RESEARCH DOSSIER
Claim Notification Time Metrics and Timeliness Governance
Abstract. Claim notification is the institutional transfer of evidence about a potentially insured event or entitlement, and timeliness governance organizes the responsibilities and controls that determine how this information becomes actionable. This monograph distinguishes occurrence, discovery, notice dispatch, receipt, registration, acknowledgment, substantive first action, proof, decision, and performance. Its method combines comparative documentary analysis of insurance statutes, conduct rules, supervisory standards and industry codes, event-history statistics, actuarial reporting-delay research, a regulatory review of Australian home insurance claims, and constructed numerical examples. The analysis separates duration measurement from legal consequence, including the conditional significance of notice prejudice, and distinguishes right censoring of observed claims from truncation of unreported losses. It examines access, clock provenance, competing events, capacity, extensions, external service providers, performance incentives, and claimant outcomes. The resulting synthesis explains why reliable timeliness governance depends on evidence of the actual triggering and completing events, correct denominator design, context-specific legal interpretation, and accountability for both speed and fairness. Findings are a comparative and methodological synthesis; no original empirical effect estimates or new doctrine are asserted.
Keywords: Claim notification; first notice of loss; reporting delay; discovery lag; acknowledgment; first substantive action; claims timeliness; notice prejudice; claims-made insurance; conduct supervision; event history; right censoring; right truncation; cumulative incidence; insurance claims operations; timestamp audit; claimant outcomes.
Introduction and Research Design
Claim notification is the institutional transfer of information about a potentially insured occurrence, loss, allegation, or entitlement to a recipient empowered to receive it. Timeliness governance is the arrangement of responsibilities, measurement rules, communication, escalation, and review through which an insurer and other participants manage the time between relevant events. The two concepts meet at the moment a claim becomes actionable within an organization. That moment often differs from the date of an underlying accident and from the timestamp at which a claim record is created.
The scientific problem is the relation between elapsed time, knowledge, legally defined obligation, evidence availability, and claimant outcomes. A measure of elapsed time acquires a clear interpretation only through its starting event, terminal event, unit of observation, applicable clock convention, eligibility rule, and treatment of uncertain or incomplete observations. A report received quickly may remain unattended. An old accident may be reported promptly after discovery. A claim may be acknowledged within a prescribed period while an important inquiry is still outstanding. Each finding refers to a different transition.
Three questions organize this study. First, how should event-to-notice, discovery-to-notice, receipt-to-acknowledgment, and first substantive action be separately measured? Second, when and how does delay influence the legal effect of notice, including the relationship between late notice and prejudice? Third, what evidence would establish whether claims-timeliness interventions improve speed, fairness, and accountability?
The study uses comparative legal and supervisory documentary analysis, selected actuarial and statistical literature, a regulatory case study, and reproducible illustrative calculations. The legal documents are analyzed as particular instruments with identifiable jurisdiction, version, scope, and legal force. The statistical methods are applied to hypothetical observations to expose denominator errors, selection, censoring, and competing outcomes. The Australian regulatory study is treated as published evidence, with its own sampling and reporting boundaries. No insurer microdata were obtained, no consumer records were processed, and no original effect sizes or market estimates are asserted. The legal examples are documentary comparisons, not individualized legal advice or a complete survey of current national law.
The parent discipline is Claims Science. The dossier develops the narrow scientific object of notification and early-response time rather than reproducing the parent monograph's broad treatment of indemnity, reserves, settlement, or fraud. Its contribution is an integrated interpretation of existing law, actuarial reasoning, event-history analysis, and operational governance. It does not present a new theory, named model, or doctrinal proposal.
Historical Development and Institutional Setting
The institutional history of claims handling can be understood as an expansion in the number of parties entitled to observe and act on a reported loss. A claimant had to communicate the fact of an event to a person capable of investigating or paying it. The development of specialized adjusters, intermediaries, delegated administrators, and internal claims functions introduced additional transmission and assignment events. Digital intake adds new channels, automated confirmations, and system integrations; the central scientific distinction between receipt and knowledge remains consequential regardless of medium. This is an interpretation of organizational development, rather than a chronology claiming dates for the invention of particular practices.
Modern conduct supervision treats the claims process itself as a subject of governance. The IAIS (2024) ICP 19 claims-handling standard and associated guidance connect timeliness with fairness, transparency, documented steps, and claimant information. The standard addresses the responsibilities of supervisors; its realization depends on the institutional and legal framework of a jurisdiction. It does not create one international numerical deadline.
The United States illustrates the role of jurisdiction-specific instruments. Under the NAIC (1997) Model 900, acknowledgment of material communications and standards for investigation and settlement form elements of the model treatment of unfair conduct. Sections 3 and 4 must be read together: the model's prohibited-practice formulation has flagrant-disregard or general-business-practice conditions. A model law is a model; an enacted rule, available remedy, and its scope require separate jurisdictional determination.
The United Kingdom provides a more principle-oriented comparator. FCA ICOBS 8.1.1R, effective from 2008, combines the handling of claims and provision of information with the treatment of rejection and agreed settlement. The institutional logic is that a claim should advance in a manner comprehensible to the claimant and consistent with legitimate investigation. Numerical monitoring may help an insurer demonstrate practice under such an obligation, but its internal target cannot exhaust the regulatory judgment.
The development of actuarial event data provides a different historical perspective. An insurance portfolio can contain losses that have already occurred but have not yet reached the insurer. Continuous-time and cohort-based reporting-delay research developed to support estimation of this unseen population. Verrall and Wüthrich (2016) connect individual event and reporting records with estimates of claims not yet reported. Conduct regulation and actuarial reserving thus approach a shared passage of time from different evidentiary positions: one concerns obligations and treatment; the other concerns incomplete knowledge of a portfolio.
Claim Clocks and Units of Observation
Event, discovery, notification, receipt, and registration
A loss-generation event is the physical, legal, medical, or other occurrence relevant to a possible entitlement. For some products it has a relatively clear time; in gradually developing damage or occupational disease the available observation may be a range, and a trigger may depend on policy language. Discovery is the claimant's acquisition of relevant knowledge. Actual and constructive knowledge, when legally relevant, must be separately represented; the system cannot infer a legal conclusion merely from an event date.
Notice is a communication that reports the relevant circumstance or demand. Sending and receipt are distinct: a letter may be sent on one day and arrive later; a transfer by a broker may follow the claimant's initial contact. Receipt is the event at which the applicable authorized recipient receives the communication. Record creation, validation, and assignment are operational events. Treating record creation as a substitute for receipt transfers intake delay out of the measured insurer interval.
Acknowledgment is a communicated confirmation of receipt, potentially accompanied by a claim reference and assistance. A technical delivery receipt, a system-generated ticket, and acknowledgment of the claim are separately observable. A substantive first action is an action with a defined relationship to advancement of the claim: assessment, evidence-specific guidance, urgent assistance, investigation, or another qualified step. Assignment without action cannot demonstrate that the claimant's substantive needs were addressed. An automated action may qualify if its actual function and correct delivery satisfy the definition. Separate the dispatch, delivery, and substance of the communication.
Temporal identity and measurement
Let E be the occurrence time, D the discovery time, S the first notice sent, R the first relevant receipt, C the administrative creation time, A the acknowledgment time, and F the first qualifying substantive action. Let P denote proof or information completeness as defined by a particular rule, J the decision communicated, and M the first payment or performance event. Individual clocks are:
- E to D: occurrence-to-discovery delay.
- D to S: known-loss-to-dispatch delay.
- S to R: communication or intermediary transmission interval.
- R to C: intake registration interval.
- R to A: acknowledgment interval.
- R to F: substantive first-action interval.
- P to J: proof-to-decision interval.
- J to M: decision-to-first-performance interval, subject to any relevant condition.
For a sequence where these events are distinct and ordered, R − E = (D − E) + (S − D) + (R − S). Similarly F − R = (C − R) + (F − C) only when creation precedes the first qualifying action. Some cases involve action before registration, such as a telephone emergency instruction; separate timestamps must preserve that ordering. A and F can occur in either order or in one communication. The identity describes a sequence; it is not a claim that every product follows this sequence.
Three types of time expression matter. Elapsed duration is calculated in seconds or hours on a common timeline; a calendar-day rule is interpreted with the legally applicable date counting; a business-day rule depends on the specific calendar, holidays, cutoffs, and inclusion convention. A weekend may count in a claimant's lived waiting experience while being excluded from a particular business-day obligation. A defensible system stores both unadjusted elapsed time and the independently computed regulatory clock.
Claim identity and multiple clocks
The unit of observation must be chosen before computing a distribution. A single physical loss may give rise to several policies, coverages or injured parties. A single policy claim may have multiple subclaims or separate proofs. A catastrophe event may generate thousands of claim episodes. An insurer might record one claim per beneficiary or group several invoices in one case. A first notice may report an incident before a legal third-party demand exists. These units should be linked rather than collapsed.
A primary metric for intake measures first receipt per eligible claim episode. A separate event-level measure counts time from catastrophe occurrence to the first report for the affected unit. A coverage-level measure begins at notification of that coverage if that is the scientifically defined start. Denominator counts and joins must prevent duplicates and avoid disappearing repeated contacts. A reopening is linked to the original claim and to a new transition; closing and reopening an administrative case does not erase the first episode's age or legal receipt.
A portfolio measure is typically grouped by accident date, reporting date, or valuation date. Each grouping answers a different question. Reporting-month throughput may be useful for staffing; accident cohorts are essential for identifying delays and reporting incompleteness; due-date cohorts support regulatory controls. Neither a timely claims reserve nor a service completion figure alone demonstrates whether all legitimate individual entitlements received timely consideration.
Reporting Lag Decomposition and Evidence Integrity
Sources of delay and attribution
The sequence of occurrence, discovery, sending, and receipt separates delays that are routinely confounded. A latent leak may be known only after visible damage; an injured person may be incapacitated; a family may lack documents or an accessible communication channel; a report may be left with a representative and reach the insurer later. A claim may then wait for identity verification, coverage confirmation, priority assignment, or review. Responsibility for each interval is an evidentiary question. An attribution category describes an established operational state or actor; it does not itself establish legal fault.
A useful claim timeline separates insurer queue time, active processing time, external service time, claimant response time, and communication gaps. Parallel processes create overlapping intervals. An assessor's work may continue while a form is being completed. Summing all such intervals would double count elapsed duration. For the total span, compute the union of intervals; for workload, count active labor separately. Promised callbacks and regulatory update duties may continue even when an external task is in progress.
For each recorded pause, require start and end events, reason, responsible owner, evidence supporting the reason, rule authority if the pause changes a contractual or statutory clock, and the next obligation. A status field such as awaiting customer does not establish that an authorization to extend or suspend a deadline exists. Evidence of a delivered, intelligible, relevant request matters. A repeated request for information already present in the file requires a different explanation from a necessary new request. Gray areas of concurrent responsibility remain marked uncertain rather than assigned for convenience.
Timestamp audit design
The essential event record contains an episode identifier; event type; occurrence timestamp or bounded interval; recording timestamp; original timezone and offset; channel; source artifact; authorized recipient; actor and role; version and correction record; relationship to previous event; and evidence quality. A timestamp audit distinguishes real-world time from transaction time. An event's original record is retained even if a later investigation corrects its effective time; the correction has a separate timestamp and provenance.
Time zone conversion must preserve local legal context. UTC normalizes elapsed hours across systems; local dates may be essential for business-day rules. Ambiguous times near daylight-saving changes require an explicit offset or an uncertainty flag. For date-only observations, store the date precision; midnight is not a measured hour. For an unknown occurrence date, interval bounds may come from the last known intact condition and first known damage, without imputing one point as an observed event.
Consider a loss discovered Tuesday between 08:00 and 12:00 with insurer receipt at 15:00, all in the same timezone. Discovery-to-receipt lies between three and seven hours. If a hypothetical policy uses a six-hour notice interval, that evidence alone does not settle compliance. A six-hour breach cannot be established from the worst bound, and compliance cannot be asserted from the best bound. Further evidence or a qualified conclusion is required.
Audit samples should compare structured timestamps with message delivery metadata, contemporaneous notes, call logs, scans, and the actual text of acknowledgments. Select a stratified probability sample of cases by source channel, product, recorded lateness, geography, and status, deliberately oversampling rare and high-risk cases. Apply inverse sampling-probability weights for overall error rate estimates; report separate high-risk findings unweighted where that is their defined population. Include open cases, voided records, abandoned attempts where lawfully documented, and manually corrected records. An extract limited to accepted claims omits important intake failures.
Measure the absolute difference between two timestamp sources, exact match at their common resolution, disagreement in deadline classification, missingness by field, invalid temporal ordering, and missing event artifact. Report both numerator and denominator. A mismatch in source-clock timestamps does not prove misconduct; it triggers reconciliation. Missingness should be described by channel and claimant circumstance, since the probability of missing data may be related to the outcome.
Data controls and minimization
For routine performance analytics, the minimum dataset needs pseudonymous identifiers, event categories, relevant timestamps, product and jurisdiction, exposure or cohort identifiers, status, and sufficiently coarse explanatory variables. Full medical information, narrative evidence, identification documents, and other sensitive records can remain in authorized claim systems. Access-controlled joins for an audited subset may be necessary for attribution, but the aggregate analysis should avoid unnecessary duplication of sensitive evidence.
Separate append-only evidence history, operational state, and reporting extracts. Build data lineage from channel logs to claim record to dashboard. Reconcile counts and dates during system migration; keep imported retrospective events distinguishable from new communications. Version the measure and the applicable rule and record the extraction date, because historical results can change when previously missing reports arrive. A later correction should produce a documented restatement, rather than silently rewriting historical compliance figures.
Measurement of Timeliness and Distributional Performance
Descriptive measures and denominator design
For an observed interval T, the empirical distribution of completed durations is a concise description of the recorded completed cases. Mean, median, and percentiles answer different questions. The mean is sensitive to a long tail; the median can remain stable while a minority experiences prolonged delays; upper quantiles and aged inventory make that minority visible. State the quantile convention and precision. A single percentile cannot establish legal compliance or adequacy of individual treatment.
A threshold measure for a fixed mature cohort is the number of eligible intervals ending by the specified threshold divided by all eligible intervals with known status at that threshold. A case still waiting at the threshold is a miss for that threshold, even if it later completes. Cases observed for less than the threshold with no completion are not yet classifiable. Report their count and the reason they are excluded or use an appropriate event-history approach. Where an eligible case's start or end is disputed, report a separate uncertain classification with sensitivity bounds.
A due-date compliance measure differs from a process target. The denominator comprises cases to which a particular legal rule applies and for which the due time is established. If N such cases include k timely, m late, and u genuinely unresolved classifications, N = k + m + u; the observed-case proportion is k/(k+m), and the bounds over the full eligible denominator are k/N and (k+u)/N. If N = 100, k = 80, m = 10, u = 10, the observed-case percentage is 88.9%; overall compliance lies between 80% and 90% before further information. Publishing the 88.9% alone hides measurement uncertainty.
An alternative is the due-date cohort: include each obligation due within a reporting period, then assess whether completed by its deadline. That metric is insensitive to how recently a file was opened, but it may lag operational diagnosis. In addition, display receipt cohorts, time-specific inventory, overdue count and age distribution, and the number of pending obligations with no due date determined. A dashboard should preserve the separate populations rather than averaging their percentages.
Censoring, truncation, and competing outcomes
Suppose all claims enter observation at first receipt, and the target is first substantive action. A claim with no action by the extraction date is right-censored at its current duration. A data set containing open and completed claims can estimate the waiting-time survival function under a suitable independent-censoring assumption. Kaplan and Meier's (1958) product-limit estimation provides the classical treatment of incomplete duration observations. For ordered distinct action times tⱼ, with Yⱼ cases at risk just before tⱼ and dⱼ qualifying actions, the estimator is
Ŝ(t) = ∏ over tⱼ ≤ t of (1 − dⱼ/Yⱼ).
Consider six hypothetical claims, each under observation from receipt. At age 1 day, two receive their first action; one case is censored at day 2; one receives action at day 3; one at day 5; and one is censored at day 6. The risk set at the action times is 6, 3, and 2. Estimated survival after day 1 equals 4/6 = 2/3; after day 3 it equals (2/3)(2/3) = 4/9; after day 5 it equals (4/9)(1/2) = 2/9. The estimated median age to action is 3 days; the complement by day 3 is 5/9, or 55.6%. The raw proportion with action by day 3 is 3/6 = 50%; the censoring assumptions explain the difference. No inference about an actual insurer follows from these numbers.
If closure or withdrawal definitively precludes the first action being measured, it is a competing event. Austin, Lee, and Fine (2016) explain in their competing-risks primer why one minus the conventional Kaplan–Meier survival estimate is unsuitable for crude event-specific incidence in this situation. For a transition type k, a standard cumulative incidence estimator sums Ŝ(tⱼ−) times dₖⱼ/Yⱼ, where Ŝ uses all terminal event types. If a withdrawal can be reopened, define the first episode and treat subsequent transitions as separate states rather than assume an absorbing event without justification.
In a hypothetical cohort of 100 received claims, 20 withdraw before any substantive action at day 1; 40 of the remaining 80 receive action at day 2; the remaining 40 are observed through day 3 with none receiving action. Cumulative incidence of first action at day 2 is (80/100)(40/80) = 40%. Treating withdrawals as simple censoring would give a 50% complement of Kaplan–Meier, thereby estimating a different quantity. Withdrawal should not be portrayed as proof of satisfactory service, particularly if the initial report never received meaningful help.
Reporting delay measured from occurrence creates another difficulty. A database of reported claims observed at date v includes losses only if R ≤ v. For a claim from occurrence date e, a delay longer than v − e prevents its appearance. This is right truncation of the recorded reporting-delay sample, unlike the right censoring of already observed receipt cohorts. Treating unavailable reports as open records in an insurer's claim system invents observations that do not exist. Where the underlying losses are independently enumerated, delayed claims can be followed as a cohort and censored; otherwise a model of reporting and incidence or a truncation-corrected analysis is required.
Modeling considerations
A discrete daily hazard can estimate the probability of a transition on a day, conditional on remaining at risk at its start. Explanatory factors may include event type, channel, calendar, geography, observed complexity, and current queue load. Results depend on correct temporal alignment: an investigation flag created after the first-action target cannot serve as a baseline predictor. A model should retain event and report calendars separately because weekends or catastrophe changes may affect both arrival and processing mechanisms.
Continuous-time proportional-hazards analysis requires an examination of proportionality and the meaning of covariates. Where hazards vary over time, time-varying covariates or flexible time interactions may be more appropriate. Cause-specific models answer questions about transition rates among those still at risk; cumulative-incidence analysis addresses the absolute probability of outcomes in the presence of other transitions. The modeling choice follows the stated research question.
Report uncertainty with intervals suitable to the design. A simple independent binomial standard error can be illustrative for a mature binary outcome under independent sampling, but related claims from one catastrophe, household, or intermediary are clustered. Cluster-respecting resampling or robust inference should be considered; effective sample size is smaller when records are strongly dependent. A fixed threshold plus an uncertainty interval should not conceal an unknown denominator or measurement bias.
Actuarial Reporting Delay and Portfolio Information
Observation of aggregate liabilities
For a cohort with n expected eventual reportable losses and distribution F(a) of reporting delay, the expected number observed by age a is nF(a), provided the assumptions defining n and the reporting distribution are coherent. The expected unreported count is n[1 − F(a)]. This relationship is elementary conditional expectation, not a proposed reserving method. Claims not yet reported and outstanding amounts on already reported claims are distinct components of claim development.
Verrall and Wüthrich's (2016) reporting-delay study calibrates a continuous-time framework on property and casualty claim records, considers changes in reporting over time including seasonality, and compares individual-data predictions with aggregated Poisson chain-ladder predictions. It demonstrates the relevance of the reporting mechanism to estimation of counts not yet observed. This dossier uses that methodological insight to interpret the intake process, without importing the paper's empirical parameters.
A simple illustration uses 200 hypothetical reported losses in an occurrence cohort evaluated at age ten days. Under an assumed reporting fraction of 0.80, the plug-in expected total is 200/0.80 = 250, leaving 50 unreported. If the reporting fraction were instead 0.60, the corresponding values would be 333.3 and 133.3. The quantities are model-based expected counts; fractional expected counts are acceptable. They are sensitive to the reporting assumption and cannot by themselves produce monetary reserves.
If a change in reporting channels raises the fraction reported by ten days from 0.60 to 0.80 without changing ultimate frequency, recorded early claims rise by one third. A naive comparison of reported counts could interpret better visibility as higher loss incidence. Conversely, an inaccessible intake channel may lower observed frequency without lowering physical losses. A study of frequency therefore needs exposure, accident cohorts, reporting development, and versioned channel policy.
The linkage to severity is similarly conditional. If minor claims report quickly and serious latent claims report later, the mean severity of early reports need not represent ultimate severity. The direction can reverse when urgent high-loss events are preferentially reported. A count adjustment multiplied by a common mean cost relies on a specific severity–delay assumption. Monitoring claim-level timing alongside portfolio development exposes this assumption without replacing actuarial valuation.
Changes in reporting and validation
Policy renewal, geographic expansion, disaster conditions, claim-channel redesign, fraud screening, and changed record systems may alter the observed relationship between occurrence and receipt. Some changes affect actual event incidence; others affect the probability or time of report; others affect timestamp definitions. A simple comparison cannot identify which mechanism operated.
Validation should split by accident cohort or calendar time, retaining sufficiently mature evaluation periods. Track observed versus expected development at several ages, calibration by product or event class, and errors for high-report-delay segments. Prospective validation requires reconstructing data available at each historical valuation, not using later corrections known only in retrospect. A developing catastrophe may invalidate normal calendar assumptions; scenario sensitivity should preserve the difference between expected frequency and delay.
The actuarial use of notification information has a different objective from claims conduct monitoring. A portfolio model may be accurate in aggregate despite biased individual processing, while a high acknowledgment rate may coexist with material unreported liabilities. Both perspectives benefit from reliable event records, but neither substitutes for the other.
Notice Duties and Comparative Legal Consequences
Claimant duties and the legal trigger
The legal significance of delayed notice depends on the instrument and contract, the triggering event, the identity and knowledge of the obligated party, the nature of the notice clause, and the consequence sought. A time metric can establish a factual interval; a coverage determination requires legal interpretation. For example, a prompt-notice covenant designed to preserve opportunities for investigation has a different function from a date-certain condition defining the coverage period. An operational late-notice label should never automatically be transformed into a coverage result.
France's Code des assurances, Article L113-2, in the version in force from April 2018 displayed for November 2024, relates notice to knowledge of a potentially covered loss and to the contractual period. The ordinary minimum specified period is five working days, with shorter minimums of two working days for theft and 24 hours for livestock mortality. The relevant contractual forfeiture for late notice depends on proof that the delay prejudiced the insurer, and the text protects cases of fortuitous event or force majeure. The relevant obligations exclude life insurance. This comparison concerns the examined provision, not every special product rule or later judicial interpretation.
Germany's VVG § 30 identifies prompt notification after knowledge and addresses a third party entitled to the benefit. It restricts reliance on an agreed loss-of-benefit consequence for breach of the specified notice duty if the insurer acquired timely knowledge by other means. The nearby VVG § 28 differentiates consequences of contractual-obligation breach under its stated conditions, including intention, gross negligence, causal relevance, and fraudulent conduct. The provisions have distinct conditions; neither should be reduced to a universal deadline or a blanket rule that all delay is harmless.
In the December 2022 version of Japan's Insurance Act, Article 14, in the non-life chapter, notification follows awareness of loss arising from an insured accident and is to be given without delay. This documentary example shows again why a discovery-related trigger may be more informative than elapsed time from occurrence. The study does not infer the consequence of breach or current Japanese case law from Article 14 alone.
The investigation interest and prejudice
A notification duty can protect an insurer's opportunity to inspect damage, identify witnesses, preserve records, defend a liability claim, control further loss, pursue recovery, or evaluate the appropriate benefit. The importance of an opportunity varies with the case. A delayed report may leave reliable digital records intact; an early report may concern a scene already altered before the insured obtained knowledge. The question of prejudice therefore requires more than the measurement of duration.
A disciplined analysis identifies the opportunity allegedly lost, its timing, the available alternatives, the causal relationship with the delay, and the materiality of its consequence under the applicable law. Evidence may include dated photographs, contemporaneous documentation, expert observations, the availability of witnesses, earlier third-party information, and insurer action after actual receipt. If a specific investigative opportunity remained available but was not used by the insurer, that fact is relevant to the causal assessment. The applicable legal burden of establishing prejudice or its absence must be separately determined; the analytic sequence does not assign that legal burden universally.
The Supreme Court of Colorado in Craft v. Philadelphia Indemnity (2015) distinguished occurrence-policy prompt-notice provisions from a date-certain reporting condition in a claims-made policy. Its refusal to extend the notice-prejudice rule to the latter arose from the function of that date-certain condition in defining the scope of coverage. This holding illustrates why the timing of the claim made, the timing of notice, and the relevant reporting period must be distinct fields in a liability data system. It also demonstrates the limits of transplanting a prejudice doctrine across policy types and jurisdictions.
Suppose an occurrence-policy insured discovers a loss five days after the event and immediately sends notice. A recorded event-to-notice time of five days says little about delay after knowledge. Suppose another policy requires a claim to be made and reported within a particular policy period or permitted extended reporting period. With the same five-day duration, a reporting date crossing a specified boundary may matter independently of investigative difficulty. These are illustrative contract structures only; any entitlement determination depends on the actual text and controlling law.
Separate notification obligations and governance
The insurer, intermediary, claimant, third-party administrator, regulator, and reinsurer can each have a distinct notice obligation. Reporting a claim to an insurance intermediary may or may not satisfy the particular receipt condition; reporting a cyber incident to a regulator may trigger a different legal clock. A reinsurance notice concerns a relationship between insurer and reinsurer and cannot automatically establish the timing of the original claim notification.
An evidence system should therefore record the sender, recipient, legal role, message content, channel, and contract or instrument associated with every notice event. A first notice of a potential incident and a later claim demand are distinct where the product requires that distinction. This arrangement provides a neutral factual chronology for legal review; interpretation remains a separate documented decision. Cross-border or multi-policy claims may carry multiple obligations that have different triggering events and date rules.
Insurer Acknowledgment and Regulatory Timeframes
Comparative evidence matrix
The following limited comparison records the applicable document and the represented obligation, not a universal schedule. The numerical standards are version-specific examples; local legal applicability, exclusions, holidays, and later amendments must be checked before operational use.
| Jurisdiction or institution | Instrument and force | Trigger and selected timing rule | Qualification |
|---|---|---|---|
| International supervision | IAIS ICP 19; supervisory framework | Timely fair transparent claim handling | Expected timeframes and status communication; implementation through jurisdiction |
| United Kingdom | FCA ICOBS 8.1.1R; regulator rule | Prompt handling and settlement; guidance and progress information | No single numerical deadline in this rule |
| United States, California | Title 10 CCR §§ 2695.5 and 2695.7; administrative regulations | Notice to acknowledgment generally within 15 calendar days; proof to decision generally within 40 calendar days | Payment alternative, legal-action notice exception, product and extension provisions |
| United States, Texas | Insurance Code §§ 542.055–542.059; statute | Receipt to acknowledgment/investigation generally by 15th day; complete required proof items to written decision generally 15 business days | Surplus lines, arson, extensions, scope exclusions and catastrophe provisions |
| France | Code des assurances L113-2; statutory provision | Loss known to insured, contract notice period with general statutory minimum five working days | Special minima, prejudice, force majeure, life exclusion |
| Germany | VVG §§ 28 and 30; statutory provisions | Notification promptly after knowledge | Consequences governed by specified statutory conditions |
| Japan | Insurance Act art. 14; non-life statutory provision, dated text | Notice after awareness of loss without delay | No numeric limit specified in examined provision |
| Australia | General Insurance Code of Practice, October 2023 version; industry code | Specified early information steps, periodic updates, separate decision clocks | Scope and exceptions of Code; distinct licensing obligations |
The matrix compares functions and clock triggers, rather than ranking legal systems by nominal days. A statutory notice condition on the claimant cannot be compared directly with an insurer acknowledgment deadline or an expected period after completed proof. The row definitions are supported by the instruments discussed in this and the preceding section.
Under California § 2695.5, a licensee or claims agent must immediately transmit notice to the insurer; subsection (e) establishes an outer fifteen-calendar-day period from notice for specified steps, with an exception for notice of legal action. Receipt must be acknowledged unless payment occurs within the period. An oral acknowledgment requires a dated record, and the text includes a qualified imputation provision regarding an agent's failure to transmit. An operational clock that starts upon creation in a central claim system may therefore be misaligned with receipt through the relevant intermediary.
California § 2695.7 establishes a generally forty-calendar-day decision standard following receipt of proof of claim, with enumerated exclusions. When additional time is needed under subsection (c), written reasons and further information requirements are specified, with renewed notices generally every thirty calendar days until the defined terminal event. Subsection (d) requires diligent, objective investigation and addresses immaterial information requests. Payment and exceptional fraud provisions are separately stated. A timely acknowledgment is accordingly compatible with an overdue later obligation; the two should be independently audited.
In Texas, Insurance Code chapter 542, subchapter B, § 542.055 distinguishes receipt-based steps from the final-proof-based determination in § 542.056. The ordinary initial deadline is the fifteenth day; eligible surplus lines insurers have a thirty-business-day period. The general proof-to-decision deadline is fifteen business days, with a thirty-day arson provision and a reasoned further-time procedure ending at the forty-fifth day after the extension notice. Under § 542.057, a separate payment clock follows notice or any specified act by the claimant, subject to the provision's qualifications. Section 542.059 extends relevant deadlines by fifteen days for defined weather catastrophe or major natural disaster. These examples refer only to the subchapter's applicable claims and conditions.
The ordinary Australian process in the October 2023 General Insurance Code of Practice illustrates simultaneous clocks. For claims requiring more information or assessment, paragraph 68 establishes early information, appointment where necessary, and timeframe-related steps within ten business days. Paragraph 70 provides a twenty-business-day progress-update interval; paragraphs 76 and 77 use the completion of relevant information and inquiries and overall receipt respectively as different starts. Paragraph 78 provides specified exceptional circumstances with a different twelve-month pathway. The Code is an industry instrument with its own applicability; statutory and licensing obligations have separate authority.
ASIC's June 2025 Information Sheet 253 uses examples to explain obligations for regulated claims-handling services, including the burden of unnecessary information requests and transparent assistance. A claim can be delayed before an asserted proof-completion date by inappropriate demands. A governance system should therefore test the reasonableness and history of requests instead of treating proof completeness as an exogenous date. This conclusion is an application of the document's rationale, not a new legal duty.
Clock implementation and extension governance
A rule engine requires the jurisdiction, instrument identifier, effective version, products and parties covered, trigger event type, time unit, counting convention, exception logic, mandatory communication, evidence requirements, and escalation. A system record of the legal basis should allow reconstruction as of the time of decision. Internal targets are separate configurations and can be tighter than legal outer timeframes; they must not silently redefine the latter.
A nonlegal service example clarifies day counting. Assume an insurer's internal standard of three business days excluding the receipt date, Saturdays, and Sundays, with no holidays or cutoff exception. A receipt on Friday is due on Wednesday. The lived duration spans five calendar dates beyond the receipt date. With a Monday holiday the internal due date shifts to Thursday. A production rule must use the actual jurisdiction and prescribed counting convention, rather than apply this illustrative assumption to every instrument.
Extension events are explicit. Record when an extension is authorized, whether it changes the deadline or simply describes expected service, notice sent and received, reason, new due date, outstanding evidence, and any recurring status obligation. Maintain the original deadline for audit and extraordinary-exception reporting. An extension cannot be inferred merely because an automated task remained open.
For open-and-overdue control, calculate which obligations have matured as of the evaluation date, whether they are satisfied, and whether any extension is valid and evidenced. Investigate missing due dates, especially where the claim is unassigned or proof completion is delayed. A missed mandatory update may coexist with a still-valid decision extension. Keeping those events separate enables supervisors and internal reviewers to see the actual pattern.
Comparative Interventions and Operational Governance
Capacity, triage, and routing
The key operational transitions are receipt to triage, triage to assignment, assignment to substantive action, and ongoing activity toward resolution. Acknowledgment may be automated, but substantive advancement needs a defined task owner, information access, appropriate authority, and resources. Emergency assistance and vulnerability-sensitive support may have a different priority from documents that can safely wait.
Queue analysis provides an established conceptual connection between inventory and time. Little's (1961) queuing identity, under its relevant conditions, relates the long-run average number of units in a system to arrival rate and average residence time. A stable queue receiving 40 eligible claims a day with mean time to first action of 2.5 days would have a time-average inventory of 100 claims in that consistently defined system. This is a steady-state calculation; it is not a prediction for a nonstationary shock. Arrival rate, population, and time boundaries must match.
In a surge illustration, 120 claims arrive per day and 80 qualifying first actions can be performed per day for five days, with an initial backlog of 30 and no other departures. The backlog becomes 30 + 5(120 − 80) = 230. If the arrival rate later returns to 60 per day, and the capacity remains 80, eleven and a half days of net clearance would be required in a fluid approximation, or twelve whole days if counting daily batches until zero. The calculation assumes equal service units, constant rates, and no priority effects or capacity losses. In practice complexity and geographic constraints matter, and the repeated application of priority can starve routine cases unless protected capacity or age escalation is used.
Use workload in terms of estimated service demand and role-specific capacity rather than raw claim counts where feasible. A single serious claim may require more work than many straightforward claims. Staffing must consider catastrophe readiness, sickness, telephony capacity, specialist bottlenecks, external repair networks, and communication obligations. A routing intervention can improve dispatch speed but move waiting time to an external expert. The outcome measure must follow the full relevant transition.
Access and assisted reporting
Several interventions can reduce the burden of initial reporting: clear instructions, multiple accessible reporting channels, assisted entry, language assistance, simplified intake, confirmation of the original receipt time, and continuity when a user changes channels. Their value should be tested through completion rates, effort, evidence completeness, time to substantive action, and representation of claimants who were unable to complete ordinary self-service.
A web form may record an opening session, the first attempted submission, an actual delivery, a failed validation, and a successful case creation. All are useful operational observations. Whether an attempt amounts to legally effective notice requires the relevant law and facts. The system should preserve attempted-communication evidence where lawful and proportionate, especially if an outage prevented proper receipt. Channels cannot be compared solely by speed among completed submissions; exclusion of unsuccessful attempts creates selection bias.
Assisted reporting should not become a source of additional clock resets. If a claimant calls after a failed digital submission and the loss was already received through an authorized channel, the later entry is linked to the first relevant receipt. If the prior attempt never reached the insurer, its separate evidence may explain the access failure. The legal determination and operational response are separately recorded.
Communication, substantive action, and prevention
Acknowledgment should accomplish a meaningful function in the process. It can confirm receipt, reference, contact method, expected next steps, and urgent pathways appropriate to the claim. Progress communications should identify the actual state, outstanding information, responsible party, expected next action, and update channel. Communication alone cannot discharge an unmet substantive commitment; it can reduce avoidable uncertainty while the necessary work proceeds.
A useful first-action definition includes evidence that the activity addressed an immediate or investigative requirement. Examples are a dispatched qualified assessor with a communicated plan, an evidence-specific request, a completed coverage assessment, or emergency support where indicated. A bulk generated update with no claim-specific work belongs to a separate communication metric. The definition must be fixed before evaluating staff or technology, and periodic sampled review should determine whether the recorded action corresponds to the actual action.
A delay-reduction intervention may generate unintended behavior when a target becomes a reward criterion. Employees can improve apparent speed by marking a claim complete before the decision is communicated, creating a second claim record with a new clock, issuing immaterial requests, or moving difficult cases into excluded categories. Controls include independently verified event times, original receipt age, a missing-status report, sampled correspondence, reversal and reopening rates, and measures of claimant burden. Governance aligns performance incentives with legitimate entitlement and documented action.
Outsourcing, complaints, and escalation
An external administrator or adjuster may control the time of an activity while the insurer retains an oversight obligation defined by applicable law and arrangement. Contractual service levels should identify receipt and assignment times, expected work product, communications, exception handling, and delivery artifacts. Performance evidence must reconcile both parties' clocks. A vendor turnaround measure starts at accepted assignment; claimant waiting begins earlier and may continue after the vendor report is completed.
A complaint can be a later communication about an unresolved claim, separate from the notice itself. EIOPA's complaints-handling guidelines represent a separate institutional topic; the title and official publication establish this distinction, without importing a complaint deadline into claim notification. A case management system should link the complaint clock and original claim chronology. Closing the complaint does not automatically establish that the underlying claim has been paid, or vice versa.
Escalation is meaningful when triggered by age, material risk, vulnerability, failed communication, repeated requests, or a missing responsible owner. The escalation record includes the original issue, responsible authority, expected next step, and resolution evidence. High-risk overdue cases should be reviewed even when average figures meet targets. A governance body should assess tail outcomes, not reduce a claimant's experience to a monthly mean.
Regulatory Case Study and Evaluation Design
Australian home insurance review
The ASIC (2023) Navigating the Storm report provides a documented example of the measurement problem. It collected information from six participating insurers on 218,256 home insurance claims lodged between January and March 2022, with progress information through September. The claims included severe weather and other events. ASIC explicitly described the snapshot's restricted observation period, different data supply, and consequent undercounting of later actions. The denominator for some measures was smaller than the headline cohort. At the reporting snapshot, 76% were categorized as accepted or partly accepted, 8% declined, 12% withdrawn, and 4% not yet decided. These rounded proportions describe the reviewed cohort at that observation boundary, rather than ultimate outcomes or an international prevalence estimate.
The report connects communication and third-party project management with documented consumer experience and identifies difficulties in consistently recording decision dates and potential Code breaches. For governance, this suggests that apparently precise performance statistics need an audit of the operational definitions and their evidence. A complete-period estimate of the actual prevalence of breaches cannot be reconstructed merely from the report's point-in-time snapshot.
The distinction between documentary evidence and inference is important. The report provides observational and qualitative evidence of problems and examples of better handling. It does not, by itself, identify the causal effect of any single communication practice, quantify the causal influence of outsourcing, or establish that every insurer experiences the same difficulties. The contribution to this dossier is evidence for measurement scope and institutional mechanisms, not a causal parameter.
Research question and target population
An evaluation of a proposed early-response intervention should specify the intervention, assignment mechanism, comparator, eligible population, primary outcome, observation horizon, and adverse consequences. For example, evaluate an assisted reporting channel introduced for all newly reported property claims in selected operations. The primary endpoint could be time from the first authenticated receipt to a precisely defined first substantive action within seven calendar days. Secondary outcomes would include completion of initial contact, repeat contacts, complaints within a fixed follow-up, reopened claims, documentation quality, and time to emergency support for those identified as eligible.
Define the comparison at receipt, before potentially intervention-affected variables such as completed assessment or accepted status. Include all eligible reports, even if subsequently denied, withdrawn, or unresolved, and specify how multiple reports and duplicate claims are joined. A reporting-channel intervention can change who makes a report, so a comparison conditional on observed claims may also reflect population change. Where possible and lawful, measure exposure or initial contact attempts to distinguish accessibility effects from processing effects.
If random allocation is operationally and ethically feasible, randomize a supplemental service or encouragement to use the new channel while maintaining all legally required services. Stratification by product and catastrophe status can improve balance. If implementation is staggered by office, use a design appropriate to a cluster-level treatment and explicitly consider spillover and shared resources. Required protections and urgent assistance are maintained independently of study allocation.
When an intervention is not randomized, a before-and-after comparison has competing explanations: seasonal claim patterns, changes in product mix, inflationary or catastrophe conditions, recruitment and staffing, altered thresholds, and migration of record definitions. Difference-in-differences designs depend on credible comparison trends and lack of simultaneous differential changes, while interrupted time-series designs require adequate pre- and postintervention observations, modeling of seasonality and autocorrelation, and examination of co-occurring events. These methods are established research designs; this study does not claim to have estimated their effects.
A transparent illustrative comparison
Suppose two fully observed groups of 100 otherwise eligible claim episodes each are followed for seven days. In a hypothetical comparator group, 70 receive substantive action by day seven; in an intervention group, 85 do. The raw absolute difference is 15 percentage points, and the risk ratio is 85/70 = 1.214. Under independent binomial sampling, an approximate unpooled standard error of the difference is √[(0.85×0.15)/100 + (0.70×0.30)/100] = 0.0581. The corresponding illustrative normal 95% interval is approximately 3.6 to 26.4 percentage points. This arithmetic assumes independent observations and a fixed denominator; the sample is invented and supplies no real treatment-effect evidence.
Now consider a different hypothetical comparison constrained by case mix. In period A, 80 simple claims have a 90% on-time rate and 20 complex claims a 50% rate: 82 of 100 are on time. In period B, 20 simple claims attain 95% and 80 complex claims attain 60%: 67 of 100 are on time. Both strata improved, but the combined raw figure fell from 82% to 67%. If period B is standardized to period A's 80/20 mix, the result is 0.8×0.95 + 0.2×0.60 = 0.88, an improvement of six percentage points against the 82% baseline. The example demonstrates why reporting both observed overall outcomes and a justified standardized comparison is important. Standardization describes a weighted estimate; it does not establish causal attribution.
A statistically detectable improvement does not establish fair handling by itself. Measure error, appeal, denial reason, and claimant burden alongside faster action. A subgroup with systematically worse access may be hidden by the average. If the intervention is a prediction or automation system, examine calibration by relevant subgroup, drift, missingness, override patterns, and whether the automation's actions meet the predeclared substantive definition. Preserve the identity of the accountable decision maker and the ability to contest an adverse decision.
Integrated Analysis, Reproducibility, and Limitations
The relation between elapsed time and responsibility
The comparative materials and methodological illustrations support three linked findings. First, knowledge, receipt, registration, acknowledgment, substantive action, proof, decision, and payment are separate events with different functions. Second, elapsed duration is an observation whose legal consequence depends on the relevant rule and evidence; notice prejudice and date-certain coverage conditions cannot be treated as universal variants of one numerical deadline. Third, operational efficiency requires measurement of both timely advancement and legitimate claimant outcomes. A metric that rewards one event while erasing another can improve a dashboard without improving the overall claim experience.
These findings arise by combining established disciplinary tools. Their central relation is practical: an institution can measure an obligation only if its data represent the event that starts the obligation, the activity that satisfies it, and the genuine exceptions. If the start is recorded late, the reported compliance rate is biased favorably; if the terminal action is recorded prematurely, the same bias arises from the opposite boundary. The solution is evidence reconstruction, not adjustment by an arbitrary constant.
A full governance assessment therefore studies the claim journey as a sequence with multiple clocks and responsibilities. It keeps both the lived waiting span and each applicable legal span. It preserves evidence for competing interpretations of delay and avoids assigning blame solely through a field label. It examines the distribution and the tail rather than reporting only a mean. It relates the insurer's queue and capacity to the entrant population, and examines the portfolio implications of changed reporting behavior.
Reproducible analytic specification
A reproducible study begins with a frozen extraction date and a documented query of the eligible base population. Retain a dataset version, cohort dates, claim unit, join rules, deduplication strategy, selected jurisdiction and product, calendar version, and definitions of each event. Represent missing endpoints explicitly. Specify whether the event of interest is first meaningful contact, first substantive action, or communication of a decision. These endpoints are not interchangeable.
An implementable analytical sequence is to identify original relevant receipt, determine whether the target event occurred, validate event artifacts and order, calculate elapsed and rule-based durations, and classify the outcome or censoring. Then tabulate eligibility, known classification, uncertainty, and missing status. Use a receipt-cohort survival analysis for observed cases still waiting, competing-risk analysis for precluding outcomes where justified, and a separate actuarial model or mature occurrence sample for right-truncated reporting data. Publish outcome definitions, exclusions, and sample counts with the tables.
Versioning matters for comparative governance. Store legal rules with validity periods; audit the mapping between a claim and its rule; keep a history of changes to the metric; run sensitivity analysis using reasonable alternate clock definitions; and reproduce historical results using information known at the historical extraction date. Material corrections are transparent restatements. An institutional review can reconstruct the frequency of timely receipt, substantive action, communication, and any validated deadline breach from the independently documented artifacts.
A technical timestamp quality check can test permitted ordering and timezone conversions, but its pass does not establish source truth. A sampled files review must inspect whether the recorded event corresponds to the communication or action. Conversely, a good qualitative case review does not establish the population distribution without a representative sample. The two forms of evidence serve complementary roles.
Limits of evidence and international comparison
The legal study uses selected instruments from Europe, Asia, North America, Australia, and the international supervisory framework. It does not exhaust other jurisdictions or special products such as mandatory motor compensation, workers' compensation, public health benefits, or formal reinsurance notices. Certain sources were examined in dated versions, particularly the Japanese provision, the French provision, and the October 2023 Australian Code. Consequently the comparison establishes their documented institutional design within the stated versions, not a complete current compliance schedule. Legal operationalization requires product-specific and date-specific verification.
Primary regulatory documents were read in direct official text where available, with California and Colorado texts accessed in public reproductions. Some PDF images could not be rendered by the retrieval service even though the relevant extracted text was available. The original Kaplan–Meier article was checked through its publisher-indexed abstract, supported by the examined competing-risks article; the original proof was not reexamined. Little's original paper was used through its publisher abstract, and the reporting-delay study through the available publisher-version text of its abstract and introduction. EIOPA is cited only for the scope of a complaints-handling document. These are explicit access and inference boundaries.
All numerical exercises other than the quoted ASIC cohort facts are constructed and reproducible arithmetic, not observed insurance outcomes. There is no representative claimant survey, proprietary claims-file data, experimental estimate, independent replication, or external peer review. The manuscript establishes analytical distinctions and practical implications rather than new empirical generalizations. The research is written for an international scholarly encyclopedia and does not recommend a particular insurer or provide individualized claims advice.
Conclusion
Claim notification time is a family of measurements grounded in distinct events and obligations. Occurrence-to-discovery describes the passage from loss to knowledge; discovery-to-notice and notice-to-receipt describe the transfer of information; receipt-to-acknowledgment and receipt-to-first-action describe institutional responsiveness. Proof-to-decision and decision-to-payment remain related but separate transitions. The scientific value of each metric depends on transparent event identity and a well-defined eligible population.
Legal comparisons establish the importance of the role of the obligated party, knowledge, authorized receipt, the instrument's force, and the purpose of the relevant provision. Prejudice may matter to the legal consequence of delay within a specified doctrine, while a different condition defines a particular coverage boundary. A numerical late-notice classification is evidence for examination, not sufficient grounds for denial in every setting. Statutory, contractual, supervisory, industry-code, and internal service timeframes should remain identifiable in the same claim history.
Statistical analysis requires the appropriate time origin, treatment of current waiting, and explicit competing events. The portfolio reporting-delay distribution is additionally affected by the nonappearance of unreported losses. A claims system should preserve the difference between the reported cohort and the underlying occurrence process. The actuarial implications of faster reporting can be evaluated only with attention to both incidence and reporting behavior.
Governance is credible when a claimant's experience, the law's defined obligation, and the administrative evidence can all be reconstructed. Capacity, accessible intake, timely communication, and meaningful first action are complementary aspects of that task. Measurement should include the aged tail, uncertain records, exceptions, and adverse outcomes alongside headline service rates. This integrated conclusion follows from the selected legal, institutional, and statistical evidence and the illustrated calculations. It provides an analytical basis for research and verification rather than a new theoretical or doctrinal system.
References
Austin, P. C., Lee, D. S., & Fine, J. P. (2016). Introduction to the analysis of survival data in the presence of competing risks. Circulation, 133(6), 601–609. DOI: 10.1161/CIRCULATIONAHA.115.017719. Source text
Australian Securities and Investments Commission (ASIC). (2023, August 16). Navigating the Storm: ASIC’s Review of Home Insurance Claims (Report 768). Source text
Australian Securities and Investments Commission (ASIC). (2025, June 16). Claims Handling and Settling: How to Comply with Your AFS Licence Obligations (Information Sheet 253). Source text
California Department of Insurance. (n.d.). California Code of Regulations, title 10, § 2695.5, Duties upon Receipt of Communications. Legal Information Institute, Cornell Law School. Source text
California Department of Insurance. (n.d.). California Code of Regulations, title 10, § 2695.7, Standards for Prompt, Fair and Equitable Settlements. Legal Information Institute, Cornell Law School. Source text
European Insurance and Occupational Pensions Authority (EIOPA). (2012). Guidelines on Complaints-Handling by Insurance Undertakings. Official publication page; English file listed January 31, 2023. Source text
Financial Conduct Authority (FCA). (2008–). FCA Handbook, Insurance: Conduct of Business Sourcebook, ICOBS 8.1, Insurers: General, rule 8.1.1R (effective January 6, 2008). Source text
France. (2018). Code des assurances, art. L113-2 (version in force from April 1, 2018; text displayed for November 10, 2024). Légifrance. Source text
Germany. (2007). Versicherungsvertragsgesetz (VVG; Insurance Contract Act), § 28, Verletzung einer vertraglichen Obliegenheit. Gesetze im Internet. Source text
Germany. (2007). Versicherungsvertragsgesetz (VVG; Insurance Contract Act), § 30, Anzeige des Versicherungsfalles. Gesetze im Internet. Source text
Insurance Council of Australia. (2023). General Insurance Code of Practice (2020 Code, October 1, 2023 version), part 8, paragraphs 67–78. Code Governance Committee. Source text
International Association of Insurance Supervisors (IAIS). (2024). Insurance Core Principles and ComFrame, revisions adopted December 5, 2024. ICP 19, standards 19.10–19.10.4. ICP and ComFrame Online Tool. Source text
Japan. (2008). Insurance Act, Act No. 56 of 2008, art. 14, Notice of Occurrence of Loss (December 19, 2022 version). e-Gov Laws. Source text
Kaplan, E. L., & Meier, P. (1958). Nonparametric estimation from incomplete observations. Journal of the American Statistical Association, 53(282), 457–481. DOI: 10.1080/01621459.1958.10501452. Source text
Little, J. D. C. (1961). A proof for the queuing formula: L = λW. Operations Research, 9(3), 383–387. DOI: 10.1287/opre.9.3.383. Source text
National Association of Insurance Commissioners (NAIC). (1997). Unfair Claims Settlement Practices Act. Model 900, §§ 3–4. Source text
Supreme Court of Colorado. (2015). Craft v. Philadelphia Indemnity Insurance Co., 2015 CO 11, 343 P.3d 951, No. 14SA43. Source text
Texas Legislature. (n.d.). Texas Insurance Code, chapter 542, Processing and Settlement of Claims, subchapter B, §§ 542.055–542.059. Texas Constitution and Statutes. Source text
Verrall, R. J., & Wüthrich, M. V. (2016). Understanding reporting delay in general insurance. Risks, 4(3), article 25. DOI: 10.3390/risks4030025. Published July 8, 2016; publisher version archived by EconStor. Source text