AIG 2006 Annual Report Download - page 94

American International Group, Inc. and Subsidiaries
Management’s Discussion and Analysis of
Financial Condition and Results of Operations Continued
class of business must consist of homogeneous types of claims more than by claim frequency. Severity trends have varied
for which loss severity trends from one year to the next are significantly from accident year to accident year.
reasonably consistent. Generally these methods work best for Workers Compensation: AIG generally utilizes loss development
high frequency, low severity classes of business such as personal methods for all but the most recent accident year. Expected loss
auto. AIG utilizes these methods in pricing subclasses of ratio methods generally are given significant weight only in the
professional liability. However, AIG does not generally utilize most recent accident year. Workers compensation claims are
frequency/severity methods to test loss reserves, due to the generally characterized by high frequency, low severity, and
general nature of AIG’s reserves being applicable to lower relatively consistent loss development from one accident year to
frequency, higher severity commercial classes of business where the next. AIG is a leading writer of workers compensation, and
average claim severity is volatile. thus has sufficient volume of claims experience to utilize
Excess Casualty: AIG generally uses a combination of loss development methods. AIG does not believe frequency/severity
development methods and expected loss ratio methods for excess methods are as appropriate, due to significant growth and
casualty classes. Expected loss ratio methods are generally changes in AIG’s workers compensation business over the years.
utilized for at least the three latest accident years, due to the AIG generally segregates California business from other business
relatively low credibility of the reported losses. The loss experi- in evaluating workers compensation reserves. Certain classes of
ence is generally reviewed separately for lead umbrella classes workers compensation, such as construction, are also evaluated
and for other excess classes, due to the relatively shorter tail for separately. Additionally, AIG writes a number of very large
lead umbrella business. Automobile-related claims are generally accounts which include workers compensation coverage. These
reviewed separately from non-auto claims, due to the shorter tail accounts are generally priced by AIG actuaries, and to the extent
nature of the automobile related claims. The expected loss ratios appropriate, the indicated losses based on the pricing analysis
utilized for recent accident years are based on the projected may be utilized to record the initial estimated loss reserves for
ultimate loss ratios of prior years, adjusted for rate changes, these accounts.
estimated loss cost trends and all other changes that can be Excess Workers Compensation: AIG generally utilizes a combina-
quantified. The estimated loss cost trend utilized in the year-end tion of loss development methods and expected loss ratio
2006 reviews averaged approximately 6 percent for excess methods. Loss development methods are given the greater weight
casualty classes. Frequency/severity methods are generally not for mature accident years such as 2000 and prior. Expected loss
utilized as the vast majority of reported claims do not result in a ratio methods are given the greater weight for the more recent
claim payment. In addition, the average severity varies significantly accident years. Excess workers compensation is an extremely
from accident year to accident year due to large losses which long-tail class of business, with loss emergence extending for
characterize this class of business, as well as changing propor- decades. Therefore there is limited credibility in the reported
tions of claims which do not result in a claim payment. losses for many of the more recent accident years. Beginning with
D&O: AIG generally utilizes a combination of loss development the year-end 2005 loss reserve review, AIG’s actuaries began to
methods and expected loss ratio methods for D&O and related utilize claims projections provided by AIG claims staff to help
management liability classes of business. Expected loss ratio determine the loss development factors for this class of business.
methods are given more weight in the two most recent accident General Liability: AIG generally uses a combination of loss
years, whereas loss development methods are given more weight development methods and expected loss ratio methods for
in more mature accident years. Beginning with the year-end 2005 primary general liability or products liability classes. For certain
loss reserve review, AIG’s actuaries began to utilize claim classes of business with sufficient loss volume, loss development
projections provided by AIG claims staff as a benchmark for methods may be given significant weight for all but the most
determining the indicated ultimate losses for accident years 2004 recent one or two accident years, whereas for smaller or more
and prior. For the year end 2006 loss reserve review, claims volatile classes of business, loss development methods may be
projections for accident years 2005 and prior were utilized. In given limited weight for the five or more most recent accident
prior years, AIG’s actuaries had utilized these claims projections years. Expected loss ratio methods would be utilized for the more
as a benchmark for profitability studies for major classes of D&O recent accident years for these classes. The loss experience for
and related management liability business. The track record of primary general liability business is generally reviewed at a level
these claims projections has indicated a very low margin of error, that is believed to provide the most appropriate data for reserve
thus providing support for their usage as a benchmark in analysis. For example, primary claims made business is generally
determining the estimated loss reserve. These classes of busi- segregated from business written on an occurrence policy form.
ness reflect claims made coverage, and losses are characterized Additionally, certain subclasses, such as construction, are gener-
by low frequency and high severity. Thus, the claim projections ally reviewed separately from business in other subclasses. Due
can produce an accurate overall indicator of the ultimate loss to the fairly long-tail nature of general liability business, and the
exposure for these classes by identifying and estimating all large many subclasses that are reviewed individually, there is less
losses. Frequency/severity methods are generally not utilized for credibility in the reported losses and increased reliance on
these classes as the overall losses are driven by large losses expected loss ratio methods. AIG’s actuaries generally do not
utilize frequency/severity methods to test reserves for this
44 AIG 2006 Form 10-K