AMERICAN INTERNATIONAL GROUP INC

The preparation of financial statements in accordance with GAAP requires the application of accounting policies that often involve a significant degree of judgment.

The accounting policies that we believe are most dependent on the application of estimates and assumptions, which are critical accounting estimates, are related to the determination of:

•loss reserves;•reinsurance assets, including the allowance for credit losses and disputes;•allowance for credit losses on certain investments, primarily on loans and available for sale fixed maturity securities;•fair value measurements of certain financial assets and financial liabilities;•income taxes, in particular the recoverability of our deferred tax asset and establishment of provisions for uncertain tax positions; and•goodwill impairment.

These accounting estimates require the use of assumptions about matters, some of which are highly uncertain at the time of estimation. To the extent actual experience differs from the assumptions used, our consolidated financial condition, results of operations and cash flows could be materially affected.

LOSS RESERVES

Loss reserves represent the accumulation of estimates of unpaid claims, including estimates for claims incurred but not reported and loss adjustment expenses, less applicable discount. We regularly review and update the methods used to determine loss reserve estimates. Because these estimates are subject to the outcome of future events and because loss trends vary and time is often required for changes in trends to be recognized and confirmed, changes in estimates are common.

The estimate of loss reserves relies on several key judgments:

•the determination of the actuarial methods used as the basis for these estimates;

•the relative weights given to these models by product line;

•the underlying assumptions used in these models; and

•the determination of the appropriate groupings of similar product lines and, in some cases, the disaggregation of dissimilar losses within a product line.

Numerous assumptions are made in determining the best estimate of reserves for each line of business, in consideration of expected ultimate losses, loss cost trends and loss development factors, where appropriate. The importance of any one assumption can vary by both line of business and accident year. Because such assumptions may differ from actual experience, there is potential for significant variation in the development of loss reserves. This estimation uncertainty is particularly relevant for long-tail lines of business.

All of our methods to calculate net reserves include assumptions about estimated reinsurance recoveries and their collectability. Reinsurance collectability is evaluated independently of the reserving process and appropriate allowances for uncollectible reinsurance are established.

Overview of Loss Reserving Process and Methods

Our loss reserves can generally be categorized into two distinct groups: short-tail reserves and long-tail reserves. Short-tail reserves consist principally of U.S. Property and Special Risks, UK/Europe Property and Special Risks, U.S. Personal Insurance, and UK/Europe and Japan Personal Insurance. Long-tail reserves include U.S. Workers’ Compensation, U.S. Excess Casualty, U.S. Other Casualty, U.S. Financial Lines, and UK/Europe Casualty and Financial Lines.

Short-Tail Reserves

In short-tail lines of business, such as property or personal insurance, where the nature of these claims tends to be higher frequency with short reporting periods, with volatility arising from occasional severe events, the actual losses reported make up a greater proportion of the ultimate loss estimate. During the first few development quarters of an accident year, the expected ultimate losses generally reflect the average loss costs from a period of preceding accident quarters that have been adjusted for changes in rate and loss cost trends, mix of business, known exposure to unreported losses, or other factors affecting the particular line of business. For more mature quarters, specific loss development methods and/or frequency/severity methods may be used to determine the incurred but not reported (IBNR). IBNR for claims arising from catastrophic events or events of unusual severity would be determined taking into account information known by the claims department, using alternative techniques or expected percentages of ultimate loss emergence based on historical emergence of similar events or claim types.

Long-Tail Reserves

Estimation of loss reserves for our long-tail business is a complex process and depends on a number of factors, including the product line and volume of business, as well as estimates of reinsurance recoveries. Experience in more recent accident years generally provides limited statistical credibility of reported net losses on long-tail business. That is because in the more recent accident years, a relatively low proportion of estimated ultimate net incurred losses are reported or paid. Therefore, IBNR reserves constitute a relatively high proportion of loss reserves.

For our long-tail lines, we generally make actuarial and other assumptions with respect to the following:

•Loss cost trend factors, which are used to establish expected loss ratios for subsequent accident years based on the projected loss ratios for prior accident years.

•Expected loss ratios, which are used for the latest accident year and, in some cases, for accident years prior to the latest accident year. The expected loss ratio also generally reflects the average loss ratio from prior accident years, adjusted for the loss cost trend and the effect of rate changes and other quantifiable factors on the loss ratio.

•Loss development factors, which are used to project the reported losses for each accident year to an ultimate basis. Generally, the actual loss development factors observed from prior accident years would be used as a basis to determine the loss development factors for the subsequent accident years.

•Tail factors, which are development factors used for certain long-tail lines of business to project future loss development for periods that extend beyond the available development data. The development of losses to the ultimate loss for a given accident year for these lines may take decades and the projection of ultimate losses for an accident year is very sensitive to the tail factors selected beyond a certain age.

We record quarterly changes in loss reserves for each product line of business. The overall change in our loss reserves is based on the sum of the changes for all product lines of business. The quarterly loss reserve changes are based on the estimated current loss ratio for each subset of coverage less any amounts paid. Also, any change in estimated ultimate losses from prior accident years deemed to be necessary based on the results of our latest detailed valuation reviews, large loss analyses, or other analytical techniques, either positive or negative, is reflected in the loss reserve and incurred losses for the current quarter. Differences between actual loss emergence in a given period and our expectations based on prior loss reserve estimates are used to monitor reserve adequacy between detailed valuation reviews and may also influence our judgment with respect to adjusting reserve estimates.

Details of the Loss Reserving Process

The process of determining the current loss ratio for each product line of business is based on a variety of factors. These include considerations such as: prior accident year and policy year loss ratios; rate changes; and changes in coverage, reinsurance, or mix of business. Other considerations include actual and anticipated changes in external factors such as trends in loss costs, inflation, employment rates or unemployment duration or in the legal and claims environment. The current loss ratio for each product line of business is intended to represent our best estimate after reflecting all relevant factors. At the close of each quarter, the assumptions and data underlying the loss ratios are reviewed to determine whether they remain appropriate. This process includes a review of the actual loss experience in the quarter, actual rate changes achieved, actual changes in reinsurance, quantifiable changes in coverage or mix of business, and changes in other factors that may affect the loss ratio. The loss ratio is changed to reflect the revised estimate if this review suggests that the previously determined loss ratio is no longer appropriate and, generally, shorter tailed lines of business are more likely to experience changes than longer tailed lines for immature accident years unless the information is directionally unfavorable.

We conduct a comprehensive loss reserve detailed valuation review at least annually for each product line of business in accordance with Actuarial Standards of Practice. These standards provide that the unpaid loss estimate may be presented in a variety of ways, such as a point estimate, a range of estimates, a point estimate based on the expected value of several reasonable estimates, or a probability distribution of the unpaid loss amount. Our actuarial best estimate for each product line of business represents an expected value generally considering a range of reasonably possible outcomes.

The reserve analysis, globally, for each product line of business is performed by a credentialed actuarial team in collaboration with claims, underwriting, business unit management, risk management and senior management. Our actuaries consider the ongoing applicability of prior data groupings and update numerous assumptions, including the analysis and selection of loss development and loss trend factors. They also determine and select the appropriate actuarial or other methods used to develop our best estimate for each business product line, and may employ multiple methods and assumptions for each product line. These data groupings, accident year weights, method selections and assumptions necessarily change over time as business mix changes, development factors mature and become more credible and loss characteristics evolve. We consult with third-party specialists to help inform our judgments as needed. Through the execution of these detailed valuation reviews an actuarial best estimate of the loss reserve is determined. The sum of these estimates for each product line of business yields an overall actuarial best estimate for that line of business.

A critical component of our detailed valuation reviews is an internal peer review of our reserving analyses and conclusions, where actuaries independent of the initial review evaluate the reasonableness of assumptions used, methods selected, and weightings given to different methods. In addition, each detailed valuation review is subjected to a review and challenge process by specialists in our Enterprise Risk Management (ERM) group.

For certain product lines, we measure sensitivities and determine explicit ranges around the actuarial best estimate using multiple methodologies and varying assumptions. Where we have ranges, we use them to inform our selection of best estimates of loss reserves by product line of business. Our range of reasonable estimates is not intended to cover all possibilities or extreme values and is based on known data and facts at the time of estimation.

Actuarial and Other Methods for Our Lines of Business

Our actuaries determine the appropriate actuarial methods and segmentation. This determination is based on a variety of factors including the nature of the losses associated with the product line of business, such as the frequency or severity of the claims. In addition to determining the actuarial methods, the actuaries determine the appropriate loss reserve groupings of data. This determination is a judgmental, dynamic process and refinements to the groupings are made every year. The groupings may change to reflect observed or emerging patterns within and across product lines, or to differentiate risk characteristics (for example, size of deductibles and extent of third-party claims specialists used by our insureds). As an example of reserve segmentation, we write many unique subsets of professional liability insurance, which cover different products, industry segments, and coverage structures. While for pricing or other purposes, it may be appropriate to evaluate the profitability of each subset individually, we believe it is appropriate to combine the subsets into larger groups for reserving purposes to produce a greater degree of credibility in the loss experience. This determination of data segmentation and related actuarial methods is assessed, reviewed and updated at least annually.

The actuarial methods we use most commonly include paid and incurred loss development methods, expected loss ratio methods, including “Bornhuetter Ferguson” and “Cape Cod,” and frequency/severity models. Loss development methods utilize the actual loss development patterns from prior accident years updated through the current year to project the reported losses to an ultimate basis for all accident years. We also use this information to update our current accident year loss selections. Loss development methods are generally most appropriate for lines of business that exhibit a stable pattern of loss development from one accident year to the next, and for which the components of the product line have similar development characteristics. Expected loss ratio methods rely on the application of an expected loss ratio to the earned premium for the product line of business to determine the liability for loss reserves and loss adjustment expenses. We generally use expected loss ratio methods in cases where the reported loss data lacked sufficient credibility to utilize loss development methods, such as for new product lines of business or for long-tail product lines at early stages of loss development. Frequency/severity models may be used where sufficient frequency counts are available to apply such approaches.

A key advantage of loss development methods is that they respond more quickly to any actual changes in loss costs for the product line of business. Therefore, if loss experience is unexpectedly deteriorating or improving, the loss development method gives full credibility to the changing experience. Expected loss ratio methods would be slower to respond to the change, as they would continue to give more weight to a prior expected loss ratio, until enough evidence emerged to modify the expected loss ratio to reflect the changing loss experience. On the other hand, loss development methods have the disadvantage of overreacting to changes in reported losses if the loss experience is anomalous due to the various key factors described above and the inherent volatility in some of the lines. For example, the presence or absence of large losses at the early stages of loss development could cause the loss development method to overreact to the favorable or unfavorable experience by assuming it is a fundamental shift in the development pattern. In these instances, expected loss ratio methods such as Bornhuetter Ferguson have the advantage of recognizing large losses without extrapolating unusual large loss activity onto the unreported portion of the losses for the accident year.

The Cape Cod method is a hybrid between the loss development and Bornhuetter Ferguson methods, where the historic loss data and loss development factor assumptions are used to determine the expected loss ratio estimate in the Bornhuetter Ferguson method.

Where appropriate, supplemental analysis for the given line of business may be performed in addition to the above described techniques such as Shareholder Class Action suit analysis for Directors and Officers (D&O) coverages.

Frequency/severity methods generally rely on the determination of an ultimate number of claims and an average severity for each claim for each accident year. Multiplying the estimated ultimate number of claims for each accident year by the expected average severity of each claim produces the estimated ultimate loss for the accident year. Frequency/severity methods generally require a sufficient volume of claims in order for the average severity to be predictable. Average severity for subsequent accident years is generally determined by applying an estimated annual loss cost trend to the estimated average claim severity from prior accident years. In certain cases, a structural approach may also be used to predict the ultimate loss cost. Frequency/severity methods have the advantage that ultimate claim counts can generally be estimated more quickly and accurately than can ultimate losses. Thus, if the average claim severity can be accurately estimated, these methods can more quickly respond to changes in loss experience than other methods. However, for average severity to be predictable, the product line of business must consist of homogenous types of claims for which loss severity trends from one year to the next are reasonably consistent and where there are limited changes to deductible levels or limits. Generally these methods work best for high frequency, low severity product lines of business such as personal auto. However, frequency and severity metrics are also used to test the reasonability of results for other product lines of business and provide indications of underlying trends in the data. In addition, ultimate claim counts can be used as an alternative exposure measure to earned premiums in the Cape Cod method.

The estimation of liability for loss reserves and loss adjustment expenses relating to asbestos and environmental pollution losses on insurance policies written many years ago is typically subject to greater uncertainty than other types of losses. This is due to inconsistent court decisions, as well as judicial interpretations and legislative actions that in some cases have tended to broaden coverage beyond the original intent of such policies or have expanded theories of liability. In addition, reinsurance recoverable balances relating to asbestos and environmental loss reserves are subject to greater uncertainty due to the underlying age of the claim, underlying legal issues surrounding the nature of the coverage, and determination of proper policy period. For these reasons, these balances tend to be subject to increased levels of disputes and legal collection activity when actually billed. The insurance industry as a whole is engaged in extensive litigation over these coverage and liability issues and is thus confronted with a continuing uncertainty in its efforts to quantify these exposures.

We continue to receive claims asserting injuries and damages from toxic waste, hazardous substances, and other environmental pollutants and alleged claims to cover the cleanup costs of hazardous waste dump sites, referred to collectively as environmental claims, and indemnity claims asserting injuries from asbestos. The vast majority of these asbestos and environmental losses emanate from policies written in 1984 and prior years. Commencing in 1985, standard policies contained absolute exclusions for pollution-related damage and asbestos. The current environmental policies that we specifically price and underwrite for environmental risks on a claims-made basis have been excluded from the analysis. Nevertheless, most of these legacy exposures have been heavily reinsured with very highly rated reinsurers.

The majority of our remaining exposures for asbestos and environmental losses are related to excess casualty coverages, not primary coverages. The litigation costs are treated in the same manner as indemnity amounts, with litigation expenses included within the limits of the liability we incur. Individual significant loss reserves, where future litigation costs are reasonably determinable, are established on a case-by-case basis.

Key Assumptions of our Actuarial Methods by Line of Business

The following sensitivity analysis table summarizes the effect on the loss reserve position of using certain alternative loss cost trend (for accident years where we use expected loss ratio methods) or loss development factor assumptions rather than the assumptions actually used in determining our estimates in the year-end loss reserve analyses in 2024:

(a)Tail factor increase of 2.5 percentage points for guaranteed cost business and 3 percentage points for deductible business.

(b)Tail factor decrease of 1 percentage point for guaranteed cost business and 1.5 percentage points for deductible business.

For additional information on our reserving process and methodology, see Note 13 to the Consolidated Financial Statements.

REINSURANCE ASSETS

In the ordinary course of business, our insurance companies may use both treaty and facultative reinsurance to minimize their net loss exposure to any single catastrophic loss event or to an accumulation of losses from a number of smaller events or to provide greater diversification of our businesses. Reinsurance assets include the balances due from reinsurance and insurance companies under the terms of our reinsurance agreements for paid and unpaid losses and loss adjustment expenses incurred, ceded unearned premiums and ceded future policy benefits for life and accident and health insurance contracts and benefits paid and unpaid. The estimation of reinsurance recoverables involves a significant amount of judgment. Reinsurance assets include reinsurance recoverables on unpaid losses and loss adjustment expenses that are estimated as part of our loss reserving process and, consequently, are subject to similar judgments and uncertainties as the estimation of gross loss reserves. For additional information on reinsurance, see Note 8 to the Consolidated Financial Statements.

ALLOWANCE FOR CREDIT LOSSES ON CERTAIN INVESTMENTS

We maintain an allowance for the expected lifetime credit losses of commercial and residential mortgage loans and available for sale securities. The sufficiency of this allowance is reviewed quarterly using both quantitative and qualitative considerations, which are subject to risks and uncertainties. These considerations and the overall methodology used to estimate the allowance for credit losses are discussed in more detail in Note 6 and Note 7 to the Consolidated Financial Statements for available for sale securities and Commercial and residential loans, respectively.

FAIR VALUE MEASUREMENTS OF CERTAIN FINANCIAL ASSETS AND FINANCIAL LIABILITIES

Assets and liabilities recorded at fair value in the Consolidated Balance Sheets are measured and classified in a hierarchy for disclosure purposes consisting of three levels based on the observability of inputs available in the marketplace used to measure the fair value. We classify fair value measurements for certain assets and liabilities as Level 3 when they require significant unobservable inputs in their valuation. We consider unobservable inputs to be those for which market data is not available. Our assessment of the significance of a particular input to the fair value measurement of an asset or liability requires judgment.

For additional information about the valuation methodologies of financial instruments measured at fair value, see Note 5 to the Consolidated Financial Statements.

INCOME TAXES

Deferred income taxes represent the tax effect of the differences between the amounts recorded in our Consolidated Financial Statements and the tax basis of assets and liabilities. Our assessment of net deferred income taxes represents management’s best estimate of the tax consequences of various events and transactions, which can themselves be based on other accounting estimates, resulting in incremental uncertainty in the estimation process.

Deferred Tax Asset Recoverability

The evaluation of the recoverability of our deferred tax asset and the need for a valuation allowance requires us to weigh all positive and negative evidence to reach a conclusion that it is more likely than not that all or some portion of the deferred tax asset will not be realized. The weight given to the evidence is commensurate with the extent to which it can be objectively verified. As such, changes in tax laws in countries where we transact business can impact our deferred tax asset valuation allowance. We consider multiple factors to reliably estimate future taxable income so we can determine the extent of our ability to realize net operating losses, foreign tax credits, realized capital loss and other carryforwards. These factors include forecasts of future income for each of our businesses, which incorporate forecasts of future statutory income for our insurance companies, and actual and planned business and operational changes, both of which include assumptions about future macroeconomic and AIG-specific conditions and events. We subject the forecasts to stresses of key assumptions and evaluate the effect on tax attribute utilization. We also apply stresses to our assumptions about the effectiveness of relevant prudent and feasible tax planning strategies. In performing our assessment of recoverability, we consider tax laws governing the utilization of net operating loss, capital loss and foreign tax credit carryforwards in each applicable jurisdiction. These tax laws are subject to change, resulting in incremental uncertainty in our assessment of recoverability.

Uncertain Tax Positions

Uncertain tax positions represent AIG’s liability for income taxes on tax years subject to review by the Internal Revenue Service (IRS) or other tax authorities. We determine whether it is more likely than not that a tax position will be sustained, based on technical merits, upon examination by the relevant taxing authorities before any part of the benefit can be recognized in the financial statements. A tax position is measured at the largest amount of benefit that is greater than 50 percent likely to be realized upon settlement. The completion of review, or the expiration of federal statute of limitations for a given audit period could result in an adjustment to the liability for income taxes.

For a discussion of our framework for assessing the recoverability of our deferred tax asset and other tax topics, see Note 21 to the Consolidated Financial Statements.

GOODWILL IMPAIRMENT

Goodwill represents the future economic benefits arising from assets acquired in a business combination that are not individually identified and separately recognized. Goodwill is tested for impairment annually, or more frequently if circumstances indicate an impairment may have occurred. A qualitative assessment may be performed, considering whether events or circumstances exist that lead to a determination that it is not more likely than not that the fair value of a segment is less than its carrying value. If management elects to perform a quantitative assessment to determine recoverability of carrying value or is compelled to do so based on the results of a qualitative assessment, the estimate of fair value involves applying one or a combination of common valuation approaches. These include discounted expected future cash flows, market-based earnings multiples and external appraisals, among other methods, all of which require management judgment and are subject to uncertainty, primarily as it relates to assumptions around business growth, earnings projections, and cost of capital.

For additional information on goodwill impairment, see Part I, Item 1A. Risk Factors – Estimates and Assumptions and Note 12 to the Consolidated Financial Statements.