Innovation and Controls on Wall Street

Ryan began by thinking he ought to compare AIG’s control systems with the best of its trading partners. If AIG was going to trade with J.P. Morgan and Goldman Sachs, it would need risk management that could stand up to business with such counterparties.

When most people think of J.P. Morgan, they think of a bank that makes loans. When they think of Goldman Sachs, they think of an investment bank that takes clients to capital markets. By the mid-1990s, if they looked deep inside each firm, they would have been very surprised. Both firms had evolved into something more akin to hedge funds with giant balance sheets and banking sidelines. By 1993, Morgan, the commercial bank, derived 75% of revenues from investment banking and trading.⁴ Around the same time, trading revenues outstripped investment banking fees at Goldman.

Several trends spurred these evolutions. For one, core franchises at both commercial and investment banks had commoditized. Consequently, fees, spread, and margins had eroded. Starting in the 1980s, banking leaders set off in search of new, more lucrative endeavors.

Their efforts fostered a dramatic era in financial products innovation. Throughout the 1980s and into the 1990s, Wall Street’s banks produced one new product after another. This tidal wave of innovation generated fees, but also produced a second effect. Innovators and their competitors discovered that the new instruments made perfect trading vehicles. Both new and complex, the instruments were hard for markets to value. This gave the innovators and their quantitatively adept imitators informational and trading advantages. These trading opportunities soon saw the most skillful players reaping immense profits. In 1993, Goldman made its largest profit ever, $2.7 billion pre-tax. More than 50% came from proprietary trading.⁵

Trading, however, is a zero sum game. For every winner there is a loser, and it is difficult for any player to win consistently; even the best traders hit cold streaks. Goldman found this out with a vengeance, as it began to lose hundreds of millions of dollars from misplaced trades during 1994.⁶

As a result of such experiences, the better managed banks revisited their risk management systems. Trading was not a new activity. Basic control principles and frameworks for trading already existed. The question was how to adapt controls in light of trading’s enhanced size and importance.

The most basic control principle is to limit the capital allocated to trading. Such practices contain the position sizes traders could put on, limiting potential for losses. Capital limits are applied on multiple levels. Individual traders have limits. So do major trading groups, e.g., currencies, convertible securities, distressed debt, etc. By themselves, however, capital controls had not proved to be adequate for managing trading risks. Traders by nature believe in their “views.” Inevitably, they press for more capital to bet. Inevitably, the more successful of them get larger allocations. Capital allocations thus tend to ratchet up over time until some disastrous bet sends the process spiraling in reverse.

To avoid boom/bust trading cycles, Wall Street firms typically use a second control, the “stop loss,” to complement capital limits. Stop losses require traders to unwind trades if their positions hit a predetermined loss figure. As an example, consider a $10 M “long” position in Ford stock. Applying a stop loss might require the stock to be sold if prices decline such that the position is $200,000 “underwater.” Such stop loss controls force traders to rethink their strategies and reposition at better prices while also keeping losses in bounds. Stop loss controls also function under mark-to-market accounting. Under these rules, banks reflect new market prices for securities on their books at the close of each day. Doing so allows for a daily assessment of gains/losses by trading position. Those positions which have hit their stop loss are then slated for unwinding during subsequent trading.

Traders generally dislike “stop loss” controls. Most believe in their trades, and favor giving them time to prove correct. Many times they will react to losses by wanting to increase the bet. For traders, stop loss controls can mean the worst of both worlds—their trade goes down in the books as a loser, even if the trader’s strategy is eventually vindicated.

To move beyond the constant battles with traders about capital and stop losses, financial institutions found it useful to manage risk on a firmwide basis. This allows a powerful risk management committee to look across all sectors and decide how much risk appetite it has by sector. Thus, a given institution can decide to increase risk in oil and gold commodities, while cutting back in European sovereign debt. Sector limits are set, and group managers derive capital and stop loss limits for their subgroups and traders. This approach tends to curtail some of the arguments at the trading desk level. Decisions come down from on high. To reverse them, traders must arm their managers with compelling reasons to go back up the management chain. In the meantime, controls are in place and they stick.

To practice aggregate risk management, firms need both good information systems and a framework that permits “apples-to-apples” comparisons across different products. Not all financial products display equal price sensitivity. Some, like equity warrants, may show a 5% price response to a 1% move in the underlying stock. The advent of “derivative” securities brought more of the same. Instruments like options and exchange traded indexes reference underlying securities. However, their prices also are sensitive to other factors, i.e., the large potential payoff relative to capital at risk, and in the case of options to the time for exercising before expiration. Such factors make derivative prices more volatile than the underlying asset being referenced. Some way had to be found to compare such outsized gain/loss potential with that of more standard securities.

J.P. Morgan developed a way to do so during the early 1990s. Dubbed “Value at Risk” (VAR), it measured the firm’s potential for loss on all holdings during the approaching 24 hour period. VAR asked a simple question—assuming “normal” trading markets, what was the maximum that the firm could lose during the next trading day?⁷ “Normal” trading markets was a rubric for statistical price data capturing 95% of the historic daily price volatility for all relevant trading. Price movement way up or down, representing 2.5% of outcomes in either direction, were excluded from consideration. The analysis then took the worst price case within the 95% confidence band, applied it to the firm’s holdings, and computed aggregate potential losses by sector and for the firm.

VAR has many attractions. For one, it permits an “apples-to-apples” comparison across trading positions whose nature and behavior may differ greatly. VAR also gives risk managers a convenient tool to spot risk “hot spots” where the firm’s exposure to loss might exceed management’s appetite. VAR is easy to understand, and hard to discredit as useful information.

However, VAR has its limits. For one thing, it ignores the “fat tail” risks, i.e., how securities would perform under stressful conditions. For another, VAR only looks at the next 24 hours. Risk managers sometimes need to look at exposures to distressing trends. Markets develop their own momentum. Bad price trends can persist. Liquidity squeezes can intensify price declines. Smart risk management needs to spot bad trends, cut positions and take losses as early as possible.

Bearing this in mind, the best risk management firms didn’t content themselves with daily VAR. Their risk managers subject firmwide positions to stress tests and scenarios. Often they test for the potential “Black Swan” events—those occurrences which have no place in historical statistics, but which might comprise a new event that could upend carefully calibrated trading positions. Some remembered the lessons of Long-Term Capital Management (LTCM). A giant hedge fund founded by several Nobel prize-winning economists, LTCM was brought to its knees by unexpected events.

The LTCM story reminded those inclined to pay attention that three other controls are foundational to risk management. Firms need to keep an adequate capital base relative to their aggregate trading positions. This is known as maintaining a sound leverage ratio. Commercial banks typically limited leverage to about 11:1, i.e., $8.5 of capital behind every $100 of assets. Prudent investment banks might double that ratio. Second, firms need to be careful about mismatching the duration of assets and liabilities. Borrowing “overnight” to buy 30-year bonds involves risk. If financing cannot be rolled over, the firm’s only recourse will be to sell the long bond. Typically, firms experiencing such stress are quickly spotted by other traders and forced to liquidate assets at fire-sale prices.

Above all else, sound financial control requires top management support for the risk managers. Traders uniformly chafe under restraints. They also enjoy strong incentives to bet and win big in the next bonus cycle. Given half a chance, traders will run around or roll over most control structures. Only the knowledge that risk management has the support of top management will restrain their efforts to dismantle the constraints.

With this history in mind, Tim Ryan turned his attention to the control structure within AIG. His focus, in particular, settled on a part of AIG that was far from its core insurance business—AIG Financial Products (AIG-FP).

Management and Controls at AIG

Tim Ryan only ascended to the Global Relationship Partner role after the departure of AIG’s legendary leader, Maurice (Hank) Greenberg. He had heard the stories, however. Greenberg had built AIG from nothing into the dominant firm in the insurance industry. He also, it was said, carried AIG’s strategy, positions, and risk management around in his head.

Greenberg had taken a little known insurance firm founded in Shanghai and in a matter of decades transformed it into the indispensable player in the industry. Greenberg was innovative and relentless. Where other insurance firms were content to stay in their niche, Greenberg made AIG the cutting edge for new types of insurance. He pioneered writing cover for political risk. Greenberg also originated cover for the liabilities of corporate directors and officers.⁸ He pushed AIG into foreign markets, forming subsidiaries one after another to cover more and more jurisdictions. The firm became a giant in reinsurance, the practice of taking on part of the exposure written by a primary underwriter. Greenberg always maintained a keen eye for risk and sound underwriting. Important rewards followed. AIG acquired a coveted AAA credit rating. This meant a low cost of capital and fattened the margins on the coverage AIG wrote. By 2000, AIG was recognized as the uncontested leader of the insurance industry. Attachment 2 provides a summary of AIG’s financial results, 2000–2003.

Part of Greenberg’s success came from fostering diversification. Gradually AIG acquired a number of well-performing non-insurance businesses, e.g., aircraft leasing. In 1987, Greenberg launched another diversification, this one into financial products. AIG-FP, as it would soon be known, was originally led by Howard Sosin, a Drexel Burnham banker with a proprietary approach to trading interest rate and currency swaps.⁹ Sosin’s FP quickly made a lot of money; then some of its trades went sour. Greenberg forced Sosin’s departure in 1993, turning the business over to Thomas Savage and FP’s CFO Joseph Cassano. Both FP and Cassano would be heard from again.

Greenberg’s domination of AIG led to an idiosyncratic management system. On the one hand, AIG was immense and complicated. It had many business units operating in hundreds of jurisdictions using thousands of subsidiaries. The insurance business was also heavily regulated. Each jurisdiction had its own rules, specific protections designed to insure that policyholders got valid claims paid. All of this made AIG bureaucratic and legally driven within individual business units.

From another perspective, AIG’s management system best resembled a collection of silos. Almost all communication was “up/down.” Greenberg had little interest in training a successor and even less in encouraging peers or rivals to emerge. Consequently, he maintained nearly 30 direct reports.¹⁰ Business unit heads talked to Greenberg, but seldom to each other. Firmwide perspective was not encouraged or developed. Greenberg also chaired weekly risk committee meetings at which he personally reviewed exposures directly with business unit heads. He kept on top of all risks and gave frequent directions to adjust positions.

Greenberg’s management style had important consequences for the firm. To the outsider, AIG looked like a giant success story crafted by a brilliant, driven leader. Inside the firm, other issues were visible. Because he preferred to get information directly from business unit heads, Greenberg also saw little reason to invest in expensive management information systems. Consequently, AIG’s internal systems were seriously antiquated. Risk management was, thus, critically dependent on Greenberg’s weekly meetings and his own keen eyes.

Greenberg’s “indispensable man” style also led him to delay development of a serious succession plan. Right up to 2005, Greenberg made it clear he had no plans to leave. This became so apparent that two of his sons departed AIG to run lesser insurance firms.¹¹ All of this meant that AIG was carrying considerable “what if Greenberg gets hit by a bus?” risk.

The third issue was harder to spot—AIG’s parent holding company was, in fact, a hollow giant. It carried a AAA rating, but had few financial resources of its own. Rather, it depended upon distributions from its many subsidiaries. Since most of these were highly regulated insurance businesses, it was unclear how much internal cash AIG parent could muster in a crisis. For the present that risk was muted by AIG’s reports of growing profits and cash flow. Only someone like Hank Greenberg, with an intimate knowledge of the firm’s internal landscape, could project how the AIG parent might rally to meet a liquidity crisis.

A final controls issue took shape at AIG-FP and was late in developing. Starting around 2004, FP began writing CDS protection. When doing so, AIG-FP guaranteed a counterparty, usually a bank, against the risk of default on designated loans. AIG promised to compensate the bank for all losses above loan recovery value; in return AIG received annual fees during the life, typically 5 years, of the CDS. Banks liked to buy CDS protection from AIG; doing so brought “capital relief” that allowed them to make new loans. In AAA-rated AIG, they also had a counterparty that seemed sure to pay out loss reimbursements. AIG-FP also liked the business. Fees more than compensated for expected losses. In AIG-FP’s eyes, CDS was good business.

Joseph Cassano, now in charge at FP, saw possibilities to grow the CDS business. Wall Street firms brought AIG other opportunities to write credit protection on asset-backed securities and leveraged loans. Among these opportunities was the chance to write protection on subprime Residential Mortgage-Backed Securities (RMBS) and the Collateralized Debt Obligations (CDOs). Cassano and FP waded cautiously into the subprime CDS business. They decided to concentrate their activity in the AAA and “super-senior” tranches of such securities.¹² AIG-FP’s internal models calculated a 99.85% probability that the firm would never have to pay any claims on these CDS exposures.¹³ To Cassano, this made the subprime CDS business look like free money. In return for writing protection on a virtually risk-free security, AIG would get a stream of annual fees.¹⁴

Cassano thus decided that no loss reserves should be set aside for this business. At first, this looked like a great bet. FP’s combination of growing CDS fees and no loss reserves made for contributions to AIG’s bottom line that approached $1 billion.

As 2004 went into the books, events began to unfold that would expose AIG’s less than robust management system. These began when Hank Greenberg “got hit by a bus.” The same bus also ran over AIG’s auditor, Price Waterhouse, and brought Tim Ryan onto the account.

Greenberg Takes a Fall for AIG’s ‘Cooked Books’

Another thing about Hank Greenberg’s management style was that he liked to keep Wall Street happy by cranking out 15% annual profit growth. AIG did this with amazing regularity. When asked how this was possible, Greenberg would give his questioner a penetrating look and inquire: “What do you want? Do you want steady growth? Or do you want up 60 percent one quarter and down 15 percent the next?”¹⁵ Since few Wall Street analysts could penetrate deep into AIG’s labyrinth of subsidiaries, most assumed the firm’s performance was some combination of Greenberg’s genius with a bit of financial statement massaging to round off the edges.

The truth turned out to be more disturbing. Starting in 2004, New York Attorney General Eliot Spitzer began investigating the insurance industry. His inquiries soon turned up a series of AIG transactions having no purpose other than to affect reported financial results. In one instance, Greenberg had created a subsidiary to hold the AIG stock used to compensate top executives. Greenberg had this entity enter into transactions with AIG over 15 years that allowed profits to be shifted from one year to another. AIG had also entered into similar types of transactions with Berkshire Hathaway’s General Re firm—deals with no economic purpose other than generating flexible reserve provisions for AIG. Spitzer moved to prosecute AIG for securities law violations. In effect he charged AIG with misleading investors about the true nature of reported profits and equity.¹⁶

A special committee of AIG’s Board launched an investigation. PW was brought in to review the books. One AIG executive later recounted: “They were finding problems everywhere”—e.g., derivatives accounting, reinsurance transactions, all deals PW had once certified.¹⁷ PW eventually told the Board it could not certify AIG’s books if Greenberg stayed as CEO.

The AIG Board looked to settle. Spitzer told the directors that there could be no settlement without Greenberg stepping down. Faced with Spitzer’s ultimatum and PW’s verdict, the Board agreed. Greenberg stepped down in February, 2005. Martin Sullivan, a long-time insurance executive, succeeded him as CEO. AIG then paid a $1.6 billion fine to New York State. It also restated earnings for 2000–05, lowering them by $3.4 billion.¹⁸ More fallout came from the rating agency S&P. In May 2005 it downgraded AIG to AA+. S&P instituted the downgrade more in response to the accounting scandal and Greenberg’s departure than any particular decline in AIG’s financial fortunes.¹⁹

These events raised questions about AIG’s auditor, Price Waterhouse. Where were PWC’s auditors when Greenberg was concocting fictitious transactions? How could PWC have missed such behavior and certified AIG financial reports? AIG’s shareholders were outraged. Eventually the Ohio Public Employees Retirement System would sue PWC in Delaware’s Chancery Court.

In part to defend itself against such litigation, PWC had undertaken a huge investigation of AIG’s financial reporting. AIG-FP was included in their audit. PWC also changed up its account team, putting Tim Ryan into the Global Relationship role with Mike McColgan as the Engagement Partner.²⁰ Such efforts did not prevent the Ohio Public Employees’ suit against PWC from being expensive. Eventually, PWC paid out $97.5 million to dispose of the case.²¹ AIG did, however, continue to keep PWC in the audit role.

As these events unfolded, FP continued to pile up subprime CDS exposure. Some voices began to raise concerns. Gene Park and Andrew Forster spotted deteriorating underwriting standards in the mortgage market. Did this trend warrant any change in direction? Here is where AIG would miss Greenberg’s presence. Greenberg knew about FP’s CDS exposures.²² Sullivan, however, knew nothing of FP’s business and showed little interest in learning. Shortly after taking over, Sullivan cancelled Greenberg’s recurring risk management meetings. PWC was consumed with restating AIG’s books.

Cassano realized, and was glad, that FP would make the CDS exposure decision on its own.

AIG-FP Confronts a Subprime Market Decline

As 2005 neared a close, Cassano heeded Park’s and Forster’s warnings. He decided to cap AIG’s subprime exposure. FP told its counterparties it would no longer write CDS protection on subprime instruments. In calling a halt, Cassano capped AIG’s subprime CDS exposure at $110 billion.²³

Cassano did, however, ignore two risks embedded in this exposure. FP’s CDS contracts had what are known as “collateral triggers.” Most of these required AIG to deliver cash collateral to their CDS counterparty if: (1) AIG was downgraded below AA-; (2) the insured securities were downgraded; or (3) prices of the insured debt security fell below some designated level, e.g., 92% of par.²⁴ Cassano and FP considered all such events highly unlikely. AIG was still rated AA+; dropping below AA would require a three-stage downgrade. The securities being insured were also rated AAA, which should minimize any price volatility related to credit.

It was important that FP not face collateral demands because of a second risk. FP did not have many financial resources of its own. FP lacked operating businesses and hard assets. Moreover, if it needed support, FP could only look to the AIG parent; FP had no call on the capital retained within AIG’s insurance businesses. So long as AIG parent kept its AA, this would not be a problem. FP could have the parent borrow and funnel funds over to it.

Still, if AIG parent was something of a hollow giant, FP could only be considered more so. At the beginning of 2006, FP had a huge balance sheet by any measure. Embedded within that balance sheet were exposures that could produce sudden demands for large amounts of cash. For the moment, those contingencies seemed remote. Cassano never discussed them with Sullivan, and no attempt was made to quantify the potential demands for sudden liquidity that might appear. PWC was aware of the CDS collateral triggers, but took comfort for the moment in AIG’s credit rating and the high apparent quality of what FP had insured.²⁵

Events during the rest of 2006 and the first half of 2007 were about to shatter all sense of comfort.

FP Faces Collateral Calls on Subprime CDS

Throughout 2006 the subprime mortgage market displayed strong symptoms of decay. Underwriting standards hit new lows. “No doc” and negative amortization mortgages made up a higher percentage of vintage 2006 mortgages. Housing prices in hot markets peaked and began to decline. Mortgage delinquencies rose rapidly; especially concerning were the increasing number of “early payment defaults,” where mortgages defaulted after making only a few, or even no monthly payments.

Prices of subprime mortgage securities began to slip in sympathy with underlying fundamentals. Initially, RMBS and CDO prices fell only 1–2 points. Yet, this decline started to destabilize certain players. Two Bear Stearns hedge funds heavily invested in subprime securities reported their first loss in February 2007. By July, they had shut down entirely.

July also saw the rating agencies begin to downgrade scores of subprime securities. Their market prices again fell on the news. Most Wall Street firms continued to maintain price marks in the mid-90s. Those marks were theoretical; actual trades had virtually ceased. Noting this and sitting on a “big short” position, Goldman Sachs marked down its securities much more severely. Whereas Merrill Lynch might hold a subprime security’s value at 95, Goldman would insist its real value was 80 or even lower. When a counterparty protested, Goldman would offer to sell them securities priced at the level of Goldman’s marks. There were no takers.

During the first half of 2007, Goldman began to discuss possible collateral calls with FP. Whether this reflected genuine worry about AIG-FP’s ability to perform or a conscious strategy to cause its “big short” to prosper is unclear. What is clear is that in July, Goldman looked at its $20 billion in CDS contracts with AIG. Noting the collateral triggers and its own price marks, Goldman sent AIG-FP a demand for $1.8 billion in cash.²⁶ FP and Cassano were stunned. No other Street firm was carrying Goldman’s price marks or asking for cash. AIG resisted Goldman’s demands before reluctantly handing over $450 million as a good faith effort on August 10.²⁷ News of Goldman’s demands reached PWC and Ryan. For the moment, they accepted Cassano’s interpretation that this was just “Goldman being Goldman.”

Unbeknownst to Cassano, conversations were taking place between PWC and Goldman. PWC was also Goldman’s auditor. Goldman executives asked: “How could the same auditor certify accounts at two different clients based upon widely different values for the same securities.” Goldman’s PWC partner promised to take up the valuation question with headquarters.²⁸

Fall 2007 brought AIG no respite. The rating agencies continued to downgrade subprime mortgage securities. Trading in such instruments was almost nonexistent. Banks, however, still had to price their securities each day. With credit ratings in free fall, there was no avoiding lower carrying prices. Goldman continued to be the most aggressive in lowering its marks. Some of its prices were now down to 60% of par. This intensified Goldman’s calls for AIG to post collateral. By late November, other firms joined Goldman in requesting collateral. Total cash demands were now north of $6 billion.²⁹ AIG-FP’s capacity to meet such demands was questionable. FP argued, delayed, made partial payments, and ultimately turned to AIG’s parent for help.

Matters came to a head on November 29. Cassano was on a phone call with CEO Martin Sullivan. Tim Ryan was listening in. Cassano informed Sullivan that FP had paid out almost $2 billion in collateral to Goldman and others. It quickly became apparent that FP had made this decision without informing anyone at AIG’s parent. Sullivan then asked about the potential profit impact of declining market prices on FP’s CDS exposure. Cassano responded that if FP agreed to Goldman’s marks, a “worst case” scenario, the negative impact on AIG’s 4th quarter earnings could reach $5 billion. A startled Sullivan later commented that the news almost gave him a heart attack.³⁰

It was shortly after this phone call that Ryan informed Sullivan of the possibility that PWC might issue a finding of a “material controls weakness.”

Ryan and PWC Approach a Decision

Ryan’s warning initiated a process of intensified interaction between PWC and AIG-FP. In January 2008, AIG Chairman Robert Willumstad called PWC to a secret meeting, and asked it to review FP’s operations in detail. The findings were not comforting.³¹

For starters, FP’s CDS contracts were quite clear. AIG owed cash collateral if the insured securities’ prices fell below designated thresholds. By any measure, prices were underwater on a large number of insured securities. Cassano’s argument that the securities in question were likely to pay off over time was irrelevant. FP’s exposure to more collateral calls was also enormous. Subprime prices were likely to continue to fall. Marks at other Street firms would draw closer to Goldman’s. FP didn’t have any good estimates of the ultimate potential cash call, but PWC could see it reaching $10 billion or more. Where would FP, or for that matter, AIG find that kind of cash?

Ryan had also concluded that FP’s marks were going to have to come down for the CDS portfolio. PWC found FP’s valuation models to be grossly flawed. That Cassano believed the insured securities would ultimately pay out 100% again was irrelevant. Market uncertainties and illiquidity had dropped market prices to Goldman-like levels. Under mark-to-market accounting, that had to be reflected in reported profits. Cassano tried to argue in favor of an alternative approach, one Ryan told him “had no basis in accounting rules.” The net results of the portfolio revaluation would drop 2007 earnings by the $5 billion Cassano had earlier described as a “worst case.”³²

Elsewhere in AIG other storms were brewing. AIG had a wealth management business of considerable size. This business involved custody of stocks and bonds for clients. As a side business, AIG would lend such securities out to “short sellers” who would post some cash collateral as security. Sound practice called for the lending firm to invest such cash in secure, short-term instruments. These could be easily liquidated and the cash returned to the short seller when the lent securities were returned. With interest rates low, however, AIG had decided to invest a portion of these security deposits in subprime securities. Those securities were now worth considerably less than par. Moreover, they could not be easily sold. Indeed, trying to force a sale of any size would only send prices plummeting. AIG was thus exposed to billions more in calls for cash that it could not readily provide. Potential funding requirements ranged as high as $73 billion, depending upon when short sellers returned securities. As 2007 ended, AIG actually owed $6.7 billion in cash to investors who had already returned their borrowed securities.³³

Perhaps even more disturbing to Tim Ryan was the fact that AIG’s securities lending business made its decision to begin purchasing mortgage instruments at the same time that FP decided to cease writing CDS protection on similar securities.³⁴

As January merged into February, Ryan began to take stock of where AIG stood on managing its risks associated with subprime derivative instruments. Sullivan and colleagues, it seemed, were still struggling to get accurate information out of AIG-FP. Ryan added these observations to certain other facts, coming up with a list of possible justifications for a finding of a material controls weakness. The items on this list, in order of priority, were:

1. AIG FP lacked reliable valuation models for estimating prices on its huge derivatives portfolio, and thus could not reliably inform AIG parent as to its P/L performance.
2. FP had not established processes for aggregating subprime derivatives exposure, or for reporting such exposure to AIG parent.
3. FP had neither a process for informing its parent about potential collateral demands, nor any protocols for assuring that FP could raise funds to meet such liabilities.
4. AIG lacked a central risk management office that would monitor exposures in different parts of the firm. Such an office would have spotted and prevented the securities lending business from taking on risks which FP had ceased underwriting.
5. AIG parent had no guidelines for capital commitments and stop losses at FP—guidelines which might have prevented FP from growing its derivative exposures to the levels now responsible for massive collateral calls.
6. AIG FP had no guidelines for reserving capital in anticipation of losses on its derivative positions, losses which can easily result from being forced to sell securities at inopportune times to meet collateral calls.

Before concluding that AIG had a material weakness, Ryan weighed his list of concerns against the fact that most issues concerned AAA-rated securities. Given the high apparent quality of these instruments, extensive risk management processes may not have seemed necessary. Cassano was probably right that these securities would recover in the long run. AIG’s own high credit rating may also have implied that liquidity backup plans were not a high priority.

Then there was the question of what constituted a material weakness? Would better controls have altered AIG’s risks and protected its stock price? If shareholders had known more about AIG’s internal processes, would they have priced the stock differently? Ryan felt the answers to questions such as these touched on what would have been material.

Ryan did feel that he needed to consider AIG’s delicate financial condition. AIG was being squeezed severely by its counterparties. This was happening at the same time that capital markets were freezing up. Insurance regulations raised strong barriers to AIG transferring capital internally. Increasingly the firm was dependent upon short-term borrowing to meet the next collateral call. Would forcing a material weakness admission simply confirm counterparties in their belief that only cash collateral could assure that AIG would perform on its contracts? Would it convince lenders that AIG had no handle on its financing needs? Attachment 3 provides a summary of AIG’s financial condition as of December 31, 2007.

Inaction on PWC’s part also had its risks. Ryan remembered the firm’s being sued over AIG’s 2005 accounting scandal. AIG’s stock was now being battered. Eventually shareholders would go looking for a deep pocket to blame. Unless PWC performed its auditing role in impeccable fashion, it would rank high on their target list. Once another AIG accounting scandal hit the headlines, there was no telling where matters might go. Subsequent political pressures could even push the PCAOB into disciplinary action.

Ryan began to consider whether PWC had any options. One possibility would be to lead the Board Audit Committee into instituting major control revisions. These would include, at a minimum, new FP management, reconstituting a central risk management function, providing new information systems, and redesigning the information channels between FP and senior management. Loss reserves and liquidity backstop measures would also need attention. Ryan also began to consider how broad or narrow to make any “material weakness” finding. If such an announcement was made, should it come only after AIG had lined up additional financing?

Lastly, Ryan looked around at the conduct of the other big accounting firms. So far, none had made any “material weakness” findings at any big financial firm—this despite such revelations as Citibank’s stashing billions of subprime securities in off-balance-sheet SIVs, only to take them back later when “liquidity puts” were triggered. KPMG had said nothing at all about Citi’s controls, and continued to issue clean certifications. How different was that from AIG’s discovery of its collateral triggers? Did PWC want to be the first public accounting firm to force a material weakness finding, and then watch its client implode?

Ryan decided to sleep on his decision. He put his control concerns list in a desk drawer and headed out the door, wondering if sleep would be possible with so many questions to ponder.

Author’s Note

This is a dual case. On the one hand, it is about the adequacy of controls at AIG. The case examines the causes and potential consequences of AIG management’s stunning lack of understanding of its derivative risk exposures. Here again one can see the connection between weak controls and management blindness to its true financial condition.

Yet, in a deeper sense, the case is also about Price Waterhouse Coopers and the auditor/client relationship post-SOX. Sarbanes-Oxley was supposed to enable CPAs to stand up to their clients when there were important disclosures to be made. It also sought greater focus on good controls. Here we get to examine how this revamped relationship is working out. PWC must make a tough call on AIG.

Students of this case should stand in Tim Ryan’s shoes. Ultimately it will be his decision whether to force AIG to admit publicly to a material controls weakness. Ryan must consider this matter on the merits, and in the full light of PWC’s legal responsibilities to investors. He also knows that his client is wounded, and that a material weakness admission will render it even more vulnerable. Should AIG survive, they may later fire PWC for having taken a hard line. On the other hand, PWC has already been sued for missing an earlier AIG accounting scandal. PWC is likely to pay out tens of millions of dollars to settle that suit. Thus, the case involves a familiar, if more intense, set of auditor issues. Audit and business concerns pull in different directions. What takes precedence? Once that’s decided, how best does one manage the remaining business, client and legal complications?

This case draws heavily on All the Devils Are Here’s account of the AIG’s entry into credit default swap writing and Hank Greenberg’s management of the company. McLean and Nocera present a quite favorable portrait of Tim Ryan. Additional facts were provided by William D. Cohan’s book on Goldman Sachs. Goldman stood on the other side of many AIG swaps and was the leading aggressor in making collateral calls. Important balance was provided by Francine McKenna, a former PWC auditor. Now the author of a blog, re: The Auditors, she correctly pointed out PWC’s failure to catch Greenberg’s earlier accounting fraud and its delay in confronting AIG senior management about the situation at AIG-FP. McKenna has documented that PWC knew of FP’s exposures and Goldman’s collateral calls during the summer of 2007. Yet, it only began to force the issue with management in November. What might more timely and assertive action by PWC avoided in the longer run?

The list of weaknesses in controls at the end of the case is not based on a historical document. Rather, it reflects a list which Ryan might have made based upon the control problems PWC uncovered in its special audit, and which subsequently became public information. Ryan probably did make such a list and if anything, it had more items on it.

PWC and Tim Ryan appear to have been more proactive than other auditors during the financial crisis. As students read the cases about Citigroup, they should also be wondering what KPMG was doing in its audits. The other CPA firms fared no better. Ernst & Young’s performance, prominently on display in the Lehman Brothers case, is especially noteworthy for its technical defensiveness and blindness towards fundamentals. Whatever improvements SOX may have achieved in corporate reporting, the auditor/client relationship still seems an area in need of further reform.