AIG 2007 Annual Report Download - page 108

American International Group, Inc. and Subsidiaries
Management’s Discussion and Analysis of
Financial Condition and Results of Operations Continued
Actuarial methods used by AIG for most long-tail casualty A key advantage of loss development methods is that they
classes of business include loss development methods and respond quickly to any actual changes in loss costs for the class
expected loss ratio methods, including ‘‘Bornhuetter Ferguson’’ of business. Therefore, if loss experience is unexpectedly deterio-
methods described below. Other methods considered include rating or improving, the loss development method gives full
frequency/severity methods, although these are generally used by credibility to the changing experience. Expected loss ratio meth-
AIG more for pricing analysis than for loss reserve analysis. Loss ods would be slower to respond to the change, as they would
development methods utilize the actual loss development patterns continue to give more weight to the expected loss ratio, until
from prior accident years to project the reported losses to an enough evidence emerged for the expected loss ratio to be
ultimate basis for subsequent accident years. Loss development modified to reflect the changing loss experience. On the other
methods generally are most appropriate for classes of business hand, loss development methods have the disadvantage of
which exhibit a stable pattern of loss development from one overreacting to changes in reported losses if in fact the loss
accident year to the next, and for which the components of the experience is not credible. For example, the presence or absence
classes have similar development characteristics. For example, of large losses at the early stages of loss development could
property exposures would generally not be combined into the cause the loss development method to overreact to the favorable
same class as casualty exposures, and primary casualty expo- or unfavorable experience by assuming it will continue at later
sures would generally not be combined into the same class as stages of development. In these instances, expected loss ratio
excess casualty exposures. Expected loss ratio methods are methods such as ‘‘Bornhuetter Ferguson’’ have the advantage of
generally utilized by AIG where the reported loss data lacks properly recognizing large losses without extrapolating unusual
sufficient credibility to utilize loss development methods, such as large loss activity onto the unreported portion of the losses for
for new classes of business or for long-tail classes at early stages the accident year. AIG’s loss reserve reviews for long-tail classes
of loss development. typically utilize a combination of both loss development and
Expected loss ratio methods rely on the application of an expected loss ratio methods. Loss development methods are
expected loss ratio to the earned premium for the class of generally given more weight for accident years and classes of
business to determine the loss reserves. For example, an business where the loss experience is highly credible. Expected
expected loss ratio of 70 percent applied to an earned premium loss ratio methods are given more weight where the reported loss
base of $10 million for a class of business would generate an experience is less credible, or is driven more by large losses.
ultimate loss estimate of $7 million. Subtracting any reported paid Expected loss ratio methods require sufficient information to
losses and loss expense would result in the indicated loss determine the appropriate expected loss ratio. This information
reserve for this class. ‘‘Bornhuetter Ferguson’’ methods are generally includes the actual loss ratios for prior accident years,
expected loss ratio methods for which the expected loss ratio is and rate changes as well as underwriting or other changes which
applied only to the expected unreported portion of the losses. For would affect the loss ratio. Further, an estimate of the loss cost
example, for a long-tail class of business for which only 10 per- trend or loss ratio trend is required in order to allow for the effect
cent of the losses are expected to be reported at the end of the of inflation and other factors which may increase or otherwise
accident year, the expected loss ratio would be applied to the change the loss costs from one accident year to the next.
90 percent of the losses still unreported. The actual reported Frequency/severity methods generally rely on the determination
losses at the end of the accident year would be added to of an ultimate number of claims and an average severity for each
determine the total ultimate loss estimate for the accident year. claim for each accident year. Multiplying the estimated ultimate
Subtracting the reported paid losses and loss expenses would number of claims for each accident year by the expected average
result in the indicated loss reserve. In the example above, the severity of each claim produces the estimated ultimate loss for
expected loss ratio of 70 percent would be multiplied by the accident year. Frequency/severity methods generally require a
90 percent. The result of 63 percent would be applied to the sufficient volume of claims in order for the average severity to be
earned premium of $10 million resulting in an estimated unre- predictable. Average severity for subsequent accident years is
ported loss of $6.3 million. Actual reported losses would be generally determined by applying an estimated annual loss cost
added to arrive at the total ultimate losses. If the reported losses trend to the estimated average claim severity from prior accident
were $1 million, the ultimate loss estimate under the ‘‘Bornhuet- years. Frequency/severity methods have the advantage that
ter Ferguson’’ method would be $7.3 million versus the $7 million ultimate claim counts can generally be estimated more quickly
amount under the expected loss ratio method described above. and accurately than can ultimate losses. Thus, if the average
Thus, the ‘‘Bornhuetter Ferguson’’ method gives partial credibility claim severity can be accurately estimated, these methods can
to the actual loss experience to date for the class of business. more quickly respond to changes in loss experience than other
Loss development methods generally give full credibility to the methods. However, for average severity to be predictable, the
reported loss experience to date. In the example above, loss class of business must consist of homogeneous types of claims
development methods would typically indicate an ultimate loss for which loss severity trends from one year to the next are
estimate of $10 million, as the reported losses of $1 million reasonably consistent. Generally these methods work best for
would be estimated to reflect only 10 percent of the ultimate high frequency, low severity classes of business such as personal
losses. auto. AIG also utilizes these methods in pricing subclasses of
professional liability. However, AIG does not generally utilize
54 AIG 2007 Form 10-K