SecProbe.io

CORRESP
1
filename1.htm

corresp

[AIG Letterhead]

February 18, 2010

Mr. Jim B. Rosenberg

Senior Assistant Chief Accountant

United States Securities and Exchange Commission

Division of Corporation Finance

100 F Street, NE

Mail Stop 4720

Washington, D.C. 20549

    Re:

    American International Group, Inc.

Form 10-Q for the Fiscal Quarter Ended September 30, 2009

File No. 001-8787

Dear Mr. Rosenberg:

We are in receipt of your letter dated February 3, 2010 with respect to American International
Group, Inc.’s (AIG) Quarterly Report on Form 10-Q for the fiscal quarter ended September 30, 2009
(Form 10-Q). This letter sets forth AIG’s responses to the Staff’s comments contained in your
letter.

AIG acknowledges that the adequacy and accuracy of the disclosure in the Form 10-Q is the
responsibility of AIG, that Staff comments or changes to disclosure in response to Staff comments
do not foreclose the Securities and Exchange Commission (the “Commission”) from taking any action
with respect to the Form 10-Q and that Staff comments may not be asserted by AIG as a defense in
any proceeding initiated by the Commission or any person under the Federal securities laws of the
United States.

We have repeated your comments below to facilitate your review.

Notes to Consolidated Financial Statements

12. Federal Income Taxes, page 79

    1.

    Please refer to prior comment two. Regarding the tax expense associated with
investment in subsidiaries, revise to disclose the information provided in your
response. Ensure that your revised disclosure complies with paragraph ASC 740-30-50-2.

     AIG Response:

AIG intends to provide the information included in its response to prior comment two
in its Annual Report on Form 10-K (Form 10-K) for 2009, and will ensure that this
disclosure complies with paragraph ASC 740-30-50-2.

Management’s Discussion and Analysis of Financial Condition and Results of Operations

Liability for Unpaid Claims and Claims Adjustment Expense, page 123

    2.

    Please refer to prior comment three. Please provide us with the following
information.

    a.

    Tell us whether the Company has or intends to request the
“actual analysis” underlying the Bernstein report. If not, tell us why you
believe further investigation of this issue is not necessary.

    b.

    Describe the specific analysis and assertions included in the
Bernstein report and provide us with your evaluation of them.

    c.

    Explain why you believe that the actual analysis underlying the
Bernstein report may have been based on Schedule P data that you have cautioned
is not appropriate for reserve adequacy determinations.

    d.

    Tell us whether you have any reason to believe that your
reserves as reported were not adequate other than as you have disclosed for
changes in prior year estimates.

     AIG Response:

a.       Tell us whether the Company has or intends to request the “actual analysis”
underlying the Bernstein report. If not, tell us why you believe further
investigation of this issue is not necessary.

AIG requested the actuarial analysis underlying the Bernstein report (Report)
immediately after the report was issued and was informed that the analysis was
proprietary and could not be made available to AIG. However, AIG was able to engage
in oral discussion with the authors regarding sources of data and certain aspects of
their study as described below.

b.       Describe the specific analysis and assertions included in the Bernstein report
and provide us with your evaluation of them.

AIG does not have access to the analysis underlying the Report and therefore it is
not possible for AIG to provide a comprehensive evaluation of the analysis
underlying the Report. The following evaluation is necessarily speculative and AIG
can give the Staff no assurance that the evaluation is correct.

AIG contacted the authors to inquire about the actuarial data, methodologies, and
assumptions used in the analysis to determine the reserve estimates. The authors
confirmed that they had used Schedule P data, which refers to the “Analysis of Loss
and Loss Expense” schedules filed by AIG with state regulators for individual legal
entities (Schedule P data). The authors also confirmed that they

2

used
a reserving methodology outlined in a Bernstein background paper
published in October 2008.

AIG attempted to recreate the Bernstein reserve indications using the published
Schedule P data and the methodology described in the background paper, modified as
described by the authors during subsequent discussions. Based on this work AIG
identified errors in both the data and the methods used:

    •

    Data — As described in AIG’s previous comment letter response dated
January 15, 2010, AIG had cautioned users of Schedule P data that “It is not possible
for any person to accurately determine the adequacy of the company’s loss and
loss expense reserves using the Company’s Schedule P data as the sole source.”
Statutory annual statements, including Schedule P, must be prepared as
specified by rules and regulations of the state of domicile. There is little
ability to modify schedules for facts and circumstances unique to a filer.
Further, the accounting conventions used for such statements can vary
significantly from U.S. GAAP (e.g., gross vs. net presentation of reinsurance).
Application of the rules and regulations by AIG’s U.S.- based insurance
companies resulted in the anomalies that necessitated the cautionary language.

    •

    Specifically, in 2008 AIG transferred certain portfolios of business from
certain of its U.S.-based insurance companies to certain of its foreign-based
insurance companies. In accordance with the requirements of annual statement
preparation, such transfers were recorded as paid losses and reductions in loss
reserves. These portfolio transfers significantly distorted both paid loss
statistics and loss reserve information. Because the foreign-based entities do
not file annual statements in the U.S., the offsetting transactions were not
included in the data used by the Bernstein analysts.1 Further,
transfers between two U.S.-based entities can also create anomalies in paid
loss and reserve data and in 2008 there were similar portfolio transfers among
U.S.-based subsidiaries as well as reinsurance between wholly-owned
subsidiaries. Moreover, there were also changes in the mix of AIG’s businesses
during 2008 that the Bernstein analysis did not appropriately consider, which
AIG believes resulted in inaccurate reserve estimates.

    •

    Methods — AIG uses several standard actuarial methods, which differ by major
class of business based upon the claims reporting nature of the classes of
business, as described in AIG’s Form 10-K and Form 10-Q reports. AIG also
employs the assistance of claims staff and external actuaries to perform
independent calculations of reserves to confirm AIG’s estimates of reserves.
The Bernstein analysis is premised on a loss reserve methodology that differs
somewhat from most standard actuarial methodologies. Most actuarial analyses
are premised on the employment of multiplicative loss development

    1

    The portfolio transactions were eliminated in AIG’s
consolidated financial statements included in its filings with the Commission.

3

    factors, which are derived from historical development patterns for each class
of business being analyzed. For longer tail lines of business, where recent
accident years have immature loss experience, expected loss ratio methods that
do not rely on the multiplicative factors would generally be employed. The
Bernstein approach relies instead on the employment of incremental loss ratios,
which are derived based on a statistical analysis of historical premiums and
losses for each line of business being analyzed. The Bernstein statistical
analysis employs a logarithmic transformation of the incremental loss ratio
data. The effect of this logarithmic transformation is that increased
variability in historical incremental loss ratios results in higher estimates of
future incremental loss ratios. Any distortions that increase the apparent
variability of historical results, such as those that would be created by
relying solely on the Schedule P data cited above, will increase the estimates
of reserves even when the variability is unrelated to the underlying development
of losses. Also, it is important to ensure that data is segmented into groups
that are homogeneous with respect to development characteristics. If two
populations of claims, each with different development patterns, are combined
and the mix between the two populations varies by period, this will create
variability in the observed incremental loss ratios, resulting in overestimation
of reserves. This combination of populations occurs frequently when aggregating
classes of business that are not considered homogeneous for actuarial purposes
(and for which different methodologies are employed) into the major Schedule P
classes of business.

    AIG believes a large portion of the difference between the Bernstein reserve
indications and AIG’s actual reserves was due to the use of the distorted
Schedule P data but AIG is not able, without access to the Bernstein model, to
determine what additional errors, if any, may exist in their calculations. By
applying the Bernstein model as understood by AIG and using the published
Schedule P data, AIG was able to approximate the reserve amounts reported by
Bernstein. Using the same model with the Schedule P data adjusted to remove the
anomalies, the result approximated AIG’s carried reserves reported in AIG’s 2008
Form 10-K. Because AIG was denied access to the Bernstein model, AIG was unable
to determine the extent to which the remaining difference is due to differences
in assumptions and judgments the authors applied, to their non-standard
methodologies, to multiple classes of business, or to other distortions that
could be created by the use of Schedule P data as described above.

    c.

    Explain why you believe that the actual analysis underlying the
Bernstein report may have been based on Schedule P data that you have cautioned
is not appropriate for reserve adequacy determinations.

As indicated in the response to (b) above, the authors confirmed to AIG that they
utilized the Schedule P data for their analysis.

4

    d.

    Tell us whether you have any reason to believe that your
reserves as reported were not adequate other than as you have disclosed for
changes in prior year estimates.

During
AIG’s 2008 year end and 2009 quarterly reserve reviews, as well as during the
process described in the response to (b) above, AIG did not identify any matters
that indicated its reserves as reported at December 31, 2008 or any 2009 quarterly
period were not adequate. In each of AIG’s 2009 Form 10-Qs, AIG disclosed both
adverse and favorable development in certain lines of business and the amount of
changes in prior years’ estimates resulting from such 2009 development. As disclosed
in AIG’s 2008 Form 10-K, a comprehensive annual loss reserve review is performed in
the fourth quarter of each year. AIG is currently finalizing this review. AIG is also considering
the results of its third party actuary’s analysis in forming its judgment regarding
reserve adequacy.

If you have any questions or require any additional information, please do not
hesitate to contact me at (212) 770-5123.

    Very truly yours,

    /s/ Kathleen E. Shannon

    Kathleen E. Shannon
Senior Vice
President,
Secretary & Deputy General Counsel

5