AIG 2009 Annual Report Download - page 100

American International Group, Inc., and Subsidiaries
the latest three accident years, and are given some weight for all years other than the latest accident year. For excess
coverages, expected loss methods are generally given all the weight for the latest three accident years, and are also
given considerable weight for accident years prior to the latest three years. For other classes of healthcare coverage,
an analogous weighting between loss development and expected loss ratio methods is utilized. The weights assigned to
each method are those which are believed to result in the best combination of responsiveness and stability. Frequency/
severity methods are sometimes utilized for pricing certain healthcare accounts or business. However, in testing loss
reserves the business is generally combined into larger groupings to enhance the credibility of the loss experience. The
frequency/severity methods that are applicable in pricing may not be appropriate for reserve testing and thus
frequency/severity methods are not generally employed in AIG’s healthcare reserve analyses.
Professional Liability: AIG generally uses a combination of loss development methods and expected loss ratio
methods for professional liability classes of business. Loss development methods are used for the more mature
accident years. Greater weight is given to expected loss ratio methods in the more recent accident years. Reserves are
tested separately for claims made classes and classes written on occurrence policy forms. Further segmentations are
made in a manner believed to provide an appropriate balance between credibility and homogeneity of the data.
Frequency/severity methods are used in pricing and profitability analyses for some classes of professional liability;
however, for loss reserve testing, the need to enhance credibility generally results in classes that are not sufficiently
homogenous to utilize frequency/severity methods.
Catastrophic Casualty: AIG utilizes expected loss ratio methods for all accident years for catastrophic casualty
business. This class of business consists of casualty or financial lines coverage which attaches in excess of very high
attachment points; thus the claims experience is marked by very low frequency and high severity. Because of the
limited number of claims, loss development methods are not utilized. The expected loss ratios and loss development
assumptions utilized are based upon the results of prior accident years for this business as well as for similar classes of
business written above lower attachment points. The business is generally written on a claims made basis. AIG utilizes
ground-up claim projections provided by AIG claims staff to assist in developing the appropriate reserve.
Aviation: AIG generally uses a combination of loss development methods and expected loss ratio methods for
aviation exposures. Aviation claims are not very long-tail in nature; however, they are driven by claim severity. Thus a
combination of both development and expected loss ratio methods are used for all but the latest accident year to
determine the loss reserves. Expected loss ratio methods are used to determine the loss reserves for the latest accident
year. Frequency/severity methods are not employed due to the high severity nature of the claims and different mix of
claims from year to year.
Personal Auto (Domestic): AIG generally utilizes frequency/severity methods and loss development methods for
domestic personal auto classes. For many classes of business, greater reliance is placed on frequency/severity methods
as claim counts emerge quickly for personal auto and allow for more immediate analysis of resulting loss trends and
comparisons to industry and other diagnostic metrics.
Fidelity/Surety: AIG generally uses loss development methods for fidelity exposures for all but the latest accident
year. Expected loss ratio methods are also given weight for the more recent accident years, and for the latest accident
year they may be given 100 percent weight. For surety exposures, AIG generally uses the same method as for short-tail
classes.
Mortgage Guaranty: AIG tests mortgage guaranty reserves using loss development methods, supplemented by an
internal claim analysis by actuaries and staff who specialize in the mortgage guaranty business. The claim analysis
projects ultimate losses for claims within each of several categories of delinquency based on actual historical
experience and is essentially a frequency/severity analysis for each category of delinquency. Additional reserve tests
using ‘‘Bornhuetter Ferguson’’ methods are also employed, as well as tests measuring losses as a percent of risk in
force. Reserves are reviewed separately for each class of business to consider the loss development characteristics
associated with the claims, the volume of claim data available for the applicable class and the applicability of various
actuarial methods to the class.
AIG 2009 Form 10-K 92