceable assumptions and post-implementation consumer outcomes. They further establish that approval, financial adequacy, actuarial classification, fairness, and realized market outcomes are separate empirical and legal questions. This research proposes no new proprietary doctrine or original statistical estimate; its contribution is a structured, evidence-grounded synthesis of established law, actuarial methods, and supervisory practice.
Keywords: Motor insurance; automobile insurance; rate filing; rate regulation; prior approval; insurance supervision; actuarial ratemaking; risk classification; loss trend; regulatory evidence; personal automobile insurance; filing transparency; affordability; consumer outcomes; insurance pricing fairness; Solvency II; SERFF; regulatory comparison.
Introduction
Motor insurance pricing is simultaneously a prospective actuarial estimate, a contractual offer, an allocation of cost across insured risks, and an object of public supervision. These dimensions intersect most clearly in rate filings. A filed schedule can specify a general premium level, a base rate, a sequence of relativities, discounts, surcharges, territorial definitions, vehicle groups, claim-history effects, and rules controlling which policyholder receives which calculated price. A supervisor assessing the filing must therefore determine more than whether the projected portfolio premium approximately balances projected losses. The supervisor must identify the proposed calculation, applicable coverage, dates and legal requirements; evaluate the supporting experience; and assess the distribution of effects on actual policyholders.
A rate is the unit price generated by a rating plan for a stated insured exposure and coverage. A premium is the amount charged for a specified policy transaction after the relevant terms and rating operations are applied. A filed average rate change is a counterfactual result calculated for a defined portfolio under old and proposed rules; it is neither the realized growth in an insurer's premium revenue nor necessarily the renewal change paid by any given driver. An actuarial indication is an estimate of the rate level consistent with explicit prospective cost and financial assumptions. An approval is a legal decision under the applicable regulatory framework. These definitions fix the analytical object throughout the dossier.
The legal frameworks differ substantially. The California rate filing review process describes a prior-approval structure with public notice and opportunities to intervene. The Ontario automobile rate approval explanation describes approval of changes to rates and risk classification systems. In the European Union, Solvency II Article 21 limits prior approval and systematic notification of premium scales. In the United Kingdom, the FCA motor insurance pricing rules address renewal pricing and consumer conduct rather than establishing an American-style prior approval of each insurer's general rate schedule. A single undifferentiated term, rate regulation, therefore obscures significant institutional differences.
The analysis addresses three research questions. First, what evidence and disclosures make a proposed motor rate reconstructible and reviewable? Second, how do supervisors evaluate prospective claims costs, risk segmentation, approval conditions, and later consumer outcomes? Third, what conclusions about regulation remain supportable after accounting for legal differences, data limitations, policy mix, and the timing of statutory changes? The research is international in subject, with a deliberately bounded set of detailed jurisdictional comparisons; it does not claim that these cases exhaust the world's systems.
The evidentiary method gives priority to enacted legislation, official supervisory rules and filing instructions, and adopted actuarial standards. Peer-reviewed actuarial and economic studies inform model choice and interpretation. Public supervisory data are used with their stated measurement restrictions. The numerical examples below are original, fully specified demonstrations rather than observations of a real insurer. A particular regulator's statutory duty is never inferred solely from a model, and a published model result is not treated as universal causal evidence.
Conceptual, Historical, and Primary-Source Foundations
From prospective insurance cost to an authorized rate
A motor insurer incurs costs only after exposure is assumed, claims occur or are reported, and obligations are evaluated under law and policy wording. At the time a premium is quoted, claim frequency, severity, defense costs, loss adjustment expenses, operations, and the cost of financing the underwriting commitment are prospective quantities. Ratemaking translates historical and external information into a forward-looking estimate of those quantities, subject to policy conditions and market mechanisms. Rate filing translates the resulting proposed price schedule into a document that can be assessed under governing law.
The professional articulation of this task is found in ASOP No. 53 on future costs, adopted by the U.S. Actuarial Standards Board and effective for work beginning August 1, 2018. It directs consideration of historical data, exposure and premium consistency, expected future costs, and relevant risk classification. ASOP No. 12 on risk classification addresses the professional analysis of systems assigning risks to classes expected to have differing outcomes. These standards are professional guidance in their respective spheres, not substitutes for a local insurance statute and not assertions that every jurisdiction uses the same legal test.
Motor contracts create several exposures that must not be consolidated without careful justification: bodily injury liability, property damage liability, no-fault or first-party injury benefits, uninsured-motorist coverage, collision, and other-than-collision physical damage. The BLS motor vehicle insurance CPI methodology provides a useful external reminder that even a broadly labeled motor insurance index covers particular private-passenger coverages and excludes various business policies from its pricing sample. A rate analysis should identify line, coverage, deductible, limit, policy term, vehicle use, jurisdiction, insured group, and the unit of exposure before any comparative inference.
Motor premium levels also respond to traffic behavior, vehicle technology, repair methods, medical and legal costs, weather, litigation, fraud control, reinsurance terms, and insurer expenses. A legislative revision to injury compensation can change expected losses without changing the number of insured vehicles. A vehicle-safety improvement may reduce certain injury claims while repair complexity increases physical-damage severity. An external price index may rise with the motor market while one insurer's expected losses fall after a change in selection, product design, or claims handling. These are different causal channels and belong in separate components of the filing evidence.
Historical forms of public control
The U.S. tradition of rate regulation is organized around legal prohibitions on excessive, inadequate, or unfairly discriminatory rates, with variation in whether prior permission is required. California's voter-approved Proposition 103, adopted in 1988, became a prominent prior-approval architecture. The California application instructions include automobile liability and physical damage among the covered lines. Under Insurance Code section 1861.05, rate approval concerns excessiveness, inadequacy, and unfair discrimination, with filing requirements and statutory procedures. California also treats the public's ability to inspect submitted information as part of the institutional design, reflected in section 1861.07.
In New York, Insurance Law section 2303 states standards for property/casualty rates, including excessiveness, inadequacy, unfair discrimination, competitive destruction, and impairment of solvency. Section 2305 and its revision history define filing and approval requirements and cannot be read accurately without attention to exceptions and effective dates. In Ontario, Ontario automobile rate statute addresses approval of private-passenger rate classifications and rates. Historical labels such as prior approval or flexible rating thus require verification of product scope and amendment dates.
European legislation took a different path. Solvency II Article 181 generally precludes routine prior approval or systematic notification of non-life premium scales while recognizing certain contract-document communications for compulsory insurance. It follows that the absence of an American-style motor rate filing requirement is a deliberate legal feature in the European setting, not evidence that price setting escapes conduct, solvency, competition, data protection, or anti-discrimination oversight.
The boundary between pricing evidence and social outcomes
Evidence that a premium covers expected costs does not establish its affordability. Evidence that a risk factor predicts claims does not establish that the factor is legally permissible. Evidence that a policyholder's renewal quote is lawful does not show that competing products offer equivalent coverage or claims service. These distinctions are necessary because public objectives coexist: the ability of insurers to meet liabilities; access to compulsory cover; predictability for buyers; differentiation of risk; avoidance of legally prohibited classifications; and efficient claims payment.
In an important empirical study, Cohen and Dehejia (2004) examined changes in American insurance and liability law and traffic fatalities over 1970–1998. Their identification strategy and historical population help explain why insurance and liability institutions may affect driving incentives. Their estimates concern safety behavior under specified laws and years. They do not directly estimate how a contemporary state rate-filing procedure changes accident frequency, nor do they establish a general effect of modern rate approval on road safety. The distinction between the actual studied intervention and the regulatory mechanism examined here is essential to sound interpretation.
Defining the review unit
The smallest reconstructible object is a filing version for an insurer or rate-service organization, covering a legally defined product and market, with a proposed effective date and linked support. It can be identified by a regulator's file number or SERFF tracking number, the submitted algorithm or manual, the actuarial memorandum, exposure and experience exhibits, and the supervisory disposition. A later amendment to the same application is a distinct version, because the decision must attach to the rates actually approved or permitted to take effect. The decision record should distinguish the insurer's original request, the regulator's inquiries, amended rate factors, final disposition, and implementation date.
This unit-of-analysis choice is an analytical synthesis of the filing instructions and legal materials. It is useful for comparing procedures without claiming that all jurisdictions issue comparable applications or reveal the same level of documentation. Where a system emphasizes principles-based conduct review rather than rate approval, the comparable observational unit becomes a pricing-policy version or supervisory review episode, with its own evidence of governance and outcomes.
Filing Evidence and Disclosures
Documentary completeness and legal identity
A scientifically reproducible rate filing starts with the insurer's legal entity, covered jurisdictions, coverage and product classes, group or affiliated-company relationships when relevant, policy forms, prior approved plan, proposed plan, filing purpose, dates, and the exact requested change. This identity prevents an apparently plausible ratemaking analysis from being applied to the wrong exposures or obsolete policy conditions. Personal automobile and commercial fleet rates must be differentiated; mandatory liability and elective physical-damage covers often obey different benefit triggers and loss distributions.
The NAIC SERFF filing system description explains the electronic infrastructure used by U.S. regulators and insurers and emphasizes that state rules and workflows remain distinct. California requires electronic P&C rate filings through SERFF. New York's property/casualty filing instructions require rate, rule, and form information in the relevant SERFF fields; a filing description should identify substantive changes to methodology or coverage. Electronic transmission standardizes the record channel, not the substantive legal approval standard.
The documentary inventory should contain a cover or filing description; prior and proposed rating manuals; a change log; classification definitions; base rates, factors, limits, and discounts; actuarial exhibits; the indication and selected implementation; and a signed certification where the specific regime requires it. The effective date must be linked to policy issuance or renewal rules. For each quantitative exhibit, the reviewer needs its period, numerator, denominator, coverage, data source, adjustments, and reconciliation to the insurer's experience. A table of rate factors without its derivation is a price schedule; a derivation without the implemented calculation is an incomplete filing.
Exposure, loss, and premium reconstruction
The foundational actuarial identity is the expected pure premium per exposure: expected claim frequency multiplied by expected claim severity, with care where multiple coverage components, large losses, or claims adjustments require more elaborate models. For coverage c and segment g, a conceptual decomposition is:
where E denotes exposure units where appropriate, N is the number of claims, Y is cost per claim, and X identifies available risk information. For a more explicit portfolio formulation, let U be selected ultimate indemnity losses, A selected allocated and unallocated loss adjustment expenses, F fixed expense per exposure, v the fraction of premium represented by variable expense, and q the selected underwriting profit and contingency fraction. Under those deliberately simplified assumptions, the indicated premium per exposure is
This formula is an instructional accounting identity, not a legal formula prescribed by an insurance authority. It excludes financing adjustments, policy fees, taxes, reinsurance effects, investment income, and capital mechanics unless expressly introduced into the indicated components. A real filing must treat those elements consistently and avoid double counting costs.
Premium history needs an on-level adjustment: earned premium generated under several historical rate schedules cannot be compared directly to losses projected under a new proposed plan without reconstructing the rate level. Exposure history needs consistent earned car-years or another justified unit; insured vehicle counts at year-end are a different measure. Ultimate losses require development of reported and paid experience, coverage-limit adjustment where material, catastrophe or large-loss treatment, and a clear distinction between accident, policy, report, accounting, and calendar years.
The ASOP No. 23 data quality standard addresses selection, review, reliance, and disclosure of data limitations. It provides the appropriate professional reason to record known data defects rather than silently fill gaps. An insurer's booked losses may be adequate for financial reconciliation while its segment-level claim coding remains insufficient to estimate a new telematics factor. Conversely, high-quality claim records for a narrow risk class may be insufficiently credible for a stand-alone rate estimate.
Prospective trend and cost components
An evidence chain should separate claim frequency trends from severity trends; bodily injury from property-damage costs; inflation in repair parts from inflation in labor or medical expenditures; and observed data from selected prospective assumptions. A severity trend is a change in expected cost conditional on the definition of a claim; a frequency trend is a change in claims per exposure. If the two factors change by f and s, the approximate multiplicative loss-cost change is (1+f)(1+s)-1, which includes the interaction term fs. Simply adding percentage trends neglects that interaction. Annual selected rates also require an appropriate projection interval to the average accident date of the prospective coverage period.
Regulatory filing support should explain why trend estimates differ from industry benchmarks, how unusual experience is treated, and whether shifts in vehicle age or policy limits explain apparent loss changes. The Ontario annual review of private-passenger auto describes regulatory use of benchmark information and the statutory just-and-reasonable assessment. This is evidence for Ontario's specified review approach, not a universally mandated projection method.
Rating factors, classification, and distributional effects
A complete filing makes the rating algorithm executable. A common multiplicative expression takes the form P(x)=B times the product of rating relativities for declared factors, with subsequent discounts, taxes, policy fees, and transaction-specific adjustments accounted for separately. It must be clear which products are compounded, capped, rounded, or applied in sequence. Otherwise identical stated factors can produce different collected premiums, especially with minimum premiums, multi-car discounts, and discounts computed against different bases.
California Insurance Code section 1861.02 establishes a defined ordering of primary private-passenger automobile factors: driving safety record, annual mileage, and driving experience, followed by other commissioner-adopted factors that meet the statutory requirements. That legal rule cannot be replaced by the variable-importance ranking of an insurer's predictive model. A model may statistically rank a proxy as informative but the filed classification remains constrained by law. Conversely, when a factor is allowed, the filing still must address its relevance, data provenance, and effect on policyholder cohorts.
Two distinct impact analyses are necessary. The overall average change holds the reference portfolio constant to compare old and new premium calculations; the individual impact distribution records percent and dollar changes across the portfolio and identifies the tails, affected vehicle groups, and vulnerable or constrained buyers where observable lawfully. If a static portfolio has 70% of prior premium in segment A and 30% in segment B, a 4% proposed change for A and a 2% reduction for B produces a 2.2% premium-weighted overall change. An individual in A faces a 4% change absent other alterations. Even that static calculation differs from realized written-premium growth after lapse, new business, revised limits, and geographic migration. The arithmetic is elementary, but the inference is operationally important.
Document-level auditability
The filing should preserve source-system extracts or reproducible aggregate tables, date-specific assumptions, versioned model code or equivalent specifications, transformations from exposure and claims to rate indications, and an explanation of how the final implemented rating manual differs from the selected actuarial indication. A supervisor need not possess an insurer's entire operating database to demand enough support to test material representations. For models using telematics or third-party scores, the boundary of auditability includes consent and lawful data use, version drift, proxies, missingness, and stability of factors; the precise legal requirements depend on jurisdiction.
The legal visibility of an application varies. The California Department of Insurance's public rate filing search resources and virtual viewing room provide avenues for access to specified public filings. The NAIC explains that SERFF Filing Access depends on participating jurisdictions and publicly available documents. The ability to locate a document proves public access to the available record, not that every detail necessary to independently replicate an insurer's proprietary pricing engine is disclosed.
Rate Approval and Monitoring
The supervisory review sequence
A rate-filing review can be reconstructed as a sequence of substantive decisions. Intake establishes jurisdiction, timeliness, submission format, and required documents. Completeness review asks whether the change can be understood and the calculations followed. Actuarial review examines historical and prospective loss and expense support. Legal review tests coverage, classification, and substantive rate criteria. Conduct review investigates how the proposed rating plan treats purchasers and how the insurer will operationalize its promises. The supervisor then approves, rejects, requests revision, holds a hearing, permits use after a statutory waiting period, or records another disposition under the applicable law. Subsequent monitoring compares implemented prices and outcomes with the filing's assumptions and applicable duties.
This sequence is an analytical description that must be tailored to each jurisdiction. For California, the Rate Regulation Division's published procedure identifies intake compliance review, public notice, a period for review, and public-intervenor hearings in Proposition 103 filings. Its stated timeframes include fourteen days for basic compliance, ten days for notice following that determination, and sixty days after public notice for a decision or notice of hearing, with additional rules and a statutory deemed-approval effect described on the page. These timeframes attach to the particular statutory procedure; they are neither a universal U.S. norm nor a guarantee that complex contested filings conclude on that schedule.
For New York, the administrative materials identify the SERFF filing description and schedules through which the regulator reconstructs a proposal. The relevant Insurance Law section 2305 includes specific classes requiring prior approval or a statutory waiting period and must be read together with exceptions, notably nonbusiness automobile flexible rating. Ontario's statutory rate application provisions require specified information, material, and evidence and allow requests for additional information. These official procedures show a common need for reasoned evidence despite different legal mechanisms.
Review of adequacy and adverse feedback
Adequacy concerns prospective capacity to pay claims and reasonable expenses under applicable legal standards. If observed frequencies rise after a reduction in road travel, a supervisor should ask whether exposure measurement shifted rather than automatically raising the selected rate. If claims severity increases, the reviewer should isolate coverage mix, settlement patterns, benefit changes, and claim development before treating the entire movement as structural inflation. The time lag between the insurer's filing and the realized insurance experience creates unavoidable uncertainty; actuarial review accordingly evaluates ranges, scenario dependence, and judgment as well as point indications.
An excessive-rate inquiry has a different evidentiary direction: does the requested premium exceed what the operative test permits when supported losses, expenses, permitted margin, investment income treatment, and other relevant factors are recognized? The policy ratio of claims paid to premiums collected cannot answer that question by itself. A low ratio might reflect unearned or unreported risk, an unusual claims period, a product with substantial servicing costs, a changed deductible, or a genuinely excessive charge. The appropriate conclusion depends on the specified observation window and financial reconciliation, as well as statutory definitions.
Discrimination analysis also has two layers. Statistical discrimination in the technical sense concerns predictions that differ across groups; prohibited discrimination concerns a legal restriction on the use of characteristics or unequal treatment. U.S. actuarial risk-classification guidance in ASOP No. 12 addresses appropriateness and effects of classification systems while expressly recognizing applicable legal constraints. The existence of predictive power cannot itself establish permission to use an attribute. Conversely, prohibiting a particular variable does not establish that the resulting portfolio rate level must become inadequate: the insurer may revise factors, cost estimates, expenses, product design, and portfolio strategy within law.
An actuarial estimate of expected claims cost may be statistically reliable in the aggregate and unstable for a thinly populated group. The supervisor should examine data support, smoothing, credibility, model calibration, exposure distributions, and whether a proposed factor produces anomalous impacts at category boundaries. Boucher, Denuit, and Guillén (2007) compare alternative claim-count specifications on an automobile portfolio, reinforcing that frequency-model form is an empirical question. Dionne and Vanasse (1992) analyze how a priori classification and claims-history experience rating interact. Their methods demonstrate why a predictive variable's incremental contribution should be studied conditional on the complete model, not by reading univariate experience tables as proof of independent causation.
Reproduced numerical illustration of a filing indication
Consider a wholly hypothetical portfolio with 100,000 earned private-passenger car-years, 6,000 claims, and $90 million of selected ultimate indemnity losses for a common coverage definition. Selected frequency is 0.06 claims per car-year, selected severity is $15,000 per claim, and the initial indemnity pure premium is $900 per car-year. Assume loss adjustment expenses of 12% of ultimate indemnity, fixed operating expenses of $120 per car-year, variable expenses of 18% of premium, and a selected underwriting profit and contingency provision equal to 5% of premium. The simplified prospective premium is then ($900 + $108 + $120)/(1 - 0.18 - 0.05), or approximately $1,464.94 per car-year. Relative to a hypothetical current on-level premium of $1,400, that produces an indicated change of approximately 4.64%.
Now suppose, independently, that a one-period projection applies 3% frequency growth and 6% severity growth before the same loss adjustment and expense assumptions are used. The compounded loss-cost multiplier is 1.03 × 1.06 = 1.0918. Indemnity pure premium rises to $982.62; its 12% loss-adjustment component becomes $117.91; and the selected premium becomes approximately $1,585.11. Relative to the same $1,400 current on-level premium, the indicated change is approximately 13.22%. Neither indication is an observed rate filing, actual approved percentage, or forecast for any real insurer. The example isolates the sensitivity of a requested rate change to the prospective trend assumptions, holding other modeled features fixed.
The illustration also exposes how different assumptions can offset. A selected 6% severity trend might be replaced by a lower figure when claims settlements show a temporary spike; a 3% frequency trend might be inappropriate if the exposure cohort changes; a different profit allowance could increase or decrease the denominator; a reinsurance provision could shift the expected cost. The reviewer needs the methods and uncertainty supporting each selection and the insurer's explanation of any departure between actuarial indication and chosen rate proposal. An indication provides evidence for a requested amount but does not itself generate legal authority to charge it.
Rate adequacy over time
A supervisor should distinguish filing-year profitability from underwriting-year loss emergence. Liability claims can develop over years; physical-damage claims may settle faster. Reinsurance payments, recoveries, litigation outcomes, and claims expense patterns can change the realized combined ratio for an underwriting cohort after approval. A rate schedule that was reasonable on the evidence and law available at review can later generate unexpected losses. Conversely, a new premium increase may coexist temporarily with an improving loss ratio if earned premium lags the effective-date change or claims experience turns favorably.
Monitoring therefore requires a temporal bridge: the filing's prospective claims period, the actual policy effective dates, earned exposure and premium, accident-year losses, and later paid and incurred development. Supervisory analysis can attribute deviations to trend error, reserve development, catastrophe events, changes in claims adjudication, selection, or shifts in coverage mix. This decomposition is more probative than a raw comparison of calendar-year profits with a single approved percentage.
Supervision after pricing implementation
Ongoing oversight may examine filed-versus-implemented algorithm conformance, consumer complaints, renewal disclosures, sales practices, cancellations, segmentation drift, insurer solvency, and concentration in residual or assigned-risk markets. California's public rate-filing visibility arrangements and Ontario's insurer-specific rate approvals database support public observation of decisions, while providing different forms of information. A posted overall average approval is one regulatory signal; its actual incidence on drivers requires the insurer's approved relativities, renewal rules, and portfolio composition.
Public conduct monitoring serves a separate function where no prior rate approval exists. The FCA's general insurance value measures require specified reporting for relevant retail general insurance products and make common outcome indicators available. A regulator can challenge a poor-value product or remuneration pattern even when its current premium was never individually signed off in a filing. This demonstrates the broader supervisory meaning of pricing review.
Jurisdictional Legal Differences
California: permission, classification, and public participation
California provides the clearest example in this comparison of a public prior-approval mechanism applied to private-passenger automobile rates. Proposition 103's statutory structure incorporates substantive rate tests, filing procedures, and mechanisms for public participation. The department's rate review process explicitly distinguishes prior-approval applications and automobile class plans. An insurer contemplating a rate or plan change must analyze both the applicable rate-level rules and the legality of its proposed classification factors.
The hierarchy in section 1861.02 is consequential for model governance. The statute directs attention to driving safety record, annual mileage, and years of driving experience in specified decreasing order of importance, followed by authorized other factors. This produces an explicit example of law determining the allowed architecture of pricing rather than merely checking the total premium produced by an unconstrained model. Material changes to telematics data, claims-history treatment, vehicle classes, or territory need analysis within the actual governing rules and approved class plan.
The rate approval standard in section 1861.05 requires attention to excessiveness, inadequacy, and unfair discrimination, while the public-inspection provision in section 1861.07 strengthens the visibility of supporting information. These features make California's filed documents useful research sources when available, although public access to a filing does not relieve the researcher of checking amendments and decision dates. An application requested by an insurer, a revised request, a proposed decision, an approved filing, and an implemented rate are five distinguishable stages.
California's online SERFF and WARFF filing access descriptions specify the existence of public filing search and different historical coverage of systems. This availability supports case studies of requested and approved motor changes. It does not justify counting search results as a complete universe of insurer prices without verifying line codes, filing types, versions, and the observation period.
New York: baseline standards, flexible rating, and a dated statutory transition
New York demonstrates why a current-date comparative analysis must include effective-law transitions. Insurance Law section 2303 provides the substantive standard for rates. Section 2305 supplies filing and approval procedures and references exceptions. Section 2350 historically permitted qualifying nonbusiness automobile rate changes within a specified five-percent band to take effect without prior approval, under its applicable conditions. It is inaccurate to describe every New York nonbusiness motor rate change as always requiring individual pre-use approval without acknowledging the flexible-rating provision.
On July 1, 2026, New York's Department of Financial Services issued Insurance Circular Letter No. 3 (2026), explaining reforms and directing insurers to reflect relevant expected changes in claim frequency, severity, and loss-adjustment expenses in their pending and future motor rate filings. The circular identifies November 27, 2026, as the effective date of a reform removing the ability to implement qualifying overall average rate increases of up to five percent without prior approval, while retaining a corresponding path for certain decreases, with a later statutory sunset described for 2030. As of this dossier's research date, November 16, 2026, the November 27 change is prospective. The preceding flexible-rating treatment remains the operative frame for filings governed by the pre-change rule; this research does not backdate the amendment.
The circular's instruction to recognize expected claims savings yields a substantive analytical insight. Supervisory review is not limited to asking insurers to justify increases when external costs rise. It may also require actuarially defensible recognition of cost reductions following legislative changes, subject to the evidence and transition period. When the underlying legislation addresses fraud or liability rules, uncertainty about claim response and implementation must be disclosed rather than encoded as a certain numerical saving. The causal effect of the reform should be tested against observed post-effective-date outcomes rather than treated as established by a policy objective.
New York also illustrates the distinction between a filed overall average change and the customer-specific bill. Even a lawful rate-level movement within a flex corridor can redistribute charges as classifications, discounts, and market composition change. Formal filing visibility and published administrative records support reconstructing the approved or permitted terms; actual distributional effects require matching the rate algorithm to relevant insured exposures.
Ontario: detailed approval of private-passenger rates and classifications
Ontario's legal requirements provide a second example of prior-rate review outside the United States. Under Automobile Insurance Rate Stabilization Act section 3, insurers apply for approval of rates and risk classification systems for the private-passenger automobile category. The Act describes evidence requirements and criteria for refusing approval, including whether the proposal is just and reasonable, appropriately predictive and distinguishing, compatible with solvency, and not excessive in the relevant circumstances. These legal elements connect aggregate premium reasonableness to the design and implications of segmentation.
The FSRA explanation of automobile rate approvals states that the regulator examines insurer applications and their assumptions using market conditions, benchmarks, and expenses. It also explains that an average approved change need not equal the increase or decrease experienced by a particular policyholder. Ontario's private-passenger annual review guidance supplies a documented example of benchmarks that inform review without eliminating the requirement to examine insurer-specific evidence.
A critical temporal qualification concerns newer Ontario guidance. FSRA's Automobile Insurance Rating and Underwriting Guidance describes an intended more outcomes-focused system with accreditation, fast-track streams, and a filing manual. The agency's published page explicitly states that the guidance's effective date has not yet been determined, even though the page carries an administrative status designation. On November 16, 2026, its future procedural details must therefore be treated as announced guidance subject to implementation timing, not asserted to have already displaced every older filing rule. This is an important example of a current website containing substantive regulatory design that is not yet a safely established operative workflow.
Ontario's public rate approval database provides useful insurer-level historical decisions dating from 2018 according to FSRA's description. Research based on that data should record the date of extraction, distinguish original approval from policy effective date, check insurer combinations and coverage mix, and refrain from treating the arithmetic average of approved percentages as an industry-wide premium-weighted change unless the correct weights are available.
European Union: freedom of tariffs with other supervisory constraints
EU law organizes the pricing issue around the prohibition of systematic price-schedule approval. Solvency II Article 21 states that Member States generally must not demand prior approval or systematic notification of premium scales and technical bases for non-life business. Article 181 for non-life insurance reinforces this rule, while allowing certain communications of compulsory-insurance conditions and preserving a narrowly framed exception for general price-control systems. Article 154 on host-state powers adds a cross-border dimension: a host Member State likewise cannot construct a routine prior-rate-approval regime for an incoming insurer contrary to these provisions.
These provisions do not make price formation unrestricted. Insurance contracts, product governance, consumer information, solvency, competition law, and anti-discrimination rules continue to shape motor insurance business. For example, Directive 2004/113/EC on equal treatment is part of the legal history of sex-based insurance rating; its text and later judicial effects require attention to the operative position rather than a selective quotation of superseded exceptions. The key comparative fact is the allocation of regulatory intervention: EU supervisors can act on lawful conduct and solvency questions without receiving each motor insurer's proposed tariff for routine pre-use approval.
A regulator operating under this regime may still demand pricing-governance evidence during supervisory assessment. A comparative study that coded 'no rate filing' as 'no rate supervision' would confuse the absence of one procedural instrument with the absence of institutional oversight. The correct observation is narrower: the legal system generally rejects systematic tariff prior approval while retaining other supervisory tools. This also means that European non-life insurance rate data may be less directly comparable with U.S. state approval records, because the European population of prices is not generated by an equivalent regulatory filing process.
United Kingdom: conduct-based renewal pricing and value measures
The United Kingdom provides a distinctive motor pricing governance example. The FCA Handbook ICOBS 6B regulates home and motor insurance pricing, including the relationship between renewal prices and equivalent new-business prices. Its operative rule prohibits a firm from setting the renewal price above the equivalent new-business price within the defined scope. This is a consumer-conduct price constraint, not a general requirement that an FCA supervisor approve each insurer's actuarial premium schedule before use. The FCA's guidance expressly distinguishes the renewal fairness objective from underlying risk pricing.
The distinction matters analytically. A motor insurer may have a valid loss-cost model and a legally problematic distribution strategy if renewal pricing exploits tenure inconsistently with the Handbook. Equally, compliance with renewal price-parity rules does not demonstrate adequate total premiums, high claims acceptance, or reasonable deductibles. Supervisory questions consequently concern comparable products, timing of offers, add-ons, fees, remuneration, and actual application of the pricing rule.
The FCA's 2025 value measures publication, first published in July 2026 and updated in September, reports that aggregate motor insurance premiums fell 7% in 2025, average policies in force rose 4%, and claims costs as a proportion of premium rose from 54% in 2024 to 59% in 2025. The FCA explicitly warns that the timing of new business premiums and claim payments can differ. These are published product-level statistics for a defined reporting system, not a calculation of the legally permitted rate increase for a particular insurer or proof that each premium decrease improved consumer welfare. The data are valuable precisely because they bring observed outcomes into an oversight system centered on ongoing conduct and value.
Comparative legal matrix
| Jurisdiction | Main mechanism studied | Object of supervisory evidence | Key legal or measurement qualification |
|---|---|---|---|
| California | Proposition 103 prior rate approval and automobile class-plan review | Rate applications, actuarial justification, rating factors, public notice and intervention | Substantive California standards and factor hierarchy apply; procedure varies by filing type |
| New York | Property/casualty rate standards with motor filing requirements and an operative flex-rating exception | SERFF documents, overall rate changes, prospective claims assumptions | November 27, 2026 restriction on increases is prospective as of November 16 |
| Ontario | Approval of private-passenger automobile rates and risk classification systems | Loss trend, expenses, actuarial support, class impacts, regulator benchmarks | Newly described outcomes-focused guidance has a separately undetermined effective date |
| European Union | General prohibition on systematic prior approval of non-life premium scales | Supervisory conduct, solvency, consumer and contractual evidence under applicable law | Absence of tariff prior approval is not absence of pricing supervision |
| United Kingdom | Renewal conduct rules and product value monitoring | Equivalent new-business comparisons, product outcomes, claims metrics | FCA renewal parity is distinct from insurer-wide prior rate approval |
The table compares legal mechanisms rather than creating an index of regulatory strictness. Filing counts, approval times, affordability measures, and customer outcomes have different generating processes across systems. A numerical ranking of regulatory 'quality' from these descriptions would exceed the evidence.
Comparative Evidence, Data Uncertainty, and Validation
A framework for evaluating what filings can demonstrate
The comparative evidence supports a hierarchy of increasingly demanding research claims. At the documentary level, an official record may establish that a company submitted a request. At the procedural level, the regulator's disposition establishes what was accepted or permitted under its applicable rules. At the actuarial level, sufficient exhibits may establish that a requested rate is internally consistent with specified data and assumptions. At the economic level, realized policy-level outcomes require observation after implementation. At the causal level, demonstrating that rate approval or a pricing rule changed consumer outcomes requires a defensible counterfactual that accounts for other contemporaneous influences. These are stages of inference within established empirical reasoning, not a proposed new theoretical system.
A filing's requested change must never be mistaken for an approved change. An approval must never be mistaken for an implemented or earned change. A permitted plan must never be mistaken for a uniform change across policyholders. A reported calendar-year loss ratio must never be mistaken for the prospective indication available on the original review date. Finally, a decline in approved rates across a selected insurer sample is not necessarily a decline in the average premium paid by all insured drivers, because risk mix, coverage selection, taxes, and insurer market shares may differ.
The identification problem can be stated explicitly. Let Y_{i,t} denote a policyholder outcome, such as paid premium for constant coverage, and R_{j,t} a rating-rule version for insurer j at time t. Observing Y after R changes does not identify the effect of the regulator's approval because many variables change simultaneously: accident exposure, vehicle choice, injury compensation, repair costs, competition, renewal behavior, and insurance mandates. A defensible policy evaluation needs a plausible control, a record of pre-implementation trends, stable definitions, and tests for endogenous timing. Publication of a series of approved percentages alone supplies none of these conditions.
Data-generating units and exposure mismatch
The choice of denominator changes interpretation. Claims per earned car-year, claims per policy-year, and claims per insured driver can diverge when households insure several vehicles, coverage terms vary, or commercial use changes. Property-damage losses per covered vehicle may trend differently from bodily injury liability losses per insured driver. Loss ratios use premium as denominator, whereas claim frequency uses exposure. Renewal retention requires a cohort of customers eligible to renew, not merely policies active on a calendar date. Each unit answers a different question.
The BLS description of motor insurance CPI pricing clarifies one important comparability limit. The CPI's sample and fixed-product pricing design are built for a consumer price index, not for an insurer's on-level loss-cost indication or a regulator's weighted average of approved requests. The market-average premium paid can change because insured buyers select different coverage, even if a standardized constant-risk price index records another trend. Researchers should therefore report which outcome is measured rather than place these series into a single unlabeled chart.
The FCA's value measures methodology and data provide a second example. The published ratio of motor claims costs to premiums is not a universally comparable statutory loss ratio. Product definitions, reporting periods, claims timing, and quality controls matter. The FCA itself emphasizes timing lags and data qualifications. Observed changes in that ratio cannot automatically be interpreted as the effect of the UK's renewal pricing rule, because claims experience and underwriting cycles also move.
Model specification and sensitivity testing
Supervisory data review should separate model-fitting success from decision robustness. A claim-frequency model can fit the sample well while producing fragile relativities for a small territory or newly introduced vehicle technology. Count models may require an offset for exposure; severity models must define the positive-claim sample and treatment of zero, censored, and large losses. Claim-level correlations within insured households and dependence among coverage components affect confidence intervals. Out-of-time validation is generally more informative for future pricing than a purely random split when market costs and rules shift over time.
The published comparison by Boucher, Denuit, and Guillén (2007) shows how alternative frequency models can be assessed with diagnostic tests and information criteria rather than chosen by habit. The Dionne and Vanasse (1992) study, based on 19,013 drivers in the authors' sample, demonstrates the value of integrating past claims with prior risk characteristics in a structured probabilistic model. These works support the methodological possibility of alternative specifications, not a claim that a particular regulator has adopted their models or that their estimated coefficients can be transplanted across countries.
A scientifically legible filing can present three sets of analyses. The first documents observed experience and all material corrections. The second provides a central actuarial indication and confidence or scenario ranges that respect estimation and model uncertainty. The third shows the requested implemented plan and its impact distribution, including cases where management selection departs from the central indication. The reviewer should trace every material departure to a stated reason. This sequence operationalizes documented professional approaches to estimating prospective P&C costs and data quality and disclosure.
Prohibited or restricted variables and the problem of proxies
Insurance risk scores can import geographical, behavioral, or demographic information through indirect predictors. A predictor's conditional correlation with losses may survive historical validation while reflecting social processes, legal exclusions, or instability under new technology. Where law prohibits a factor, using a close proxy can create additional legal and governance issues. A prudent review asks how a score is constructed, which population it represents, whether missing values have systematic patterns, whether interactions reproduce restricted classifications, and how the insurer can explain an individual's rate under the operative law.
The California statutory primary-factor hierarchy makes these questions especially concrete. In the European Union, conduct and equal-treatment controls arise through different legal instruments from tariff approval itself. The relevant review therefore has to map each modeled variable to its jurisdiction-specific rule before deciding whether its use is permissible. A global blanket claim that one motor-rating feature is legal or illegal everywhere would have no evidentiary foundation.
Timing, legal validity, and source audit
This dossier's legal comparison is anchored to November 16, 2026. The distinction is operational rather than decorative: New York's July 2026 motor reform circular identifies a November 27, 2026, change that has yet to become operative at the research date. Ontario's new rating and underwriting guidance explicitly leaves its effective date undetermined. A web document's current availability and its provision's legal effect are separate facts.
The evidence recorded in the accompanying EVIDENCE.json links material assertions to the official or scholarly source checked for each claim. The file contains source identities, observed limitations, and the research date; it is a source-to-claim ledger, not a claim that proprietary company filings or complete underlying actuarial datasets were independently audited. The bibliography preserves full identification and direct source links. Contemporary law and public data remain subject to future amendment, correction, or restatement. Such changes can affect a later publication decision even when the manuscript's historical as-of statements remain accurate.
Limits of inference
The research does not possess a jurisdiction-wide linked database containing every requested rate, regulator modification, policyholder transaction, accident occurrence, and claim payment. It therefore makes no original population estimate of the effect of regulatory prior approval on insurance prices or market affordability. The numerical ratemaking calculations are explicit hypothetical examples. Causal claims are limited to descriptions of published studies in their own designs, rather than imported as findings for motor rate filing.
The jurisdictional selection is purposeful: California illustrates public rate prior approval; New York illustrates a statutory transition; Ontario illustrates rate and classification approval with a future guidance transition; the EU illustrates freedom of tariffs; and the UK illustrates price-related conduct rules with public value outcomes. Jurisdictions not treated in depth, including most U.S. states, Australia, Japan, and emerging insurance markets, warrant additional study before any global generalization. This bounded comparison supports a substantial institutional synthesis while keeping the underlying legal heterogeneity visible.
Integrated Results and Open Research Questions
Principal research findings
The first integrated finding concerns the object of a review. A filing is an evidentiary proposal for a defined and versioned rate calculation, not simply a percentage. A percentage can be mathematically verified while concealing major changes in classification relativities, policy fees, deductibles, and renewal impacts. The scientific minimum is therefore a trace from experience and projected cost to the implemented policy-level algorithm and the legal decision made about it.
The second finding concerns adequacy and fairness. These objectives occupy different evidentiary domains: prospective expected loss cost, solvency and expense capacity, compliance with classification law, distributional rate effects, and consumer outcomes. A regulator may rationally accept a cost estimate but require a change to a proposed factor, or accept a classification scheme while questioning the assumed inflation trend. The relationship among these decisions is conditional on law and data, not captured by a single fairness or accuracy score.
The third finding concerns time. The same filing can have an application date, actuarial experience cutoff, selected projection period, approval date, permitted-use date, policy effective date, accident period, and eventual development of losses. Every aggregate percentage reported in research should carry the date to which it pertains. This finding is especially important in motor insurance because bodily injury claims can develop over lengthy periods and new safety and repair technology may alter both frequency and severity.
The fourth finding concerns supervisory architecture. California's public prior approval, Ontario's mandated rate/classification approval, New York's time-sensitive flex-rating arrangements, EU freedom of tariffs, and UK conduct-based pricing rules create different administrative records. Comparative researchers cannot treat an EU absence of approved rate filings as an absence of supervision, nor a larger U.S. count of approvals as proof of superior consumer outcomes. The institutional object being observed changes across systems.
The fifth finding concerns transparency and reproducibility. Public access to filings and publicly reported value measures are complementary, but they reveal different sides of the pricing process. California's accessible rate applications expose proposed reasoning; Ontario's rate-approval records show supervisory decisions; the FCA's value measures reveal selected realized product outcomes. A comprehensive account of pricing governance needs to connect the evidence of rate formation to the evidence of experienced coverage and claims performance, while recognizing legally protected information and data access limits.
The sixth finding concerns regulatory adaptation. Changes in legal injury remedies, anti-fraud measures, permitted factors, or renewal rules can affect expected loss and the lawful path for pricing. New York's 2026 circular letter is a particularly clear example of a regulator requiring insurers to consider expected loss savings alongside prospective premium increases. The correct empirical response is to estimate the anticipated mechanism, disclose uncertainty, and compare later data with the original assumptions.
Practical implications for scholarly and supervisory assessment
The following practices follow analytically from the verified materials, although the binding legal content of each varies by jurisdiction. A file or research dataset should record the product and legal entity, rate-version identifiers, loss and exposure definitions, historical rate-level adjustments, prospective scenarios, review communications, final disposition, and implementation dates. It should distinguish the actuarially indicated change from the requested and approved changes. Policy-level simulations should retain the reference portfolio and quantify distributional consequences. Later experience comparisons should define exposure and accident period, identify shifts in coverage or claims process, and document whether the outcome could reasonably be attributed to a filing intervention.
Actuarial sign-off needs to be intelligible to a legally trained reader; legal compliance needs to be traceable to the precise rate factors reviewed by the actuary. Both disciplines benefit from reproducible input definitions, transparent changes, and a shared chronology. The underlying ASOP No. 53 guidance supplies a professional reference for future-cost work. The NAIC's SERFF infrastructure supplies a documentary channel in participating U.S. jurisdictions. Neither by itself proves the consumer effect of a completed filing.
Open research questions suitable for separate empirical studies
A first research program could compare approval time with changes in insurer participation and policyholder premiums, using insurer-by-state panels and statutory transition dates. It would need to account for nonrandom insurer entry, catastrophe exposure, claim-cost inflation, and the direction of causality: distressed markets may generate both more filing scrutiny and higher premiums. A second could study whether public-intervenor participation changes selected rate assumptions or individual impact distributions, separating contested applications from those voluntarily amended before review.
A third could examine whether outcome-focused oversight produces different policyholder outcomes from tariff approval after controlling for product scope, market competition, liability law, and claims inflation. Such a design would require an agreed outcome definition, not simply binary coding of prior approval. A fourth could evaluate how new telematics and vehicle technology affect model stability and proxy discrimination over consecutive renewal cohorts, with clear privacy and legality constraints. A fifth could construct linked insurer-level data combining requested rate changes, approved changes, implemented renewal distributions, and later claims performance. Its primary scientific obstacle would be the difference between the public filing record and confidential operational data.
These research questions use existing actuarial, causal-inference, and regulatory-analysis methods. Their results cannot be inferred from the present documentary corpus. A future study would need specified populations, predeclared outcome definitions, source permissions, sensitivity analysis, and independent reproducibility checks.
Conclusion
Motor Insurance Rate Filing and Supervisory Review establishes rate filing as a disciplined chain from a defined exposure and legal insurance promise, through observed and projected claims costs, to a versioned rate calculation, an applicable supervisory decision, and observable consumer outcomes. Primary materials from California, New York, Ontario, the European Union, and the United Kingdom demonstrate that supervision of insurance prices takes institutionally different forms. Under American and Canadian prior-approval systems, a filed actuarial case is connected directly to permission to use specified rates. Under EU tariff freedom and UK conduct-based regulation, oversight addresses relevant pricing behavior, solvency, and outcomes through other mechanisms.
The central scientific conclusion is that every substantive judgment requires its own evidence. Prospective rate adequacy depends on credible losses, expenses, trends, and financial assumptions. Rating fairness depends on lawful variables and distributional treatment. Supervisory legality depends on the precise regime and effective date. Consumer outcomes depend on actual policies, claims, and market behavior after implementation. The numerical illustration demonstrates how material trend selections change the indicated rate while leaving the formal calculation consistent. The comparison of legal sources further demonstrates that a published regulatory reform may affect future filings without yet having entered into force.
The monograph's original contribution is an integrated comparative interpretation that clarifies the unit of review, identifies the inferential limits of public filing and supervisory datasets, and connects actuarial traceability to institutional legal accountability. It records a completed evidence-grounded research synthesis with explicit limits on proprietary-data access and causal inference. The resulting manuscript is suitable for independent editorial appraisal as a scholarly Research Dossier; its release remains a distinct technical and publishing decision.
References
Actuarial Standards Board. (2005; revised 2011). Actuarial Standard of Practice No. 12: Risk Classification (for All Practice Areas). https://www.actuarialstandardsboard.org/asops/risk-classification-practice-areas/
Actuarial Standards Board. (2016). Actuarial Standard of Practice No. 23: Data Quality. https://www.actuarialstandardsboard.org/asops/data-quality/
Actuarial Standards Board. (2017). Actuarial Standard of Practice No. 53: Estimating Future Costs for Prospective Property/Casualty Risk Transfer and Risk Retention. https://www.actuarialstandardsboard.org/asops/estimating-future-costs-prospective-propertycasualty-risk-transfer-risk-retention/
Boucher, J.-P., Denuit, M., & Guillén, M. (2007). Risk classification for claim counts. North American Actuarial Journal, 11(4), 110–131. https://doi.org/10.1080/10920277.2007.10597487
California Department of Insurance. (n.d.). Rate Filing Review Process. https://www.insurance.ca.gov/0250-insurers/0800-rate-filings/rate-filing-review-process.cfm
California Department of Insurance. (n.d.). Rate Filing Applications. https://www.insurance.ca.gov/0250-insurers/0800-rate-filings/rate-filing-applications.cfm
California Department of Insurance. (n.d.). Rate Filings: Information, Filing Access, and Electronic Submission. https://www.insurance.ca.gov/0250-insurers/0800-rate-filings/
California Department of Insurance. (n.d.). Virtual Viewing Room. https://www.insurance.ca.gov/0250-insurers/0800-rate-filings/0050-viewing-room/index.cfm
California Legislature. (2024 codification). California Insurance Code § 1861.05: Approval of Insurance Rates. Reproduced by Justia. https://law.justia.com/codes/california/2024/code-ins/division-1/part-2/chapter-9/article-10/section-1861-05/
California Legislature. (2024 codification). California Insurance Code § 1861.07: Public Inspection. Reproduced by Justia. https://law.justia.com/codes/california/2024/code-ins/division-1/part-2/chapter-9/article-10/section-1861-07/
California Legislature. (2025 codification). California Insurance Code § 1861.02: Automobile Insurance Rating Factors. Reproduced by Justia. https://law.justia.com/codes/california/code-ins/division-1/part-2/chapter-9/article-10/section-1861-02/
Cohen, A., & Dehejia, R. H. (2004). The effect of automobile insurance and accident liability laws on traffic fatalities. Journal of Law and Economics, 47(2), 357–393. https://doi.org/10.1086/422978
Dionne, G., & Vanasse, C. (1992). Automobile insurance ratemaking in the presence of asymmetrical information. Journal of Applied Econometrics, 7(2), 149–165. https://doi.org/10.1002/jae.3950070204
European Insurance and Occupational Pensions Authority. (2021 rulebook text). Solvency II Directive 2009/138/EC, Article 21: Policy Conditions and Scales of Premiums. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-2105_en?prefLang=bg
European Insurance and Occupational Pensions Authority. (2021 rulebook text). Solvency II Directive 2009/138/EC, Article 154: Prior Notification and Prior Approval. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-2241_en
European Insurance and Occupational Pensions Authority. (2021 rulebook text). Solvency II Directive 2009/138/EC, Article 181: Non-life Insurance. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-2268_en
European Parliament and Council. (2004). Council Directive 2004/113/EC implementing the principle of equal treatment between men and women in the access to and supply of goods and services. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A32004L0113
Financial Conduct Authority. (2026, handbook version retrieved November 2026). Insurance: Conduct of Business Sourcebook, ICOBS 6B: Home Insurance and Motor Insurance Pricing. https://handbook.fca.org.uk/handbook/icobs6b
Financial Conduct Authority. (2026, September update). General Insurance Value Measures Data 2025. https://www.fca.org.uk/data/general-insurance-value-measures-data-2025
Financial Conduct Authority. (2026, October update). General Insurance Value Measures: Reporting Requirements and Purpose. https://www.fca.org.uk/firms/general-insurance-value-measures
Financial Services Regulatory Authority of Ontario. (n.d.). Auto Insurance Rate Approvals. https://www.fsrao.ca/consumers/auto-insurance/understanding-auto-insurance-rates/auto-insurance-rate-approvals
Financial Services Regulatory Authority of Ontario. (2025, based on data through December 31, 2024). Ontario Private Passenger Automobile Annual Review. https://www.fsrao.ca/industry/auto-insurance/regulatory-framework/guidance-auto-insurance/ontario-private-passenger-automobile-annual-review-based-industry-data-december-31-2024
Financial Services Regulatory Authority of Ontario. (2026, published guidance with effective date undetermined as checked). Automobile Insurance Rating and Underwriting Guidance. https://www.fsrao.ca/industry/auto-insurance/regulatory-framework/guidance-auto-insurance/automobile-insurance-rating-and-underwriting-guidance
National Association of Insurance Commissioners. (n.d.). System for Electronic Rates & Forms Filing (SERFF). https://content.naic.org/industry/serff
New York Department of Financial Services. (n.d.). Property Insurers: Rate and Form Filing Instructions. https://www.dfs.ny.gov/apps_and_licensing/property_insurers/rate_form_filings
New York Department of Financial Services. (2026, July 1). Insurance Circular Letter No. 3 (2026): Motor Vehicle Insurance Reforms. https://www.dfs.ny.gov/industry-guidance/circular-letters/c32026-01
New York State Senate. (current text viewed November 16, 2026). Insurance Law § 2303: Standards for Rates. https://www.nysenate.gov/legislation/laws/ISC/2303
New York State Senate. (current and historical texts viewed November 16, 2026). Insurance Law § 2305: Rates or Rating Plans. https://www.nysenate.gov/legislation/laws/ISC/2305
New York State Senate. (current and future-effective texts viewed November 16, 2026). Insurance Law § 2350: Flexible Rating for Nonbusiness Automobile Insurance Policies. https://www.nysenate.gov/legislation/laws/ISC/2350
Ontario Legislature. (as in force at research date). Automobile Insurance Rate Stabilization Act, 2003, S.O. 2003, c. 9. https://www.ontario.ca/laws/statute/03a09
U.S. Bureau of Labor Statistics. (n.d.). Measuring Price Change in the CPI: Motor Vehicle Insurance. https://www.bls.gov/cpi/factsheets/motor-vehicle-insurance.htm