AIG 2008 Annual Report Download - page 96

nature of the automobile related claims. Claims relating to certain latent exposures such as construction defects or
exhaustion of underlying product aggregate limits are reviewed separately due to the unique emergence patterns of
losses relating to these claims. The expected loss ratios utilized for recent accident years are based on the projected
ultimate loss ratios of prior years, adjusted for rate changes, estimated loss cost trends and all other changes that can
be quantified. The estimated loss cost trend utilized in the year-end 2008 reviews averaged approximately five
percent for excess casualty classes. Frequency/severity methods are generally not utilized as the vast majority of
reported claims do not result in a claim payment. In addition, the average severity varies significantly from accident
year to accident year due to large losses which characterize this class of business, as well as changing proportions of
claims which do not result in a claim payment.
D&O: AIG generally utilizes a combination of loss development methods and expected loss ratio methods
for D&O and related management liability classes of business. Expected loss ratio methods are given more weight
in the two most recent accident years, whereas loss development methods are given more weight in more mature
accident years. In addition to these traditional actuarial methods, AIG’s actuaries utilize ground-up claim
projections provided by AIG claims staff as a benchmark for determining the indicated ultimate losses for all
accident years other than the most recent accident year. For the year-end 2008 loss reserve review, claims
projections for accident years 2007 and prior were utilized. These classes of business reflect claims made coverage,
and losses are characterized by low frequency and high severity. Thus, the claim projections can produce an overall
indicator of the ultimate loss exposure for these classes by identifying and estimating all large losses. Frequency/
severity methods are generally not utilized for these classes as the overall losses are driven by large losses more than
by claim frequency. Severity trends have varied significantly from accident year to accident year.
Workers’ Compensation: AIG generally utilizes loss development methods for all but the most recent
accident year. Expected loss ratio methods generally are given significant weight only in the most recent accident
year. Workers’ compensation claims are generally characterized by high frequency, low severity, and relatively
consistent loss development from one accident year to the next. AIG is a leading writer of workers’ compensation,
and thus has sufficient volume of claims experience to utilize development methods. AIG does not believe
frequency/severity methods are as appropriate, due to significant growth and changes in AIG’s workers’ com-
pensation business over the years. AIG generally segregates California business from other business in evaluating
workers’ compensation reserves. Certain classes of workers’ compensation, such as construction, are also evaluated
separately. Additionally, AIG writes a number of very large accounts which include workers’ compensation
coverage. These accounts are generally priced by AIG actuaries, and to the extent appropriate, the indicated losses
based on the pricing analysis may be utilized to record the initial estimated loss reserves for these accounts.
Excess Workers’ Compensation: AIG generally utilizes a combination of loss development methods and
expected loss ratio methods. Loss development methods are given the greater weight for mature accident years such
as 2002 and prior. Expected loss ratio methods are given the greater weight for the more recent accident years.
Excess workers’ compensation is an extremely long-tail class of business, with loss emergence extending for
decades. Therefore there is limited credibility in the reported losses for many of the more recent accident years. For
the mature accident years, AIG’s actuaries utilize claims projections provided by AIG claims staff to help determine
the loss development factors for this class of business.
General Liability: AIG generally uses a combination of loss development methods and expected loss ratio
methods for primary general liability or products liability classes. For certain classes of business with sufficient loss
volume, loss development methods may be given significant weight for all but the most recent one or two accident
years, whereas for smaller or more volatile classes of business, loss development methods may be given limited
weight for the five or more most recent accident years. Expected loss ratio methods would be utilized for the more
recent accident years for these classes. The loss experience for primary general liability business is generally
reviewed at a level that is believed to provide the most appropriate data for reserve analysis. For example, primary
claims made business is generally segregated from business written on an occurrence policy form. Additionally,
certain subclasses, such as construction, are generally reviewed separately from business in other subclasses. Due to
the fairly long-tail nature of general liability business, and the many subclasses that are reviewed individually, there
is less credibility in the reported losses and increased reliance on expected loss ratio methods. AIG’s actuaries
generally do not utilize frequency/severity methods to test reserves for this business, due to significant changes and
growth in AIG’s general liability and products liability business over the years.
90 AIG 2008 Form 10-K
American International Group, Inc., and Subsidiaries