(This testimony today is being delivered today to the FCIC)
Chairman Angelides, Vice Chairman Thomas, and Members of the Commission:
Thank you for having me here to provide information. I hope my testimony is helpful in
understanding fully the circumstances we faced at American International Group, Financial
Products Division (“AIG-FP”) in the period between August 2007 and February 2008.
As you may know, AIG-FP had a diverse portfolio. We were active in the capital
markets, doing business in more than 25 countries. We had offices in the United States, the
United Kingdom, Hong Kong, Japan, and France. AIG-FP was a dealer in derivatives and ran
active portfolios in interest rates, commodities, currency, credit, and equity derivatives. At its
peak, AIG-FP had more than 400 employees. When I became Chief Executive Officer in 2002,
it was my distinct honour to manage such a diverse business with so many talented employees.
Understandably, you asked about one product in particular, the credit-default swaps
(“CDSs”) on super-senior tranches of multi-sector collateralized debt obligations (“CDOs”). As
you know, a CDS is a contract that provides insurance-like protection against a risk of default.
In the case of the multi-sector CDOs, the risk was on the super-senior layer of CDOs that
contained a variety of debt — mostly residential-mortgage-backed securities (with both prime
and subprime collateral), but also commercial-real-estate loans, corporate loans, consumer loans,
and auto-loan receivables.
Often repeated are my words during an earnings call in August 2007 that I did not expect
any realised, economic losses (as opposed to unrealized accounting losses) on this portfolio. I
meant exactly what I said in August 2007. The underlying loans in the CDOs were diversified,
and we insured only the super-senior tranche, which always had a AAA layer of loans below it. I
did not expect actual, economic losses on the portfolio. That said, I was truthful at all times
about the unrealized accounting losses and did my very best to estimate them accurately, in
consultation with others at AIG-FP, as well as with my supervisors, AIG’s senior accounting
staff, and its internal and external auditors.
What is also true is this: we were “long” the housing market through the CDS contracts
because we believed until late 2005 that banks and other mortgage originators were applying
appropriate standards when writing mortgages. When we recognised — well before many others
— that changes in the mortgage market likely presented increased risk for future deals, we
decided to exit the subprime business. We thought the decision was appropriate, despite the lost
profits at the time. With hindsight, the decision looks even more prudent.
You asked about the approval and monitoring system for the multi-sector CDS portfolio.
The approval process had several stages, and the AIG parent organisation reviewed every single
transaction under its promulgated standards.
The first stage in the approval process was internal at AIG-FP. Every proposed CDS
transaction first was reviewed by an experienced AIG-FP credit officer, who performed an
independent analysis of the proposed deal. The credit officer analysed the trade fundamentals,
the proposed terms and conditions, and the asset classes.
After that initial review, University of Pennsylvania Professor Gary Gorton, who served
as a consultant to AIG-FP, worked with our experienced analysts to refine the deal structure. To
complete this review, Professor Gorton used a sophisticated actuarial model to make sure the
proposed deal was fundamentally sound and to determine an appropriate attachment point. This
process was designed to minimize risk to AIG-FP.
After AIG-FP reviewed a transaction, it could not be executed without AIG’s specific
review — and only upon its approval. AIG’s credit risk officer, part of AIG’s Credit Risk
Committee (“CRC”), which was, in turn, part of AIG’s Enterprise Risk Management group
(“ERM”), first reviewed a detailed memorandum about the transaction. The CRC consisted of
many of AIG’s most senior and experienced employees, including its CFO, Head of Enterprise
Risk, Chief Credit Officer, Vice President for Financial Services, and the Vice Chairman for
Investments. Each memorandum provided an overview of the proposed transaction, including a
description of the CDO structure, its assets and liabilities, and the terms of the proposed CDS.
The memorandum also specifically identified risk factors and provided an analysis of those risks.
Virtually every deal, including all CDS transactions above $250 million, required the CRC’s
approval, and the rare transaction with a net notional exposure of more than $5 billion also
required approval from either AIG’s Chief Executive Officer or AIG’s Chief Risk Officer. AIG’s
Chief Credit Officer oversaw this review.
After AIG approved a transaction, AIG-FP continued its diligence throughout the deal’s
lifetime by conducting rigorous reviews to identify unanticipated risk. We reviewed the
portfolio every month. We conducted a more in-depth analysis quarterly. AIG-FP ran stress
tests using its model to determine whether each transaction continued to meet AIG’s standards.
We conducted this review with actual performance data. The ERM group reviewed the results of
AIG-FP’s quarterly analyses.
In addition, AIG’s CRC conducted an annual review of AIG-FP’s portfolio. The CRC
included several of AIG’s senior executives. The CRC’s review was independent of AIG-FP’s
quarterly reviews. It was aimed at separately identifying any changes in the deals that led to
I do not believe this highly structured approval and review process changed over time in
any significant way. At bottom, AIG and AIG-FP were in frequent communication, both
formally and informally, about AIG-FP’s super-senior CDS portfolio during the approval and
You also asked about the growth of the CDS portfolio over time, and, in particular, in
2005. As you know, we made a decision at the end of 2005 to stop writing new deals that
contained subprime collateral. Although we completed deals already in the pipeline, the
portfolio grew comparatively little after 2006. Until about 2005, I suspect the pace of our growth
was consistent with the growth of the CDO business generally. In dollar terms, our business
grew every year because the CDS contracts were multi-year obligations: adding even one
increased the notional size of the book. But, on a percentage basis, our biggest increases were
prior to 2005. In fact, 2005 was a sizeable increase in dollars, but a far smaller percentage
increase than the prior years. I suspect it was much smaller than the overall growth in the
broader CDO market at the time. I recall in 2006, after we announced our exit from subprime
transactions, some market participants were just announcing expansions of their CDO
businesses. During the short period of growth and then decline in our multi-sector CDO
business, we never compromised our standards to enter into a transaction. Every deal had to
stand up to the rigorous evaluation process.
Let me explain more fully why AIG-FP stopped writing CDS protection on CDOs with
subprime collateral. In mid-2005, AIG-FP’s head trader was concerned that changes in the
subprime mortgage market may have meant more risk in new deals compared to existing ones.
We discussed taking a harder look at the market, and I asked for his recommendation. In the
third quarter of 2005, the head trader expressed continued concern and asked to convene a group to gather additional information. I agreed his course was a prudent one, and we selected a team
to assist in the fact-gathering and analysis.
After that detailed review and analysis, the head trader believed AIG-FP should stop
writing CDS protection on CDOs with subprime exposure. With that analysis, and even though
opinions varied within the group, we decided to stop writing deals with subprime exposure. Weannounced our decision to the marketplace in February 2006.
We made this decision, despite the immediate impact on our profits, for one reason: it
was the right thing to do to protect the long-term health of our business. We hope that, with this
decision, we helped limit risk.
When AIG-FP exited the business, we wondered whether there was a risk in our existing
portfolio for which a hedge might be appropriate. After internal discussion, we believed that the
existing vintages were not tainted by any slippage in underwriting standards. We remained
confident in our risk analysis for the existing deals. The decision was based on our firm view
that the portfolio would not experience realised losses (as opposed to short-term accounting
losses). As I look at the performance of some of these same CDOs in Maiden Lane III, I think
there would have been few, if any, realised losses on the CDS contracts had they not been
unwound in the bailout.
But, as you know, a decision was taken after my departure from AIG-FP to unwind the
CDS contracts due largely, so far as I can tell, to the proliferation of collateral calls under the
CDS contracts. Generally under the credit support annexes to the CDS contracts, collateral calls
were triggered by the reduction in fair value of the underlying CDOs and the reduction in the
credit rating for AIG or the CDO. For example, if AIG was downgraded, a counterparty
generally could call for the posting of collateral, but only if the diminution in fair market value
was more than a stated percentage. If the parties disagreed about the fair market value of the
bonds (which would likely happen in a period of illiquidity when bonds did not trade), the
contracts generally had dispute-resolution processes, such as a so-called “dealer poll.”
During my tenure, AIG-FP had disagreements with several counterparties relating to calls
for collateral on AIG-FP’s CDSs. I directed my team to conduct thorough diligence to determine
both the basis for the numbers underlying the collateral calls and to reach the most reliable
valuation of AIG-FP’s book. As we dealt with collateral calls, which escalated as the market
deteriorated, we used available contractual defenses, relied on fundamental analysis to challenge
the values underlying the counterparties’ demands, and were prepared to call for dealer polls as
necessary. Through this strategy, we got steep reductions in the called-for amounts from all
counterparties who made the largest calls. During my tenure, no counterparty declared us in
breach or threatened litigation, which shows our strategy was effective. I believe this strategy
was appropriate and in the best interests of the company and its shareholders.
You have asked about the way in which we calculated fair market value for our portfolio.
This was, to be sure, a challenging, first-of-its-kind process at AIG-FP, but I believe we
developed a reasonably strong analytic method for estimating fair value. We selected a model
developed by Moody’s (the Binomial Expansion Technique, or “BET”) in September 2007. We
customised it to provide the most accurate valuation estimate possible. We eventually developed
two versions of the model, as a cross-check, using different sets of inputs. As is true with all
models, our model had several assumptions and adjustments. We reviewed all the assumptions
and adjustments with a large team from AIG and its outside auditors, which included executives
with dozens of years of public-accounting experience (I am not a CPA).
You asked about one adjustment in particular, the so-called negative-basis adjustment,
including our auditors’ decision to disallow it. We believed then, and I still believe now, that it
was a wholly appropriate adjustment.
At the time, in light of the great unrest in the market, we gathered the best data we could
to evaluate the appropriateness of, and estimate the magnitude of, the negative-basis adjustment.
Within days of developing the blueprint for the version of the model that required the
adjustment, in late November 2007, we discussed the need for the adjustment specifically with
AIG and its outside auditors. We continually made specific references to the adjustment in the
weeks that followed.
No one raised concerns about the negative-basis adjustment until mid-January 2008.
Ultimately, in February 2008, the auditors decided that we did not have sufficient audit-quality
evidence for the adjustment. As a result, they disallowed it. I disagreed with that decision, as
did several others at AIG. It was, however, the auditors’ final decision and we had to abide it.
Regardless, the auditors’ decision about the quantum of evidence supporting the
adjustment had a huge impact on the company: it increased our unrealized accounting loss
substantially. The auditors also made a material-weakness finding, which I first learned about in
February 2008. In light of the auditors’ heavy involvement in the fair-market-model evolution
generally, and their prior knowledge of the existence and magnitude of the negative-basis
adjustment in particular, I also found the material-weakness finding surprising, to say the least. I
know AIG senior management argued strenuously against it.
After the material-weakness finding, I was called into a meeting with AIG’s CEO in
February 2008. Suggesting that “changes” would have to be made, he offered me an opportunity
to retire, which I took. Management asked me to stay on in a consulting role without seeking
other employment. I agreed to the arrangement.
I am glad you asked about the compensation structure at AIG-FP. We designed AIG-FP’s
compensation system to make sure that highly compensated employees had a large percentage of
their compensation deferred and tied to the company’s long-term performance. We adopted this
plan approximately 15 years ago and changed it little after that. The plan generally required
highly compensated employees to defer up to 45% of their compensation, through the purchase
of AIG-FP subordinated debt, which was not guaranteed and was at risk to the business’s
performance. The deferred compensation was paid out annually during an average-life window
(usually between four and six years). Although the money paid out over time, it vested
immediately. Employees at some other financial institutions were essentially handcuffed to their
jobs because leaving meant they would lose unvested, deferred compensation behind. Under
AIG-FP’s plan, any employee could leave without forfeiting any portion of their deferred
compensation, which would continue to pay out over the average-life window even if an
employee departed. In other words, if anyone did not like working with me, or did not like what
we were doing, they could walk out the door without losing a penny in deferred compensation.
In November 2007, it became apparent that AIG-FP’s accounting losses would be
substantial and would require a change to our compensation structure to ensure that employees
stayed with the company to help it address the issues surrounding the AIG-FP portfolio, but also
would not be immune to AIG-FP’s losses if they were actually realised. I had several discussions
with my superiors at AIG about this change, emphasising the need to recognise the accounting
losses while also noting the importance of keeping our employees together during this critical
time. For that reason, I suggested that AIG-FP adopt a special-incentive plan (“SIP”), which
would place any compensation in excess of a set amount in a special deferred-compensation
account. The funds in that account would remain subject to AIG-FP’s business performance and
the risk of realisation of the accounting losses. Unlike the standard deferred compensation, one-
third of the SIP funds would vest at the end of 2008 and the remaining two-thirds would vest at
the end of 2009, and the funds would have been paid in 2012. My goal in proposing the SIP was
to recognise the unrealized accounting losses and to put highly compensated employees’ pay at
risk to those losses becoming realised, while also providing an incentive for them to remain with
the company through the end of 2009.
Although I was pleased with the SIP and believed it was a fair and appropriate plan given
the magnitude of the accounting losses, I went one step further and volunteered to take no bonus
whatsoever for 2007. Instead, I proposed to AIG that we simply have a handshake agreement
that AIG would compensate me in the future if AIG-FP’s accounting losses reversed, as I was
confident they would. I did not make this offer because anyone asked me to — I did it because it
was the right thing to do, even though AIG management ultimately rejected my proposal.
In closing, I understand that identifying the root causes of the financial crisis is an
extraordinarily important, yet dauntingly complex, mission. I greatly appreciate the professional
and objective manner in which the Commission’s staff has dealt with me and my counsel, and I
am available for any questions.