Insurance by BogdanovaThe Encyclopedia of Insurance and Risk Science

RESEARCH DOSSIER

Reserve Selection and Booked Liability Reconciliation

Abstract. The selection of an insurer's recorded unpaid-claim liability is a consequential accounting and governance decision informed by, but conceptually distinct from, the actuarial estimation of future claim cash flows. It requires a specified measurement objective, a valuation date, a defined population of incurred claims, an assessment of prediction uncertainty, an explicit record of management adjustments, and an auditable connection between claim records, actuarial analyses, subsidiary ledgers, financial statements, and regulatory returns. This monograph develops a documentary and quantitative examination of that decision across property/casualty actuarial practice, United States statutory reporting, IFRS 17 financial reporting, and European Solvency II prudential valuation. The research distinguishes actuarial central estimates from ranges of reasonable estimates, statistical prediction intervals, booked loss and expense reserves, risk adjustments, risk margins, reinsurance recoverables, and recognized insurance contract liabilities. It supplies an internally reconciled hypothetical example involving three insurance portfolios, reserve selection adjustments, expense provisions, gross and net balances, and a calendar-year runoff test. A second illustration shows why an undiscounted recorded balance cannot be carried unchanged into a discounted, risk-adjusted measurement basis. The central finding is that reserve adequacy can be discussed coherently only after the target liability, measurement basis, entity boundary, reporting date, gross/net scope, and decision authority have been identified. A reconciliation is scientifically informative when every difference has an accountable economic, methodological, or booking explanation. The work offers an integrated analytical synthesis of established sources, without claiming a new reserving theory or a universally prescribed reserve-selection algorithm.

Keywords: Claims reserving; unpaid claims; loss adjustment expenses; booked reserves; actuarial central estimate; reserve range; management selection; IBNR; IBNER; financial reconciliation; Schedule P; statutory accounting; IFRS 17; liability for incurred claims; Solvency II; technical provisions; actuarial governance; reserve uncertainty; reinsurance; claims development.

Introduction

An insurer estimates the ultimate cost of insured events before every claim is reported, fully investigated, defended, settled, and paid. At a reporting date it also recognizes amounts that enter accounts, disclosures, solvency measures, and decisions about capital and distributions. These two activities interact, but their outputs need not share a number, a statistical target, or even a reporting perimeter. Estimation asks what is expected to remain payable under specified assumptions. Booking asks what amount the entity recognizes under a specified accounting framework, within an authorized closing process. Reconciliation explains the passage between them, and the economic and institutional meaning of every difference.

Historical descriptions of reserving often used the word reserve for both projected estimates and accounting balances. The Casualty Actuarial Society records that its historical Statements of Principles served as educational conceptual foundations and that two were rescinded in 2020 as professional standards developed. This history confirms the importance of distinguishing conceptual literature from the current standards governing an actual engagement.

The distinction is articulated unusually precisely in the Actuarial Standards Board's ASOP 43 on unpaid claims, which reserves the technical word reserve for an amount booked in a financial statement while treating an actuary's unpaid-claim estimate as an actuarial finding. The standard places actions taken by the actuary's principal concerning such estimates beyond its scope. This division of responsibility is the point of departure for the present work. A documented actuarial estimate can support a defensible booked liability; the booking decision requires further reconciliation and governance evidence.

The methods examined here are documentary comparative analysis, interpretation of official standards, synthesis of established actuarial literature, and constructed numerical examples whose arithmetic can be independently reproduced. The principal US instruments are ASOPs 43, 36, 20, 23, 41, and 56, together with the 2026 NAIC property/casualty annual statement instructions. International comparison uses IFRS 17, the European Solvency II Directive and EIOPA guidance, and the International Association of Insurance Supervisors' valuation principles. The scholarly literature includes Mack's analytic chain-ladder prediction error, England and Verrall's stochastic reserving survey, Wüthrich and Merz's reserving treatment, and paid-incurred methods. The scope focuses on nonlife incurred-claim liabilities. Life insurance contract liabilities, unexpired coverage measurement, and capital models are discussed where essential for accurate distinctions, rather than assimilated to property/casualty unpaid claims.

The researched question is operationally specific: how can an organization demonstrate that the value selected for recognition follows from a defensible estimation process and that the selected amount, financial records, and statements consistently describe the same obligations? The answer requires analysis of the estimand, the selection decision, and the reporting bridge. An apparently exact equality between actuarial and booked totals is weak evidence when the underlying populations differ. Conversely, a disclosed and quantitatively explained difference can be entirely legitimate.

Actuarial Estimates versus Recorded Reserves

The object measured and the institution recording it

An unpaid-claim estimate represents anticipated payments arising from events already incurred as of a defined accounting date. For a nonlife insurer it often encompasses expected claims payments and associated claim adjustment expenses, either in a single projection or in distinct components. An accounting liability is an amount recognized by the reporting entity under its applicable measurement framework. A reserving analysis may be performed for an insurer, a reinsurer, a self-insured institution, or another risk-bearing arrangement, while financial reporting follows the entity and contract boundary of the accounts. This scope difference explains why a projection can be accurate on its own terms yet unsuitable for direct placement into a particular balance-sheet line.

ASOP 43's scope and definitions distinguish gross estimates, net estimates, and estimates of recoverables, and require a documented intended measure. The actuary identifies whether the estimate is discounted, the exposure and claims included, the covered expense types, and how recoverability is treated. It also states that a label such as “best estimate” cannot substitute for the intended statistical or financial measure. This requirement is especially important in reserve selection, where an amount described as a “best estimate” may be a mean, a judgmental central estimate, a specific scenario, or an accounting amount assembled with further provisions.

For a contract portfolio indexed by segment i, an elementary undiscounted representation of the claims estimand is E[U_i | I_t], where U_i denotes future payment on events incurred by the accounting date and I_t denotes the information available by the valuation cutoff. A selected balance B_i may instead be expressed descriptively as B_i = A_i + M_i + J_i, where A_i is a defined actuarial reference, M_i is a documented management or selection adjustment, and J_i records adjustments required to translate between measurement boundaries or accounting classifications. These symbols are accounting explanations, not an algorithm for deriving a scientifically privileged amount. Their usefulness is conditional on a fully specified basis; otherwise different effects are liable to be double counted.

Dates, populations, and the multiple meanings of reserve

An accounting date separates incurred events and paid amounts from subsequent obligations. A data valuation date indicates the transaction cutoff used in the analysis. A review date indicates the information cutoff for actuarial judgment, including material subsequent information. A publication or filing date is a further event. Material changes between these dates can change the estimate even when the underlying historic accounting date is fixed. ASOP 43 explicitly addresses accounting, valuation, and review dates in actuarial communication, so a reconciliation must carry dates as data rather than silently infer them from file names or financial year labels.

A reporting basis may organize the same liabilities by accident year, underwriting year, report year, calendar year, policy period, product, legal entity, ceded layer, claims handling system, or statutory line of business. These are distinct partitions of a claims universe. For example, an accident-year actuarial triangle and a statutory statement's line-of-business classification may reconcile only through a documented mapping table. A change in the mapping can be as material as a change in development assumptions.

The term reserve also has different institutional uses. It can designate individual claim-handler case amounts, the insurer's aggregated booked provision, an actuarial unpaid-claim estimate, a Solvency II technical provision, an IFRS 17 liability for incurred claims, or a capital-related provision. A common vocabulary should preserve these distinctions rather than treat the terms as automatic synonyms. They can differ by discounted versus undiscounted valuation, inclusion of a risk adjustment, future expense scope, reinsurance accounting, and the grouping of coverage obligations. The valuation perspective must be stated before any comparison of magnitudes.

Claim case reserves, IBNR, IBNER, and claim expenses

A case reserve is a claim-specific amount recorded against a notified claim. It reflects the claim handler's or system's estimate under the organization's reserving instructions, including possible defense expense treatment. Case amounts may be informed by legal opinions, claims settlement practices, coverage decisions, limits, deductibles, recovery prospects, inflation assumptions, or automated rules. Their sum is an important ledger control, though not a sufficient forecast of all costs attributable to past events.

Incurred but not reported claims (IBNR) have occurred but are not yet notified; additional bulk development also reflects systematic inadequacy or overstatement in reported claim case estimates. Practitioners often use IBNR broadly to include this incurred-but-not-enough-reported (IBNER) development, and sometimes distinguish “pure IBNR” from IBNER for diagnostic purposes. The NAIC SSAP 55 text reproduced by CAS describes reported losses, IBNR, and adjustment expenses, expressly acknowledging that IBNR may include bulk provisions for deficient case reserves. The classification should therefore be governed by the actual accounting and claims system definition, not by a presumed universal taxonomy.

The booked unpaid claim liability may be displayed as case reserve plus a bulk unpaid-loss component plus unpaid loss adjustment expense. Defense and cost containment (DCC) and adjusting and other (AO) expenses are important US statutory classifications; other frameworks group acquisition, maintenance, claims handling, or directly attributable expenses differently. A reconciliation that omits expense definitions can show artificial adequacy or artificial deficiency. For example, a gross paid-plus-case triangle excluding DCC is not directly comparable to a booked claims-plus-DCC balance unless both datasets are aligned.

For each segment a simple closure identity is: expected ultimate claims = paid claims to date + expected future claims; therefore expected future claims = expected ultimate claims - paid claims to date. When the recorded case reserve is comparable in scope and currency, implied bulk reserve = selected expected future claims - recorded case reserve. Negative bulk in one segment may be economically plausible if case reserves are systematically conservative, but it requires scrutiny for scope mismatch, claim coding, and assumptions. It cannot be reclassified as zero without changing the selected total or creating an offsetting adjustment.

Reserve selection as an accountable decision

Reserve selection takes place at the intersection of technical estimation, business judgment, legal entity reporting, and internal control. Management commonly evaluates actuarial recommendations, multiple methods, emerging experience, profitability and capital implications, reporting instructions, and information unavailable in historical development triangles. Its duty is to establish a justified recorded amount under the relevant framework. This decision becomes scientifically auditable when each adjustment states its numerical amount, evidence, responsible decision maker, classification, timing, and effect on reported obligations.

The IAIS valuation principles emphasize reliable, decision-useful and transparent valuations; their articulation of ICP 14 makes clear why judgment can contribute to an informative range without excusing untraceable discretion. An opinion about reasonableness is evaluated against a stated standard and scope. A booked amount lying inside a range is not thereby proven to be a forecast mean, nor does being above a central estimate automatically demonstrate superior prudence. The meaning of the difference follows from the intended measure and the adopted financial-reporting basis.

Point Estimates and Ranges

Development-based estimates and exposure-based estimates

Claims development techniques exploit observed relationships between cohorts and their maturity. For a cumulative paid-loss triangle C(i,j), with origin period i and development age j, a chain-ladder age-to-age factor can be estimated as f_j = sum_i C(i,j+1) / sum_i C(i,j), over origin periods for which both cells are observed. A chain-ladder ultimate estimate for cohort i multiplies current cumulative paid losses by estimated subsequent development factors. The implied unpaid estimate is that projected ultimate less recorded paid losses. This familiar derivation uses development regularity as a substantive assumption; structural shifts in settlement rates, claim mix, inflation, policy terms or operational systems can undermine it.

Bornhuetter–Ferguson-type reasoning incorporates a prior expected ultimate loss, derived for example from exposure and an expected loss ratio, and an expected unreported or unpaid proportion at current maturity. If q_i is the estimated unpaid proportion and E_i the independent prior expected ultimate, a simplified expected unpaid amount is E_i q_i. In an ultimate-loss representation combining paid to date and expected unpaid, the formula becomes U_i = P_i + E_i q_i. It has lower immediate dependence on unusually high or low early paid claims than paid chain-ladder; at later development ages the exposure-based component diminishes. The prior must itself be justified and may be sensitive to pricing assumptions, inflation and underwriting selection.

A case-incurred triangle and a paid triangle encode different aspects of claims administration. Their joint analysis can reveal operational changes, case reserve strengthening, accelerations of payment, and changes in claims settlement. Merz and Wüthrich (2010) formulate a paid–incurred chain reserving model, explicitly combining payment and incurred information in a stochastic setting. Its scholarly contribution establishes that combining information requires a model of their relationship; averaging paid and incurred indications without checking dependence or development distortions can conceal disagreement.

A selection is often more defensible after segmentation than as a mechanical average of methods at portfolio level. Paid chain-ladder may be persuasive for a stable mature short-tail book, and exposure-based methods may be preferable for a rapidly growing, immature or newly restructured book. For large claims or complex litigation, claim-level qualitative information and a tail assessment may be pivotal. Method choice depends on the predictability of the phenomenon and the quality of data; method plurality is valuable when it provides independent diagnostic information rather than merely duplicating an assumption.

Estimands, parameter uncertainty, and predictive distributions

A point estimate answers a specific question only when the estimand is defined: conditional mean, median, actuarial central estimate, or selected judgmental outcome. The mean of future cash flows is appropriate for an expected-value measure, but a different accounting basis may require explicit risk compensation or market-consistent valuation. A range of estimates chosen as reasonable alternatives is not automatically a probability interval. The distinction protects both statistical inference and financial-accounting interpretation.

Mack (1993) derived a distribution-free chain-ladder standard error under specific chain-ladder model assumptions. The result makes explicit that even an apparently deterministic projected ultimate has prediction uncertainty. England and Verrall (2002) survey stochastic claims reserving approaches, examining prediction error, bootstrap and Bayesian approaches, and distributions of reserve outcomes. Wüthrich and Merz's Stochastic Claims Reserving Methods in Insurance develops related frameworks for predictive risk analysis. These sources support the measurement of uncertainty; they do not establish a unique number to book irrespective of reporting requirements.

Three uncertainties deserve separate treatment. Process risk concerns inherently random future development even with known model parameters. Parameter risk concerns incomplete information about model quantities. Model risk concerns structural choices, data transformations, development stability, and omitted drivers. Dependence risk arises when favorable or adverse outcomes across cohorts, lines or claim severities share economic and judicial causes. A single model's standard error does not cover every source of uncertainty. Multiple plausible models can therefore produce a broader and more decision-relevant account of liability than a highly precise standard error conditional on one unstable structure.

For total future loss R = sum_i R_i, the variance is Var(R) = sum_i Var(R_i) + 2 sum_{i<j} Cov(R_i,R_j). Ignoring positively correlated inflation or litigation shocks understates the variance of the aggregate. Modeling negative covariance merely to achieve diversification benefits requires empirical support. In a group of connected legal entities, the dependencies may increase when claim-management, underwriting, legal jurisdiction, and reinsurance recoverability are shared.

Interpreting a reserve range and management margins

ASOP 43 allows presentation as a point, a range, a point plus a margin, or a probability distribution and asks the actuary to explain a range's basis. The standard's discussion of ranges distinguishes a range of stand-alone reasonable estimates from a statistical confidence interval. A reasonable selection interval derived from judgment about credible methods does not carry an implicit coverage probability. Labeling every such interval a “90% confidence range” would be misleading unless a model and calibration support that interpretation.

Suppose an actuary provides an undiscounted gross estimate of 178 million monetary units with a documented range of 170–190 million. Management chooses a booked gross reserve of 188 million. The selected amount is 10 million above the actuarial central estimate, inside the documented range, and 2 million below its upper bound. A reserve opinion may regard the booked figure as within a range of reasonable estimates, given its own governing requirements. Nothing in these figures alone proves that 188 million is the mean or the 90th percentile of a predictive distribution. If management describes 10 million as a margin, its derivation and scope must be transparent, and its relation to legally required risk adjustments or margins separately evaluated.

Quantiles of a calibrated distribution can serve specified risk measurement purposes, including tail capital. If Q_p(R) denotes the p-quantile of future payments, the difference Q_p(R) - E[R] depends on p and on a defined predictive distribution. A hypothetical quantile must not be disguised as an observed market fact. Especially for severe long-tail liability claims, parameter and model uncertainty can dominate a seemingly precise frequency-severity simulation. The selection memorandum should identify whether a stated range comes from alternate actuarial methods, sensitivity bounds, probabilistic simulation, or a regulatory definition of reasonableness.

Sensitivities, dependence, and stress conditions

Sensitivity analysis evaluates the movement in reserve indications when assumptions change. For a portfolio with unpaid central losses of 72 million, an illustrative, proportional 10% severity uplift affecting the entire remaining claim amount adds 7.2 million. That calculation is an arithmetic scenario, not a prediction that inflation will reach 10%, and it cannot be added automatically to another trend provision already embedded in the estimate. Similar tests may vary payment patterns, large-loss frequency, settlement rates, reopening rates, bodily injury inflation, litigation duration, reinsurance credit risk, or exchange rates.

Stress scenarios serve a different purpose from ordinary predictive ranges: they expose concentrations and the conditions under which a selected booked amount would fail to absorb emerging obligations. A stress may be deliberately outside routine historical development and can be indispensable for governance, even where it is unsuitable for an unbiased current estimate. The board or reserving committee should receive both a defined base measurement and a candid description of material stress exposures.

The revised ASOP 20, property/casualty cash flow analysis, effective June 1, 2026 for covered work products, expands guidance on cash-flow analyses including discounting. Time value and timing assumptions affect the value of claims that may be paid over many years, and interest-rate sensitivity cannot be inferred from an undiscounted runoff triangle. Financial reporting may also apply prescribed discount rates or yield curves that differ from the economic assumptions used in an internal estimate. Selection must therefore keep base case, discounting, and explicit margins as identifiable components.

Selection and Governance

The reserve-selection process and decision rights

A reproducible reserving process separates actuarial preparation, review of estimates, management selection, financial posting, and assurance. Its sequence begins with a defined accounting date and claims population, continues with a frozen data extract and documented actuarial methods, and culminates in a reserve-selection record. The record identifies those authorized to propose changes, to challenge them, to approve their inclusion, and to post them to the general ledger. Accountable judgment is a feature of the process, and its exercise must leave a trace from underlying evidence to the final selected balance.

A reserve committee may receive a method-by-method diagnostic package: selected development factors, expected loss ratios, observed claim count and severity changes, large-loss exhibits, payment versus incurred comparisons, and explicit claim expense methods. It should also receive management information about claims operations, claims litigation, new coverage interpretations, portfolios acquired or transferred, and reinsurance disputes. Such information may legitimately change selections, provided its contribution is quantified or explained to a defensible materiality level. A discretionary override that has no identifiable mechanism, period, or portfolio attribution is difficult to distinguish from financial result management.

The distinction between decision authority and technical opinion matters especially when accounting reserves affect reported earnings or solvency. ASOP 36 on reserve opinions is directed to an actuary preparing a written opinion on carried reserves, while ASOP 43 governs the estimation work feeding that opinion. The actuary can assess a booked amount relative to relevant estimates and available evidence without being the person authorized to recognize it in the accounts. Recording an explicit allocation of responsibility improves the reliability of both the scientific estimate and the financial statement.

Selection policies should also address when committee members may consider post-valuation information. A loss reported after year-end may relate to a pre-year-end insured event, but its accounting treatment depends on the reporting framework and the nature of the new information. Decisions should be tied to the accounting date, the knowledge cutoff, and the actual ledger-close chronology. A file revised after a committee meeting should acquire a new version and a new approval trail; otherwise documentation can falsely imply that a decision maker reviewed information that became available only later.

Actuarial standards, accounting objectives, and independent opinion

The professional boundaries are explicit. ASOP 43 requires the actuary to consider the intended purpose, appropriate methods, materiality, and the reasonableness of the result. It recognizes that multiple methods, assumptions, and data limitations are part of professional practice. ASOP 36 addresses the written evaluation of reserve amounts, including the basis on which reserves are stated and the relationship of an actuary's analysis to work performed by others. Its current version is effective for opinions issued on or after October 1, 2024; using an older version without noticing the effective date could misstate the professional setting.

A reserve opinion and a financial audit have related but distinct objects. The actuary generally evaluates the reasonableness of reserves within a defined scope and stated standards. Accounting management owns the recognition and presentation of financial statement amounts. Financial auditors evaluate financial reporting evidence, controls and material misstatement within the relevant assurance framework. Regulators have a supervisory valuation objective. None of these roles should be silently substituted for another, and an unqualified opinion within its scope cannot prove that every estimate, risk margin, recoverable and capital item is certain.

The 2026 NAIC Property/Casualty Annual Statement Instructions make this interaction concrete. The Actuarial Report's comparisons require a connection between the appointed actuary's point estimates or reasonable ranges and the amounts carried, together with reconciliation of the data actually used in the analysis to Schedule P. The instructions distinguish a carried reserve inside an actuary's range from a deficiency, excess, qualification, or inability to give an opinion. These requirements are part of a specific US statutory filing regime; they should not be promoted into a universal reserve-selection algorithm for every jurisdiction or accounting basis.

A reserve committee can thus articulate its question more precisely: “Which selected amount satisfies the reporting basis, and what evidence justifies differences from the actuary's measured indication?” The answer may be a central estimate adjusted for accounting inclusions, a selection among reasonable methods, or a measurement containing explicitly required risk components. The governing framework determines how discretion enters the selection and what judgments must be disclosed.

Data provenance, controls, and basis mapping

Reserve estimates frequently traverse a chain of systems: policy administration, claim notification and adjudication, payment platforms, recoverables registers, actuarial triangles, spreadsheet selections, journal-entry interfaces, subsidiary ledgers, and reporting software. Every transformation can change population membership, currency, timing, aggregation, or sign. An audit trail identifies extract timestamps, source-system totals, mapping rules, excluded records, corrections, reruns, reconciliation owners and approvals.

ASOP 23 on data quality addresses the suitability, reasonableness, review and limitations of data, including potential changes to analytic conclusions when deficiencies exist. It does not turn a data check into proof of completeness. For a material reserve-selection exercise, good controls are targeted at economic assertions: occurrence of insured events, coverage period, gross/net identity, recorded payment amounts, the completeness of case reserves, claim status changes, loss adjustment expense allocation, recovery recognition, and the relation between current and prior extracts.

A mapping table should record how one source cohort contributes to one or more destination financial categories. For example, liability bodily injury may be analyzed in accident-year origin periods but reported across statutory automobile liability and other liability categories, depending on actual policy coding. Reinsurance treaty layers can connect multiple underlying lines and currencies. Without a crosswalk, a spreadsheet total matching a financial statement line might conceal compensating classification errors. A robust mapping is many-to-many where required, with explicit weights or identifiable case membership, and its row and column totals should reconcile under a consistent unit convention.

The 2026 NAIC instructions on Schedule P expressly insist that the reconciliation use the data actually employed by the appointed actuary rather than merely the data delivered to the actuary. This difference matters when the analytical process excludes outliers, corrects transaction dates, rebuckets claims, adjusts for large losses or adds external estimates. The trace must show the initial reported amount, the transformation, and the amount entering the selected indication. The objective is an explainable path, not cosmetic agreement between two exported totals.

Documentation, challenge, and subsequent revisions

A complete selection memorandum is both an accounting record and a record of technical reasoning. At minimum it should identify the accounting basis; exact valuation and review dates; legal entities and contract cohorts; currency conventions; gross/net inclusion; paid, case and bulk expense definitions; data version; analytic method indications; selected actuarial central amounts; range basis; management adjustments; approval signatures or equivalent records; final posted amounts; control totals and reconciliation; and all limitations materially affecting users. It should preserve prior versions rather than overwrite the reasoning chain.

ASOP 41 on actuarial communications requires communication appropriate to intended users and circumstances. Material methods and assumptions must be communicated with enough specificity for the user to understand the findings and their limitations. A board presentation may summarize evidence differently from a technical actuarial report, but a concise presentation should be traceable to a full technical basis. Excessive precision without decision relevance and undocumented rounding are both failures of usable communication.

ASOP 56 on modeling brings attention to model design, validation, reliance, and communication. A reserve projection generated from code or a commercial platform warrants review of input provenance, software version, adjustment logic, output reasonableness, model changes and intended use. An apparently automated decision can still embed discretionary choices in training populations, claims development factors, trend assumptions, data imputation and threshold parameters.

Changes in actuarial selections over time are evidence of emerging information, not automatically evidence of prior error. The reviewer needs the prior data and assumptions, subsequent paid and incurred development, prior booked liability, and contemporaneous documentation. A reserve release attributable to favorable development should be differentiated from an accounting change in discounting or from a reinsurance recoverable newly recognized. The essential scholarly requirement is causal attribution of changes rather than retrospective classification based solely on whether the ultimate proved larger or smaller.

IBNR and Case Reserve Reconciliation

An integrated hypothetical three-portfolio example

Consider an invented nonlife insurer with three segments—automobile, property and general liability—and an accounting date of December 31. Figures below are in millions of an unspecified reporting currency, with no inflation, discounting, or risk adjustment included in the initial undiscounted gross loss amounts. The data are designed for arithmetic demonstration; they are neither an observed insurer portfolio nor a fitted stochastic model. Cumulative paid amounts refer only to claim indemnity; outstanding case amounts refer to unpaid loss indemnity. Expenses are considered separately.

Portfolio Paid indemnity to date Recorded unpaid case indemnity Actuarial ultimate indemnity Actuarial unpaid indemnity Implied actuarial bulk
Automobile 120 35 181 61 26
Property 80 18 105 25 7
General liability 40 45 112 72 27
Total 240 98 398 158 60

The first identity is 398 = 240 + 158. The second is 158 = 98 + 60. These figures form a controlled analytical projection, but they do not yet state the amount that management elects to record. The original ultimate indications can result from different methods by line: mature automobile development, recent property loss information, and general-liability severity and legal-environment analysis. The use of multiple methods must be recorded before a management bridge is constructed; otherwise the selection step may conceal revisions that belong inside the actuarial model.

Suppose management selects indemnity ultimates of 184, 103, and 120 million, respectively, after reviewing claim-handling developments and the level of uncertainty. This produces selected unpaid indemnity of 64, 23, and 80 million. Their changes relative to the actuarial central indications are +3, -2, and +8 million, and the net change is +9 million. Each adjustment has an individually documentable origin. In an actual reserve selection, the +8 million long-tail liability adjustment might reflect identified litigated claims and an amended severity trend; the -2 million property adjustment might reflect substantiated faster settlement and lower cost indications. Here those rationales are examples of what evidence would be necessary, not claims that these events occurred.

Portfolio Actuarial unpaid Documented selection adjustment Selected unpaid Unpaid case Selected bulk
Automobile 61 +3 64 35 29
Property 25 -2 23 18 5
General liability 72 +8 80 45 35
Total 158 +9 167 98 69

The arithmetic is explicit: 167 = 158 + 9 and 167 = 98 + 69. The term “selected bulk” denotes the remainder of selected indemnity beyond booked case indemnity under this example's definitions. It includes both future reporting and development on previously reported claims, because this example does not provide sufficient claim-level detail to estimate pure IBNR separately from IBNER. Treating the whole 69 million as pure unreported claims would be an unsupported substantive interpretation.

Selected liability bridge and accounting control totals

Claim adjustment expenses are now added as separate unpaid amounts. Assume the actuarial central indication includes 13 million of future defense and cost containment expense (DCC) and 7 million of future adjusting and other expense (AO), producing 20 million in total. Management selects 14 million DCC and 7 million AO, an incremental one million of DCC. The final gross unpaid loss-and-expense liability is 167 + 14 + 7 = 188 million. The actuarial central on the same undiscounted gross loss-and-expense basis was 158 + 13 + 7 = 178 million. Therefore the 10 million movement decomposes into 9 million on indemnity and 1 million on DCC.

Reconciliation step Amount Running gross liability
Actuarial central unpaid indemnity 158 158
Actuarial central DCC +13 171
Actuarial central AO +7 178
Automobile selection difference +3 181
Property selection difference -2 179
General-liability selection difference +8 187
DCC expense selection difference +1 188
Final gross booked unpaid loss and expense 188

The running total is one accounting bridge across defined categories. It is distinct from the accounting components of the final reserve, which are unpaid case indemnity 98, selected bulk indemnity 69, DCC 14, and AO 7. Both paths converge on 188. A third confirmation can use a journal-entry tie-out: claim-level case balances 98 plus general-ledger bulk reserve 69 plus unpaid DCC and AO expense reserves of 21. The ledger summary is correct only if all components are counted exactly once and the case reserve extract and posting population share the same cutoff.

Suppose a preliminary financial-reporting export carries 186 million because two million of approved general-liability bulk reserve is present in the actuarial selection file but missing from the posted ledger. The reconciling difference is +2 million to the general ledger, after which the recorded amount agrees with 188. That is a posting correction, not evidence that the actuarial mean increased by two million. The memorandum must distinguish a mathematical selection change from a system-to-ledger correction; the posted 186 million should not be called the final booked amount after the adjustment.

A balancing entry cannot make a deficient estimate adequate by its mere existence. For every adjustment, the analyst asks whether it affects the actuarial cash flow, the accounting measurement, or the movement of a legitimately selected value into the correct financial statement account. This three-way classification avoids using unexplained plug entries to force reconciliation.

Gross-to-net and expense classifications

A second bridge addresses recoveries. Suppose, solely for illustration, that the insurer expects 24 million in ceded-reinsurance recoverables and 3 million in salvage and subrogation recoveries corresponding to the selected gross obligation, and that these amounts meet the assumed recognition and measurement conditions of the exercise. The simplified economic net exposure is 188 - 24 - 3 = 161 million. On the actuarial central basis, suppose eligible expected reinsurance recoveries are 22 million and salvage/subrogation recoveries 2 million, yielding 178 - 22 - 2 = 154 million. The selected net exposure exceeds the central net indication by 7 million, reflecting the gross +10 change together with +2 additional reinsurance and +1 additional salvage recovery.

Basis Gross unpaid losses and expenses Ceded recovery Salvage/subrogation Simplified net exposure
Actuarial central 178 22 2 154
Management selected 188 24 3 161
Difference +10 +2 +1 +7

This table is a deliberately simplified economic reconciliation, not a prescribed financial statement presentation. Reinsurance held can require a separately presented asset and its own measurements, and collectibility or contractual boundaries may change recognized recoveries. IFRS 17's treatment of reinsurance held distinguishes underlying insurance contracts issued from reinsurance contracts held; Solvency II also requires gross best estimates and separate recoverables. A company cannot necessarily net its underlying liabilities and its reinsurance rights within one liability line merely because an actuary calculates net expected cash flows.

The same caution applies to expenses. Payments for defense under a duty to defend may be classified as loss adjustment expense in one reporting framework, while payments under a policy insuring defense costs as the insured loss itself may enter indemnity classifications. The NAIC's SSAP 55 definitions illustrate why such classification is substantive. A financial reporting comparison must state whether it includes allocated and unallocated expense equivalents, defense expenses, coverage litigation, claims handling overhead, foreign-exchange adjustments, and future administration costs.

Runoff, backtesting, and diagnostic interpretation

Reserve selection is a dated decision under uncertainty. Its validation is informed by development after the accounting date, while the quality of the original decision must be judged against information reasonably available at that time. A simple runoff identity holds a prior origin-year cohort constant: prior booked unpaid liability + subsequent payments + change in remaining unpaid liability must reconcile across successive valuations subject to clear accounting adjustments. More directly, reestimated ultimate at the new date equals payments after the old date plus the newly estimated remaining liability, assuming the previous paid-to-date balance is properly excluded from the comparison.

For an invented prior cohort, suppose unpaid loss-and-expense liabilities booked at December 31 of Year 1 were 150 million. During Year 2, 65 million was paid on that same cohort. At the end of Year 2, the estimated unpaid amount for the cohort is 90 million. Its implied subsequent total development is 65 + 90 - 150 = +5 million, an adverse development of 5 million relative to the Year 1 booked amount. This calculation is not a statement that all Year 1 selections were unreasonable. It records what happened under new information and should be explained through payment patterns, claim emergence, estimation changes, and any relevant recognition adjustments.

A multi-period analysis can examine the direction, magnitude and timing of prior-year development by line, origin period, and selection method. Unusually persistent adverse development may indicate underestimated trends, an inadequate tail, inconsistent case reserve practice, or inaccurate assumptions about claim closure. Persistent favorable development may indicate genuine improvements, prudent selections, or over-conservative methods. The causal diagnosis should not be imposed from the sign of development alone. Observations are affected by mix changes and may become biased when blocks of claims transfer between entities or settlement systems.

Mack's prediction error framework shows why a sequence of future realizations should be interpreted within model uncertainty, and England and Verrall's stochastic reserving discussion supports looking beyond a single estimate to predictive outcomes. A reserve backtest ought to distinguish unanticipated experience from changes in economic conditions or reporting populations. Repeated backtesting can improve the calibration of estimates, but one favorable runoff year is not a general proof of model validity.

Financial Reporting Bridges

United States statutory accounting and NAIC returns

US state insurance statutory accounting measures and reports claim obligations within a prescribed regulatory framework that includes legal-entity filing, specified line allocations, statement exhibits, and actuarial opinion requirements. The NAIC 2026 Property/Casualty Instructions describe the annual statement, the actuary's opinion, the related report, and Schedule P reconciliation. These instructions establish a defined relationship between reported reserves, the actuarial analysis, and the filing's gross/net and expense classifications. They do not replace the applicable state's statutory law or approval requirements, and not every insurance entity is subject to an identical filing obligation.

Unpaid losses and claim adjustment expenses are accounted for as liabilities for incurred events under the statutory system. SSAP No. 55's reserve definitions, as reproduced in an accessible actuarial reference copy, distinguish reported losses, IBNR, and claim adjustment expenses. Since that PDF contains an older edition, it is used here for documented definitions, rather than as proof of every provision in the current NAIC Accounting Practices and Procedures Manual. The 2026 annual statement instructions are the controlling evidence in this research for the described annual filing and actuarial reconciliation requirements.

The annual statement requires careful alignment between reporting-category reserve balances, Schedule P development information, and the actuary's analysis. The 2026 instructions state that the comparison should address the actuary's conclusions—points, reasonable ranges, or both—and the carried amount, with a mapping of the actual data used in the actuarial analysis to Schedule P. The distinction matters for insurers whose management adjusts actuarial central estimates after the original analysis. The eventual statement may correctly carry a selected amount that differs from the central point, provided the purpose, reasonableness, and authority are addressed and the relevant recorded liabilities map to the return.

Consider the preceding example. A statement of the insurer's gross unpaid loss and expense liability would have to derive from the recognized 188 million, with indemnity 167 and expense 21 distinguished at the appropriate level. A separate supervisory exhibit could identify the actuary's 178 million central indication and the 170–190 million illustrative range, with a management selection of 188 million and the explicit adjustments totaling 10 million. If the accounting and filing classifications differ, the differences must be reconciled in a controlled crosswalk. The actuary cannot meaningfully compare a 188 million gross loss-and-expense carried reserve with a 154 million actuarial net-of-recoveries figure and call the 34 million spread a reserve margin. That subtraction mixes incompatible bases.

The revised ASOP 36 reserve-opinion standard adds professional guidance for the actuary evaluating the financial statement's carried reserve. A reported range may help assess whether the carried amount is a reasonable provision for specified liabilities, but an opinion is bounded by the standards and the statement's defined scope. Different accounting rules and materiality thresholds can lead to a different evaluation than an internal risk model. The scientific conclusion is a scoped relationship among an estimate, a booked number, and a regulatory assertion, rather than an invariant numerical reserve level.

IFRS 17 liability for incurred claims

IFRS 17 Insurance Contracts is a global financial-reporting framework applied where adopted, effective for annual reporting periods beginning on or after January 1, 2023. It recognizes and measures groups of insurance contracts through fulfilment cash flows and, for future service under the general measurement model, a contractual service margin. Within an issued insurance contract's carrying amount, the liability for remaining coverage relates to future insurance services and the liability for incurred claims (LIC) relates to incurred service obligations. An actuarial projection of unpaid claim cash flows is therefore an important ingredient of LIC measurement but not, by itself, necessarily the recognized LIC.

The IFRS Foundation's IFRS 17 key terms specify that fulfilment cash flows include projected cash flows, their timing, and risk; its explanation of LIC distinguishes incurred obligations from remaining coverage. The full text of an IFRS-aligned national implementation, AASB 17 Insurance Contracts, states in paragraph 40 that the carrying amount includes the liability for remaining coverage and the liability for incurred claims, the latter comprising fulfilment cash flows related to past service. The Australian text inspected is a dated compilation applicable to specified reporting periods, and national implementation should be checked at the relevant reporting date rather than presumed eternally identical in wording.

LIC cash-flow measurement may require a discounting adjustment and a risk adjustment for non-financial risk; the treatment of expenses, contract boundaries, discount rates and discount unwinding must follow the applicable standard. The contractual service margin generally relates to the liability for remaining coverage rather than an extra profit margin assigned to incurred claims. Under the premium allocation approach, AASB 17's paragraph 59 provides a relevant incurred-claims simplification: the entity need not adjust incurred-claim future cash flows for time value and financial risk when they are expected to be paid or received within one year of claim occurrence. This is a conditional exception; treating all short-tail portfolios as permanently exempt from discounting without checking the relevant claim-payment period would be overbroad.

To illustrate, suppose the selected 188 million gross claims-and-expense amount is a nominal forecast of 108 million paid one year after the measurement date and 80 million paid two years later. For a purely hypothetical flat annual discount rate of 4%, the present value is 108/1.04 + 80/(1.04)^2 = approximately 177.81 million. If an illustrative, separately determined non-financial risk adjustment is 8 million, the hypothetical discounted risk-adjusted incurred-claims value becomes 185.81 million before any further scope or presentation adjustments. These are arithmetic inputs selected for teaching, not prescribed IFRS 17 rates or an empirically estimated risk adjustment. The undiscounted 188 and discounted 185.81 are different measurement bases; their 2.19 million difference is not evidence that either estimate is inherently superior.

Under the reported framework, IFRS 17 also requires disclosures describing methods and assumptions and, in paragraph 119, the confidence level used for the non-financial risk adjustment or the corresponding level for another technique. The relevant text is reproduced in the EU's IFRS 17 endorsement regulation, with national implementation such as AASB 17 disclosure provisions. A reconciliation memorandum that identifies an 8 million risk adjustment without its method, confidence-level equivalence and specified perimeter would be incomplete for a reporting exercise to which these requirements apply.

An IFRS bridge should consequently distinguish nominal underlying cash flows, expected timing, discounting, risk adjustment, insurance versus reinsurance held, currency and grouping, and any additional presentation differences. The statutory reserve-selection table in the hypothetical example can supply a base cash-flow reconciliation, but it cannot be substituted wholesale for the IFRS group-level measurement. Accounting recognition requires the translation, not a change of label on the same number.

Solvency II best estimates and risk margins

Solvency II valuation addresses insurance and reinsurance obligations for prudential purposes. Article 76 of Directive 2009/138/EC requires technical provisions that correspond to obligations valued on a prudent, reliable and objective basis. Article 77 generally expresses technical provisions as a best estimate plus a risk margin. Its best estimate is the probability-weighted present value of future cash flows using the relevant risk-free interest-rate term structure; claims and expense inflows and outflows are considered over the lifetime of the obligations, and the best estimate is calculated gross of reinsurance recoverables, which are measured separately. The EIOPA Article 77 rulebook entry confirms this basic structure.

The best estimate is not necessarily identical to a US-style undiscounted actuarial central indication. The valuation includes expected cash flow timing, prescribed valuation principles, and regulatory technical provisions. The risk margin is a separate component linked to transferring obligations under the prudential framework. It cannot automatically be equated with a booked management margin, an IFRS 17 risk adjustment, or a percentile uplift. Those components arise from distinct institutional objectives and potentially different risk perimeters and calculation methods.

EIOPA's revised valuation guidelines, applicable from January 1, 2023, address matters including management actions, expert judgment, expenses, and the treatment of options and guarantees within the technical provision valuation framework. They reinforce a methodological lesson: expert judgment may be necessary to model complex future obligations, but it must operate within the rulebook's valuation basis. An internally selected undiscounted amount and a Solvency II technical provision need not coincide, and a divergence may be explained entirely by discounting, risk-margin construction and the treatment of recoverables.

Using the same illustrative two-year cash-flow schedule from the IFRS discussion solely for structural comparison, a stylized best estimate discounted at a hypothetical flat 4% would again be approximately 177.81 million. An invented risk margin of 6 million would produce illustrative technical provisions of 183.81 million. This is only an arithmetic demonstration of the addition of distinct components. The actual Solvency II interest-rate structure, risk margin method, contract obligations and calculation assumptions would determine the reported amount. The 6 million is not an actual calculated cost-of-capital margin, and the example makes no claim that a flat 4% rate or uniform risk margin is permitted in a real regulatory filing.

Under the IAIS ICP 14 valuation framework, consistency, reliability, objectivity, decision usefulness and transparency are critical to the comparison of insurance obligations. The existence of prudentially valid valuation differences does not eliminate the need to explain them. Instead, their separate identification is precisely what enables a regulator, auditor, board or researcher to understand why two recognized values are different.

Jurisdictional comparability and reporting basis differences

A meaningful international comparison begins with a matrix of measurement conventions: purpose, entity or contract boundary, definition of incurred event, valuation date, nominal versus present value, discount rate, non-financial risk adjustment, risk margin, expense scope, reinsurance presentation, currency conversion, and materiality. For a specified insurer, an analytic “gross undiscounted expected unpaid claims” column can serve as a common explanatory anchor. From that common base, each reporting basis follows its own permitted recognition and measurement path.

Comparison dimension Illustrative US P/C statutory framework IFRS 17 issued contracts Solvency II technical provisions
Principal object Filed unpaid losses and loss adjustment expenses Insurance-contract LIC for incurred service Prudential technical provisions for obligations
Base source Loss, case, bulk and expense records Fulfilment cash-flow projections by applicable contract group Probability-weighted future cash flows
Timing Filing-specific statutory requirements Discounting subject to IFRS 17 measurement and exceptions Discounted best estimate using relevant term structure
Uncertainty treatment Actuarial range and carried-reserve opinion under applicable rules Risk adjustment for non-financial risk Distinct risk margin with regulatory method
Recoverables Statutory gross/net and reinsurance presentation rules Reinsurance contracts held measured and presented separately Gross best estimate, recoverables calculated separately
Principal reconciliation Carried reserves, actuarial analysis, Schedule P and ledger LIC, related reinsurance-held balances, recognized financial statements Best estimate, risk margin, recoverables and prudential reporting

The matrix is intentionally functional rather than an exhaustive description of legal requirements. Accounting conventions change, and regional interpretations matter. The Canadian Institute of Actuaries standards page is one example of how applicable professional standards are maintained separately from US ASOPs; the published Canadian insurance standards and their effective dates must be checked for Canadian practice. Likewise, national financial statements under an IFRS adoption regime may incorporate reporting and regulatory overlays that do not appear in a US annual statement.

The strongest comparative claim supported by these official sources is that transparent reserve selection requires a common definitional base and a framework-specific bridge. It would exceed the evidence to say that one global percentage margin, one actuarial central estimator or one statutory opinion definition is binding across the international insurance market. A scientifically useful international reconciliation documents differences rather than erasing them.

Documentation and Independent Review

A reproducible reserve memorandum and evidence chain

A final reserve-selection memorandum can be organized so that a technically trained reviewer reconstructs each movement from source records without relying on the preparer's memory. The first section defines the exact obligor entity, populations and contract groups, dates, accounting basis and reserve scope. The second provides frozen source files or immutable extract identifiers, data dictionary, control totals and reconciliation to claims and financial systems. The third contains actuarial methods, selected parameters, reasonableness tests, assumptions and limitations. The fourth records all management selections and adjustments by portfolio, gross/net effect and affected reporting period. The fifth maps the final selection to the approved journal entries and the financial statements. The sixth records review challenges, resolutions, approval records and the version of the evidence used in the final close.

For the hypothetical example the reviewer needs evidence that the claim subledger totals 98 million case indemnity, the actuarial projections imply 60 million bulk in the central view, the selected indemnity bulk is 69 million, DCC and AO sum to 21 million, and the resulting gross booked liability is 188 million. The reviewer then confirms that reinsurance and salvage recovery assumptions are separately identified and that the simplified net economic result is 161 million. If a preliminary 186 million was incorrectly posted, the correction's journal ID and reason must be retained and the final ledger read back at 188 million. This reconstruction checks more than arithmetic; it checks that the same liability has not changed identity between stages.

An adequate working-paper record also addresses the treatment of newly reported large claims, prior-year development, catastrophe aggregates, portfolio transfers, claims reopened after closure, currency translation, recoverable disputes, and discount-rate updates when relevant. Evidence should be proportionate to materiality. The absence of detailed evidence for an immaterial item may be defensible under a documented materiality policy; the absence of evidence for a large unexplained selection override defeats the objective of auditable reconciliation.

The NAIC's reconciliation requirements supply a concrete statutory example of why the actuarial data-to-filing bridge must survive independent scrutiny. ASOP 23 and ASOP 41 support the data and communication disciplines that make such scrutiny possible. The standard of evidence is strongest when an internal control, a technical valuation and a final financial amount can each be traced to the same underlying obligation and properly stated transformations.

Independent review and decision-quality diagnostics

Independent review should begin with the scientific object and the reporting basis. A reviewer first assesses whether the reserve question has been posed correctly and whether the reported accounting categories match the actuarial analysis. Second, the reviewer examines data completeness and transformation controls. Third, methods are challenged for suitability, emerging experience and sensitivity. Fourth, the selection adjustments are tested for internal coherence, double counting and support. Fifth, the final figures are tied to ledger and statutory or IFRS reports. Sixth, the reviewer tests whether disclosures enable users to understand uncertainty and the actual authority responsible for the recorded amount.

Decision-quality indicators include the proportion of booked reserves supported by an actuarial model within its stated scope; the portion of selections attributable to explicit management overrides; the availability and coherence of independent method indications; the number and scale of unreconciled ledger differences; the timeliness of remediation; the proportion of claims data with stable source IDs; the stability of assumptions between valuation periods; and the share of reserve changes with documentary attribution. None is an independently validated universal threshold. They are audit and management diagnostics whose significance depends on portfolio size, maturity, organization, materiality, and reporting law.

A particularly important independence test concerns the same result appearing repeatedly under different analytical labels. Two projections that share input triangles, tail factors and inflation assumptions are not genuinely independent merely because they are generated by separate software packages. The reviewer should inspect dependence among methods and organizational incentives, especially where a management adjustment improves near-term earnings. Where expert judgment is necessary, its evidence, timing and sign should be documented. The IAIS valuation discussion explicitly connects objectivity with controls, independent information and review of management-influenced valuations.

Methodological boundaries and further research

This monograph is a comparative source-based analysis and mathematical illustration. Its numerical portfolios, reserve range, discount rates, expense amounts, recoveries and runoff figures were constructed for reproducibility. They establish accounting identities and illustrate decision differences; they are not empirical distributions, observed reserve outcomes, regulatory valuations, or evidence that any insurer uses the illustrated selections. A calibrated empirical study would require anonymized claims transactions, accident-year and report-year triangles, recorded case reserve changes, cash-flow timing, treaty terms, claim expense allocation, management selection records and final ledger postings. Those data would need legal permission, protection of confidential information, and enough time periods to evaluate uncertainty and model change.

The documentary inquiry is limited to selected major international and US, European and Canadian professional and accounting materials that could be inspected as of November 14, 2026. It does not assert complete coverage of every national accounting rule, supervisory filing instruction, insurance product, or recent amendment. Standards that are accessible as publisher summaries or indexed abstracts are used only for claims supported by that access level; inaccessible subscription text is not represented as reviewed in full. Statutory formulations should be checked against the exact version effective for a real reporting period. The research also does not replace insurer-specific actuarial analysis, accountancy, audit, financial advice or a regulator's legally operative determination.

Several research directions follow directly from observed methodological concerns. One is the empirical measurement of the relationship between reserve-selection overrides and subsequent development when management incentives, portfolio maturity and dependence are controlled. A second is the study of how IFRS 17 and Solvency II valuation bridges affect the interpretation of reserve stability and risk adjustments across time. A third is the design of interoperable reserve evidence records for claims systems, actuarial models and reporting ledgers, with verifiable transaction lineage. These are potential empirical questions and infrastructure tasks, not newly proposed theories or doctrines.

Integrated analytical findings

The reviewed standards and actuarial literature support a coherent explanatory synthesis. The first element is semantic precision: an actuarial estimate is specified by its estimand and scope; a booked liability is specified by its reporting basis and recognition authority. The second is quantitative discipline: actuarial method indications, case reserves, bulk provisions, expenses, recoverables, risk adjustments and discounts require separately identified components and common-basis comparisons. The third is governance: management judgment can select or alter a recorded reserve, while actuarial and assurance roles carry distinct responsibilities. The fourth is accountability over time: reconciliation must connect the actual analysis dataset to reports and must permit subsequent runoff development to be attributed to evidence rather than retrospectively rewritten assumptions.

These elements explain an apparent paradox in reserving practice. A selected reserve different from a central estimate may be defensible and accurately booked when the difference is justified and measured on a compatible basis. A balance exactly equal to the actuary's indication may still be wrong if the wrong legal entity, incurred-event period, expense scope, recoverable measure or discounting convention was used. Reconciliation quality is therefore assessed by traceable correspondence of obligations and decisions, in addition to numerical equality.

The operational consequence is that management and technical reviewers should retain a three-part record: the actuarial estimation evidence, the substantive reasons for the booked selection, and the reporting-basis and ledger mapping. The numerical example demonstrates this result without postulating any new formal reserve-selection model. It is a reasoned integration of documentary sources, accounting identities and existing reserving methods.

Conclusion

Reserve selection and booked liability reconciliation are essential to the credibility of insurer financial reporting and prudential assessment. A booked claims amount is the outcome of a specific recognition and governance process, and its relationship to actuarial indications must be explained on a common measurement basis. Properly scoped actuarial estimates make visible the amount of future payment expected from past insured events and the uncertainty associated with that amount. Financial records then recognize amounts under the applicable institutional rules, including relevant expenses, discounting, recoverable treatment, and risk compensation. The reconciliation documents the transition from evidence to recorded liability.

The hypothetical three-portfolio example establishes an auditable sequence: 158 million of central unpaid indemnity plus 20 million of expenses yields a 178 million gross actuarial indication; documented selection differences of 10 million yield 188 million of gross booked unpaid loss and expense. Case indemnity, bulk indemnity and expense components independently reconstruct the same balance. A separately disclosed economic recoverables illustration produces 161 million of simplified net exposure, while a separate runoff test records 5 million of adverse development on an earlier cohort. Each exercise demonstrates a different reconciliation identity, and each requires its own reporting scope.

US statutory reporting, IFRS 17 and Solvency II prescribe distinct measurement and presentation purposes. Their frameworks can yield different legitimate financial amounts from related claims data, and comparison requires explicit valuation bridges. Actuarial standards add the disciplines of defined purpose, data quality, model reasonableness, communication, and independent opinion. The scientific conclusion of this dossier is that reliable reserve governance combines the validity of estimates, the transparency of selection, the integrity of financial mapping, and the preservation of evidence across reporting periods. That conclusion derives from established professional sources and reconstructed quantitative examples. No claim of a new actuarial theory is made.

References