AIG 2007 Annual Report Download - page 111

American International Group, Inc. and Subsidiaries
test the loss reserves. AIG maintains a data base of detailed position of using alternative loss trend or loss development factor
historical premium and loss transactions in original currency for assumptions rather than those actually used in determining AIG’s
business written by Foreign General Insurance, thereby allowing best estimates in the year-end loss reserve analyses in 2007.
AIG actuaries to determine the current reserves without any The analysis addresses each major class of business for which a
distortion from changes in exchange rates over time. In testing material deviation to AIG’s overall reserve position is believed
the Foreign General Insurance reserves, AIG’s actuaries segment reasonably possible, and uses what AIG believes is a reasonably
the data by region, country or class of business as appropriate to likely range of potential deviation for each class. There can be no
determine an optimal balance between homogeneity and assurance, however, that actual reserve development will be
credibility. consistent with either the original or the adjusted loss trend or
loss development factor assumptions, or that other assumptions
Loss Adjustment Expenses: AIG determines reserves for legal made in the reserving process will not materially affect reserve
defense and cost containment loss adjustment expenses for each development for a particular class of business.
class of business by one or more actuarial methods. The methods
generally include development methods analogous to those de- Excess Casualty: For the excess casualty class of business, the
scribed for loss development methods. The developments could assumed loss cost trend was approximately five percent. After
be based on either the paid loss adjustment expenses or the ratio evaluating the historical loss cost trends from prior accident years
of paid loss adjustment expenses to paid losses, or both. Other since the early 1990s, in AIG’s judgment, it is reasonably likely
methods include the utilization of expected ultimate ratios of paid that actual loss cost trends applicable to the year-end 2007 loss
loss expense to paid losses, based on actual experience from reserve review for excess casualty will range from negative five
prior accident years or from similar classes of business. AIG percent to positive 15 percent, or approximately ten percent lower
generally determines reserves for adjuster loss adjustment ex- or higher than the assumption actually utilized in the year-end
penses based on calendar year ratios of adjuster expenses paid 2007 reserve review. A ten percent change in the assumed loss
to losses paid for the particular class of business. AIG generally cost trend for excess casualty would cause approximately a
determines reserves for other unallocated loss adjustment ex- $2.4 billion increase or a $1.6 billion decrease in the net loss
penses based on the ratio of the calendar year expenses paid to and loss expense reserve for this class of business. It should be
overall losses paid. This determination is generally done for all emphasized that the ten percent deviations are not considered
classes of business combined, and reflects costs of home office the highest possible deviations that might be expected, but rather
claim overhead as a percent of losses paid. what is considered by AIG to reflect a reasonably likely range of
potential deviation. Actual loss cost trends in the early 1990s
Catastrophes: Special analyses are conducted by AIG in response were negative for several years, including amounts below the
to major catastrophes in order to estimate AIG’s gross and net negative five percent cited above, whereas actual loss cost trends
loss and loss expense liability from the events. These analyses in the late 1990s ran well into the double digits for several years,
may include a combination of approaches, including modeling including amounts greater than the 15 percent cited above. Thus,
estimates, ground up claim analysis, loss evaluation reports from there can be no assurance that loss trends will not deviate by
on-site field adjusters, and market share estimates. more than ten percent. The loss cost trend assumption is critical
AIG’s loss reserve analyses do not calculate a range of loss for the excess casualty class of business due the long-tail nature
reserve estimates. Because a large portion of the loss reserves of the claims and therefore is applied across many accident
from AIG’s General Insurance business relates to longer-tail years.
casualty classes of business driven by severity rather than For the excess casualty class of business, the assumed loss
frequency of claims, such as excess casualty and D&O, develop- development factors are also a key assumption. After evaluating
ing a range around loss reserve estimates would not be the historical loss development factors from prior accident years
meaningful. Using the reserving methodologies described above, since the early 1990s, in AIG’s judgment, it is reasonably likely
AIG’s actuaries determine their best estimate of the required that actual loss development factors will range from approximately
reserve and advise Management of that amount. AIG then adjusts 3.5 percent below those actually utilized in the year-end 2007
its aggregate carried reserves as necessary so that the actual reser ve review to approximately 6.5 percent above those factors
carried reserves as of December 31 reflect this best estimate. actually utilized. If the loss development factor assumptions were
changed by 3.5 percent and 6.5 percent, respectively, the net
Volatility of Reserve Estimates and Sensitivity Analyses loss reserves for the excess casualty class would decrease by
approximately $600 million under the lower assumptions or
As described above, AIG uses numerous assumptions in determin-
increase by approximately $1.0 billion under the higher assump-
ing its best estimate of reserves for each class of business. The
tions. Generally, actual historical loss development factors are
importance of any specific assumption can vary by both class of
used to project future loss development. However there can be no
business and accident year. If actual experience differs from key
assurance that future loss development patterns will be the same
assumptions used in establishing reserves, there is potential for
as in the past, or that they will not deviate by more than the
significant variation in the development of loss reserves, particu-
amounts illustrated above. Moreover, as excess casualty is a long-
larly for long-tail casualty classes of business such as excess
tail class of business, any deviation in loss cost trends or in loss
casualty, D&O or workers compensation. Set forth below is a
development factors might not be discernible for an extended
sensitivity analysis that estimates the effect on the loss reserve
AIG 2007 Form 10-K 57