AIG 2005 Annual Report Download - page 91

AMERICAN INTERNATIONAL GROUP, INC. AND SUBSIDIARIES
$400 million change (either positively or negatively) to the sensitivity analysis for this business, there is the potential for
overall AIG loss reserve position. The comparable effect on variations far greater than this amount (either positively or
the D&O and related management liability classes would be negatively). Likewise, in the judgment of AIG’s actuaries,
approximately $200 million (either positively or negatively) if five percent is considered an appropriate benchmark for
future loss development factors differed by five percent from sensitivity analysis with respect to the loss development factor
those utilized in the year-end 2005 loss reserve review. For the assumptions used to test the reserves.
excess workers compensation class, the effect would be For D&O and related management liability classes of
approximately $150 million (either positively or negatively). business, the loss cost trend assumption has been critical in
For workers compensation reserves, other than excess workers past analyses, however, because of the utilization of claim
compensation reserves, the effect of a five percent deviation projections for all claims through accident year 2004, as
from the loss development factors utilized in the year-end 2005 described above, the sensitivity of the reserves to this
reserve reviews would be approximately $1.0 billion (either assumption has been reduced in 2005. The loss development
positively or negatively). Because loss development factors for factor assumption is important but less critical than for excess
this class have shown less volatility than higher severity classes casualty. As this coverage is written on a claims-made basis,
such as excess casualty, however, actual changes in loss claims for a given accident year are all reported within that
development factors are expected to be less than five percent. year. Actual changes in loss costs from one accident year to
There is some degree of volatility in loss development patterns the next in the 1990s ranged from double digit negative values
for other longer tail liability classes as well. However, the for several accident years in the early 1990s to nearly
potential effect on AIG’s reserves would be much less than for 50 percent per year for the period from accident year 1996 to
the classes cited above. accident year 1999. Thus, there is the potential for extreme
The calculations of the effect of the five percent change in volatility in loss costs for this business and, although five per-
loss development factors are made by selecting the stage of cent is considered a reasonable benchmark for sensitivity
accident year development where it is believed reasonable for analysis, there is the potential for variations far greater than
such a deviation to occur. For example, for workers compensa- this amount (either positively or negatively). Five percent is
tion, the $1.0 billion amount is calculated by assuming that also considered an appropriate benchmark for sensitivity
each of the most recent eight accident years develop five per- analysis with respect to the loss development factor assump-
cent higher than estimated by the current loss development tions used to test the reserves for these classes. However, as
factors utilized in the reserve study, i.e. the factor 1.05 is noted above, the effect of such a deviation is less than that of
multiplied by the incurred losses (including IBNR and loss a similar deviation in loss cost trends. For the excess workers
expenses) for these accident years. compensation class, the loss development factor assumptions
AIG management believes that using a five percent change are critical and the loss trend assumption is important but not
in the assumptions for loss cost trends and loss development as critical. Excess workers compensation is an extremely long
factors provides a reasonable benchmark for a sensitivity tail class of business, hence there is a much greater than
analysis of the reserves of AIG’s most significant lines of normal uncertainty as to the appropriate loss development
general insurance business. For excess casualty business, both factors for the tail of the loss development. As noted above,
the loss cost trend and the loss development factor assumptions the effect of a five percent change in the loss development
are critical. Generally, actual historical loss development factor is approximately $150 million. It would not be
factors are used to project future loss development. However, uncommon for the loss development factors to deviate by
there can be no assurance that future loss development greater than five percent for this class of business.
patterns will be the same as in the past. Moreover, as excess For workers compensation, the loss development factor
casualty is a long-tail class of business, any deviation in loss assumptions are important. Generally, AIG’s actual historical
cost trends or in loss development factors might not be workers compensation loss development would be expected to
discernible for an extended period of time subsequent to the provide a reasonably accurate predictor of future loss develop-
recording of the initial loss reserve estimates for any accident ment. A five percent sensitivity indicator for workers compen-
year. Thus, there is the potential for the reserves with respect sation would thus be considered to be toward the high end of
to a number of accident years to be significantly affected by potential deviations for this class of business. The loss cost
changes in the loss cost trends or loss development factors that trend assumption for workers compensation is not believed to
were initially relied upon in setting the reserves. These changes be material with respect to AIG’s loss reserves. This is
in loss trends or loss development factors could be attributable primarily because AIG’s actuaries are generally able to use loss
to changes in inflation or in the judicial environment, or in development projections for all but the most recent accident
other social or economic phenomena affecting claims. For year’s reserves, so there is limited need to rely on loss cost
example, during the lengthy periods during which losses trend assumptions for workers compensation business.
develop for excess casualty, actual changes in loss costs from For casualty business other than the classes noted above,
one accident year to the next have ranged from negative there is generally some potential for deviation in both the loss
values to double-digit amounts. Thus, there is the potential for cost trend and loss development factor selections. However,
significant volatility in loss costs for excess casualty and, the effect of such deviations would not be material when
although five percent is considered a reasonable benchmark for compared to the effect on the classes cited above.
AIG m Form 10-K 39