Citibank’s Subprime Product Flow and its SIVs

By 2006 Citibank had constructed a formidable mortgage security machine. Citi had developed an integrated mortgage business model. Consumer Lending (CL) employed an extensive U.S. branch network to originate mortgage loans. A second CL unit purchased mortgages from brokers and other lenders. A third unit purchased pools of mortgages that had already been gathered for the purpose of using them to back new security issuance.

Citi next took all of these mortgages and pools and packaged them into tranches that would collateralize new RMBS. This required some technical skill. The mortgages backing a security would have to generate flows which in aggregate would resemble normal bond payments. The tranches would also have to be composed in such a way that the highest ranking tiers, dubbed “super-senior” and “senior,” would be regarded by the rating agencies as virtually immune from losses. Once accomplished, the super-senior tiers were regarded as so remote from losses that Citi did not even submit them to the ratings process.⁴ The two senior tranches typically constituted 65–80% of a subprime mortgage pool.

After taking the various tranches through the rating agencies, Citi would sell RMBS to institutional and foreign investors. U.S. interest rates were low during this period. Citi’s RMBS offered premiums of 1–2% over comparable U.S. Treasury securities, and the bank had little trouble selling off the higher investment grade securities.

However, the lowest ranking tranches usually remained on Citi’s books. In order to make way for new mortgage inflows, Citi reapplied its structuring technique. The bank constructed pools of equity and mezzanine tranches, using them to back new securities. It then tiered these into senior and subordinate securities, known as Collateralized Debt Obligations. The more senior of these CDOs would be rated and sold to investors. Citi expanded this business rapidly; from 2003–05, Citi’s annual underwritings of CDOs grew from $3 billion to $20 billion.⁵ The RMBS/CDO business model proved hugely profitable, as Citibank’s 2001–05 results indicated (Attachment 3).

This left Citi with a variety of subprime exposures on its books. For starters it had subprime mortgages, pools and purchased RMBS that had not yet been packaged into new RMBS or CDOs. These assets were regarded as residing in a “warehouse” awaiting sale. Citi also retained certain equity and mezzanine tranches; these were either awaiting further structuring or represented a Citi decision that it was more economic to retain than to sell them. Finally, there were the super-senior CDOs; these were regarded as so remote from loss that typically they were offered at very small premiums over U.S. Treasury securities. Buyers were sometimes scarce and Citi often considered it more economic to retain the super-seniors.

All of this subprime exposure created a problem—it tied up a lot of Citibank’s Tier One Capital. This was especially the case with the equity and mezzanine tranches, whose risk ratings required high capital allocations. Citi harvested fees at many points along the mortgage pipeline. Keeping this pipeline flowing was regarded as the key to rising profits. However, as Citibank’s subprime business grew, the capital demanded to support this business was also growing—and at a rate that was beginning to curtail both the mortgage business and other business lines.

To unblock this pipeline, Citi landed on the idea of developing Structured Investment Vehicles. Starting in 2003, a sizeable portion of Citi’s subprime mortgage assets was directed into these SIVs.

Citibank Structures and Launches Subprime SIVs

Structured Investment Vehicles were introduced by Citibank in 1988, and since used for a variety of purposes. Citi’s growing mortgage business suggested another application. The idea was to construct a family of SIVs to act as dedicated buyers of Citi’s RMBS/CDOs; this held three attractions. First, it created a ready-made destination for Citi’s mortgage securities factory. Second, it promised to expand Citi’s investor base for these products. Finally, the SIVs promised to ameliorate the Citi capital constraints. In sum, SIVs as final destinations for subprime securities promised to be hugely value creating.

However, Citibank discovered there were problems to overcome before these benefits could be harvested. The problems involved issues which Citi would have to address for the SIVs to work as intended.

The basic SIV structure was crafted as follows:

1. Citi would set up a separate legal entity;
2. the bank would sell a small amount of venture equity to third parties, thus creating independent ownership;
3. the SIV would then buy subprime assets, predominately CDOs, from Citi;
4. the SIV would sell asset-backed commercial paper (CP) to investors, using the proceeds to pay Citi for the subprime assets; and
5. Citi would contract to manage the SIV going forward.

A core problem here was the mismatch between SIV assets and funding. The SIVs would own assets whose maturities stretched out to 30 years while its funding consisted of 30–60 day CP. Given the SIVs minimal equity, any failure to rollover CP would present the vehicles with a funding crisis. Either the SIVs would have to sell long-term assets, perhaps at marked-down prices, or fail to honor a commercial paper redemption.

This problem was particularly acute because Citi was planning something ambitious; it was going to target sales of SIV commercial paper to Money Market Funds (MMFs). Accomplishing this would greatly expand the pool of subprime investors. MMFs are the brokerage house equivalent of bank checking accounts, and their depositors expect their funds to be both safe and immediately available. Typically, they would never invest in something as illiquid as subprime mortgage instruments. SIV CP appeared more short term and therefore more liquid, but this was true only to the extent that investors remained confident that the subprime assets backing the CP would perform as expected.

Citi’s solutions needed to address both SIV solvency and liquidity. The bank felt that by stocking the vehicles with only highly rated CDOs, SIV solvency would be satisfied. As for liquidity, Citi decided to address the CP rollover risk by issuing “liquidity puts.” In the face of a financing crunch, the SIVs could “put” their CP to Citi. The bank would then own SIV CP backed by the vehicles’ “super-senior” CDO tranches.⁶

The liquidity puts satisfied the MMF’s concerns, and they became big buyers of SIV CP. However, the put instruments raised fundamental questions about whether the SIVs were truly independent from Citibank. If they were deemed not to be, Citi would have to incorporate the vehicles into its consolidated financial reports. Doing so would prevent the bank from harvesting the third SIV benefit—that of ameliorating pressures on the bank’s capital position.

Whether this goal could also be realized depended on how one regarded Citi’s liquidity puts in light of FASB ASC 810–10, the rules on “Variable Interest Entities” (VIE).

Citibank’s SIVs Finesse the VIE Rules

The FASB “VIE” rules were put in place in response to Enron’s collapse. Enron had made extensive use of thinly capitalized partnerships that purported to be independent. These entities received the bulk of their financing from Enron, traded only with Enron, and were usually managed by an Enron employee. Despite this, Enron treated the partnerships as independent for accounting purposes, and never consolidated their assets or debts. Trades between Enron and these vehicles were done to manufacture earnings which Enron used to sustain the impression that it was a well-oiled profit machine. Investigations subsequent to Enron’s bankruptcy described the vehicles as having no economic purpose other than to help Enron manage its reported earnings.⁷

In response to this accounting fiasco, the FASB first issued VIE rules in 2002. The essence of the rules was to establish when firms would need to consolidate the financials of an entity in which they held neither a majority of the voting stock or of owner’s votes. VIE rules sought to determine “effective control.” For example, if a company contracted with another entity to absorb losses which that entity might incur, the VIE rules required the loss-absorbing company to consolidate the entity. Other examples focused on weak or nonexistent vehicle equity, including:

1. The equity-at-risk is not sufficient to support the entity’s activities (e.g., the entity is thinly capitalized, the group of equity holders possess no substantive voting rights, etc.);
2. As a group, the equity-at-risk holders cannot control the entity; or
3. The economics do not coincide with the voting interests (commonly known as the “anti-abuse rule”).⁸

Like other FASB decisions, the VIE rules were disputed by the affected industry and eventually revised. The final version is summarized in Attachment 4.

Citibank and its accommodating auditor KPMG decided its SIVs did not trigger the VIE rules. Certain rationales supported this decision. Citi did not own the vehicle equity and did not enter into contracts to absorb vehicle losses. So long as the subprime assets performed and financing markets remained liquid, the vehicles’ economics should meet investor expectations. Since the SIVs’ assets were all high investment grade securities, it was reasonable to expect them to perform reliably. As manager, Citibank performed specified contractual roles that did not imply ownership or control. It could, for example, purchase or sell individual CDOs without informing the SIVs’ Board. It could not, however, decide to have the vehicle enter a whole new line of business, e.g., purchasing real estate.

That left the liquidity puts as the principal SIV link back to Citibank. The bank argued and KPMG agreed that the puts were not different than “backup” credit lines which banks routinely provide to support client CP programs.

There was, however, another way in which the liquidity puts could end up damaging Citibank. CDO securities also involve liquidity risk. Even if a particular CDO is performing as expected, it is possible that during its life market conditions could cause buyer demand to diminish or even evaporate. Should that happen, the CDO’s price will drop below par value. A financial institution holding that CDO would then have to record a loss. Should major events spook financial markets, e.g., unexpectedly high mortgage default rates and reports of fraud, liquidity could dry up all together. Price drops and mark-to-market losses on CDOs could then be huge.

KPMG, however, decided this risk was remote. It also was aware that the Basel agreements stipulating allocation of bank capital against specific types of assets included a very pertinent carve out. The Basel rules exempted credit lines extended to support commercial paper programs so long as their tenors were for less than a year.

Taking all this into account, KPMG agreed with Citi’s decision to treat the SIVs as non-consolidated entities. Soon Citi had launched 7 such entities, with names like Beta Finance Corp., Centauri Corp., Dorada Corp., Five Finance Corp., Sedna Finance Corp., Vetra Finance Corp. and Zela Finance Corp. By mid-2007 these entities held over $80 billion in assets, most of which involved subprime CDOs.⁹

Neither these assets nor any associated gains and losses were being reported in Citibank’s quarterly or annual financial disclosures.

Conditions Worsen in the Mortgage and RMBS/CDO Markets

Even as Citibank expanded its subprime mortgage business and its family of SIVs, America’s mortgage market began to crack. Key conditions underpinning the values of subprime RMBS/COOs were eroding. Mortgage lending and underwriting practices also deteriorated severely.

By the end of 2004 the U.S. mortgage market had reached a point where risk levels were rapidly rising. For starters, a record 69% of American households now owned their home. Of these, some $2 trillion was supported by subprime and Alt-A mortgages, i.e., loans which would not have qualified for purchase by Fannie Mae or Freddie Mac before 2000. The portion of new mortgages represented by these risky loans had risen to one third. Two thirds of subprime loans now had adjustable rates, meaning borrowers were exposed to both rising interest rates and declining home prices. As for new prime mortgages, 50% now involved second mortgages, either for withdrawing home equity or speculating in real estate.¹⁰

With the mortgage market reaching saturation, financial institutions all along the mortgage pipeline pressed harder to maintain thruput. Underwriting standards decayed. Reports began to surface of predatory lending, i.e., brokers and originators tempting borrowers with “teaser” rates while not disclosing the fees and rate adjustments that doomed the mortgage to default.

By 4Q 2006, the consequences of these practices began to appear. Subprime mortgage delinquencies were up sharply, reaching 7.75% nationwide. By October, 3% of all subprime mortgages written since the beginning of the year were delinquent. This meant that buyers were defaulting after making one or in some cases no payments. Reports began to circulate of “owners” abandoning houses in which they never lived. As delinquencies soared, house prices declined sharply in former hot markets like Las Vegas. Nationwide, 3Q housing prices declined, the first such drop in 13 years.¹¹

Conditions didn’t improve in 2007. In February two Bear Stearns hedge funds, both heavily invested in subprime CDOs, reported declines in net asset values. By June the Bear Stearns funds were widely seen to be in serious difficulty. Investors ran for the exits. Bear Stearns’ broker/dealer pumped money in to fund the redemptions. Then in July, Bear decided the funds could not be saved. Further redemptions were denied and the funds went into liquidation.

It was against this backdrop that Citibank began to consider what to disclose about subprime exposure when it reported 2Q results.

Citibank Reports Second Quarter Results

On July 20 Citibank reported record second quarter profits and immediately held an earnings call for investors and analysts (see Attachment 5 for summary results and a schematic of Citi’s business lines).

CEO Charles Prince opened the call by highlighting Citi’s strong performance during the quarter:

“I want to begin by thanking everybody on our team for delivering a record quarterly performance, the best we have ever done. I am very pleased with the underlying health and quality of our business.

I am especially pleased in this quarter with a couple of particular items. One, the growth in our international franchise; second, the performance in our capital markets related businesses, generally, especially given the market conditions we all saw in June … Credit was a drag on our bottom line results, and Gary will talk about that in more detail as he talks about credit, but I would tell you I am pleased with the continued turn we see in our U.S. consumer business.”¹²

Crittenden then took over the call, providing a more detailed account of Citi’s performance. Turning eventually to the subject of credit losses/reserves, he disclosed the following:

“The cost of credit was up 50%, and as Chuck said, I will come back and describe that in some detail in a few minutes. Despite the increase in credit costs, pretax income was up 19%, net income from continuing operations was up 18% and EPS increased by 18%. Our return on equity was 20.1% in the quarter.”¹³

Significantly, Crittenden was telling the market that Citi had incurred huge credit losses but still generated record 2Q earnings. Then after talking about consumer loans, Citi’s international business, Japan and several other items, Gary turned to the bank’s subprime exposure:

“Now, let me spend a minute talking about two topics, the subprime secured lending market and our leveraged lending activities. Our subprime exposure in Markets and Banking can be divided into two categories, which together account for 2% of the Securities and Banking revenues in 2006. The first is secured lending and the second is trading.

With regards to secured lending, we have been actively managing down our exposure for some time. We had $24 billion in assets at the end of 2006. It was at $20 billion at the end of the first quarter and $13 billion at the end of the second quarter, while adjusting at the same time collateral and margin requirements.”¹⁴

Reflecting on events during 3Q 2007, Gary wondered whether these remarks had rendered both the bank and himself personally liable as regards improperly disclosing the bank’s subprime exposure (Attachment 6 provides a summary of disclosure standards for publicly traded companies). Clearly the sense of his remarks minimized the importance of Citi’s exposure. The word subprime appeared only twice before giving way to the euphemism “secured lending.” The $13 billion figure clearly implied that Citi’s exposure was trending down. There was no inclusion or discussion of the super-senior tranches and liquidity puts. Crittenden felt that these had been excluded for good reasons—high credit quality in the former case and reasonable analogy to backup CP lines in the latter. However, 3Q developments were making those assumptions look tenuous. Would his July remarks now look bad in retrospect?

Third Quarter Events Hammer Citi’s Results

Third quarter 2007 got off to a rocky start with the aforementioned failure of the Bear Stearns funds. This was followed by Countrywide, the nation’s leading mortgage lender, filing 2Q results. Twenty percent of Countrywide’s subprime mortgages were at least one month delinquent. Some 10% were now 3 months delinquent.¹⁵ Responding to this and other developments, Standard & Poor’s began to downgrade selected vintages of subprime RMBS.

On August 6, American Home Mortgage, the nation’s tenth largest mortgage lender, filed for bankruptcy. On August 9, BNP Paribas, the largest French bank, closed three funds heavily invested in subprime CDOs. As the reason for its action, BNP cited “a complete evaporation of liquidity.”¹⁶ Interbank lending rates began to creep higher—a sign that banks were increasingly nervous about undisclosed RMBS and CDO exposure at other institutions. Trading in mortgage securities almost completely ceased.

As August turned to September, Moody’s joined S&P in downgrading more subprime RMBS and CDOs. Ratings ceased for new issuance. Then, Northern Rock, one of the U.K.’s largest mortgage banks, required a bailout by the Bank of England.

With secondary trading halted, banks were forced to recognize further markdowns of their subprime assets’ carrying value. Goldman and J.P. Morgan aggressively marked down prices, putting competitive pressure on their rivals. By any standard, 3Q financial reports for the Wall Street firms were going to be ugly.

These developments also prevented Citi’s SIVs from rolling over expiring commercial paper. MMF managers panicked at the prospect of losses on SIV CP. This might mean they could not protect the principal invested with them. Such an outcome, known as “breaking the buck,” had never before happened. Were it to occur now, MMFs might lose their reputation as a safe cash depository.

As MMFs fled SIV CP, Citi was contractually obligated to honor its liquidity puts. By the end of 3Q, Citi had purchased some $25 billion in CP from its SIV family. Since the only source of repayment for this CP was the super-senior CDO securities, Citibank was now directly exposed to their performance.

Considering Citi’s 3Q Results and IR’s Proposed Pre-Announcement

As Crittenden pondered the text from IR, he also considered another report provided by his financial reporting staff. Prominent in this report was a line item showing that Citi’s securities and banking business had incurred pre-tax losses totaling $1.3 billion. Of this total, $300 million related to leveraged loans “warehoused” in anticipation of future securitization. The remaining $1 billion derived from Citi’s subprime exposures. Some $100 million of these losses pertained to Citi’s retained super-senior tranches.¹⁷

Returning to the IR text, Crittenden noted that nowhere did it explicitly mention Citi’s super-senior tranches. The closest it came was the veiled reference that Citi “held onto most of the highest rated tranches.” From there the text seemed to imply that these were included in the $13 billion exposure figure. There was also no mention of Citi’s liquidity puts, or of the $25 billion in SIV subprime asset-backed CP that it now owned. Should the SIVs’ super-senior CDOs fail to perform, Citi would end up taking losses on the CP it now held. Finally, there was nothing in the text to indicate that Citi might need to provide further support to avoid SIV defaults on remaining CP.

When IR was asked whether it was misleading not to mention these exposures, it responded that because they had not been mentioned in previous press statements, they could not be introduced now. Two senior investment bank executives supported IRs position.¹⁸

Crittenden understood the delicacy of the situation. Senior executives were concerned about surprising the market. Interbank lending was becoming more expensive. Mishandling Citi’s pre-announcement could raise funding costs, drop the stock price, and even cause a “run on the bank.”

Gary was also aware that efforts were underway to organize an SIV rescue. He didn’t want a poorly crafted disclosure to scupper the effort. Citi was not the only bank on Wall Street that employed SIVs. The idea currently being floated, with support from Treasury Secretary Hank Paulson, would see all major banks establish a lending pool to purchase SIV CP. The idea was that if the pool purchased CP from every SIV, no individual bank could be seen to be primarily supporting a given SIV. Reconsolidation of the SIVs under the VIE rules might then be avoided. Crittenden liked the idea. However, it was far from a sure shot solution. Gary had heard that more conservative banks, notably J.P. Morgan and Bank of New York, did not employ SIVs and were not inclined to participate in any rescue fund.

Finally, Crittenden recalled the reasons why the super-senior tranches, liquidity puts, and Citi’s relationships with the SIVs had been excluded from prior reporting. The view on super-seniors had been that what constituted “materiality” for subprime was the idea of high risk, i.e., above average exposure to loss. That was what the market wanted to know. The super-senior CDO tranches were so well regarded that they ranked above AAA CDO securities. For this reason, they had not been regarded as material to a discussion of Citi’s subprime exposure. Unfortunately, they now had produced $100 million in 3Q losses. As sound as the original argument might have been, current market conditions had undermined the premise that super-seniors were not relevant to Citi’s exposure discussion.

The liquidity put rationale also had been undermined. The analogy to back-up CP lines failed to take into account that with the SIVs the CP would be “asset-backed”— and the assets doing the backing would be super-senior subprime CDOs. Now Citi’s liquidity puts had been transformed into $25 billion of CP backed by those CDOs. As the SIVs’ CP investors exited, the exercise of Citi’s liquidity puts was making the bank look like it was primarily responsible for supporting the entities. Bank market conditions had triggered the liquidity puts, and Citi had been obliged to “step up.”

Reviewing IR’s text, Crittenden felt trapped in a vice—caught between past disclosure history, a fear of spooking nervous markets, and his rising concern that he might face an SEC lawsuit. Was there a way to modify what IR had written without seeming to cast prior disclosures as material omissions? Could this be done in such a way that the markets would not savagely punish Citi, cut off its funding and turn cold on the SIV rescue efforts?

Gary Crittenden turned on his desktop computer, brought up the IR text, and began to draft a revised pre-announcement. If he could devise something that met the challenges just noted, he’d take a run at convincing CEO Prince to consider his version in place of the IR text.

Author’s Note

This case focuses on an issue that formed part of Enron’s legacy: how should “off-balance sheet” vehicles be used and accounted for? In theory, this issue was addressed by the Financial Accounting Standards Board in their rule FASB ASC 810-10: “Variable Interest Entities. The Citigroup SIV case suggests that the disclosure problems associated with such vehicles persist.

More specifically, the case discusses the dilemma of “creativity” associated with pushing the limits on accounting rules. Citi was an innovator in the use of SIVs. By 2007 it had been deploying them for 20 years. Using SIVs allowed Citi to relieve capital constraints, make more loans, book more fees and report a higher Return on Equity. For many years this must have seemed a smart practice. Rules, such as ASC 810-10, were successfully wired around, and the architects of such “financial engineering” were duly rewarded. However, the case depicts what happens when markets turn. It is at that point that accounting judgments rendered in calm markets start to look problematic. Just when credit is most precious, the requirements for revised disclosures bite hardest, and the weight of past disclosure decisions hangs heavy on the responsible management.

Gary Crittenden is one of those managers. As Citigroup’s CFO, he is responsible for the firm’s SEC filings. He’s also legally liable for material misstatements and omissions under securities law, including SOX.

Crittenden’s challenges are threefold. He must decide if the Investor Relations draft statement is factually accurate and complete. If not, he must consider revisions. Second, any revisions must be reconciled with Citi’s past disclosures, or else past errors must be admitted. Third, the firm must consider how to present its adverse news in a manner that doesn’t exacerbate its financing problems. For Crittenden, this may require action plans that go beyond rewording IR’s disclosure statement.

This case is based largely on the SEC’s “Cease and Desist” order directed at Crittenden and IR’s Arthur Tildesley. Supplemental information comes from contemporaneous reports in the financial press, especially Bloomberg’s coverage, and from investor litigation against Citigroup filed in the U.S. Southern District. The draft IR statement is based upon the actual press release issued by Citigroup on October 1, 2007. There is no public record of Crittenden questioning or revising an earlier IR draft. The case imagines him doing so in order to provide students with an opportunity to reconsider his dilemmas and devise alternate solutions. Crittenden resigned along with CEO Prince one month later.

The case raises important questions about Citi’s longtime auditor, KPMG. That firm is conspicuous by its absence in accounts of these disclosure issues. Missing too is Citi’s outside counsel, who must have been advising Prince and Crittenden on securities law requirements. Their failures here are noteworthy examples that the Enron “gatekeeper” problem has not been fixed.

Finally, this case should be considered in the light of the two other Citigroup cases presented in this volume. Collectively these present a disheartening portrait of serial violations and weak governance. Citigroup seems to tarnish the reputations of even the most formidable of financial leaders, from John Reed to Sandy Weill to Robert Rubin. The firm’s enduring inability to reform itself raises serious questions of whether Citi is not only too big to fail, but too big to manage.