The Citigroup Rule

My previous post looked at one of the two major issues in Basel III’s all-important liquidity requirements that still have to be determined by national regulators. This post will look at the other major issue that was left to national regulators, which I’m trying to get people to call “the Citigroup rule.” (Do it, it’s cool.)



To recap, the main component of Basel III’s liquidity requirements, the Liquidity Coverage Ratio (LCR), requires internationally active banks to maintain a stock of “high-quality liquid assets” that’s sufficient to cover cash outflows in a 30-day stress scenario. The cash outflows in the stress scenario are calculated by applying “run-off rates” to each source of funding (e.g., unsecured wholesale funding, repos, etc.). Most of the action/controversy here is in the different run-off rates used.



Under the heading, “Other contingent funding obligations,” the final Basel III document says: (emphasis mine)

101. These contingent funding obligations may be either contractual or non-contractual and are not lending commitments. Non-contractual contingent funding obligations include associations with, or sponsorship of, products sold or services provided that may require the support or extension of funds in the future under stressed conditions. Non-contractual obligations may be embedded in financial products and instruments sold, sponsored, or originated by the institution that can give rise to unplanned balance sheet growth arising from support given for reputational risk considerations. These include products and instruments for which the customer or holder has specific expectations regarding the liquidity and marketability of the product or instrument and for which failure to satisfy customer expectations in a commercially reasonable manner would likely cause material reputational damage to the institution or otherwise impair ongoing viability.

Did you catch that? Banks have to set aside high-quality liquid assets to cover liquidity support for clients that they’re not technically required to provide, but likely would anyway. Some people might look at that and immediately say it’s ridiculous, because how can a regulator predict what a bank will do to protect its reputation in a crisis? Or what a bank considers “material reputational damage”? But it’s not actually ridiculous. Moreover, to be truly effective, this is something the LCR absolutely has to grapple with.



The reason I call this “the Citigroup rule” is that this is exactly what happened to Citi when it bailed out its SIVs in 2007-2008. (Yes, other banks like HSBC and BofA rescued their SIVs too, but none on nearly the same scale — Citi’s decision to rescue its SIVs resulted in nearly $100bn of the most toxic assets being brought back onto Citi’s balance sheet, and was arguably the bank’s fatal mistake.)



Citi had been the largest sponsor of structured investment vehicles (SIVs), which invested in long-term assets, including MBS and CDOs, and funded themselves by selling short-dated asset-backed commercial paper (ABCP) and medium-term notes. Citi structured, underwrote, and then “advised” its SIVs. The SIVs’ senior debt was primarily sold to Citi’s institutional clients, while money-market mutual funds were the main buyers of the short-term ABCP. When the ABCP market imploded in the Fall of 2007 due to subprime concerns, the SIVs were forced to sell assets into an illiquid market in order to fund themselves, leading to large losses on their MBS holdings.



Citi’s institutional clients (i.e., the senior debt holders) were, shall we say, “not pleased.” Citi had structured, underwritten, and were now managing these SIVs, which had been pitched as ultra-safe investments, and the institutional clients damn-well expected Citi to help the SIVs out. But according to its 2007 10-K, Citi “was not contractually obligated to provide liquidity facilities or guarantees to the SIVs.”



However, had Citi simply hung the senior debt holders out to dry and refused to provide any support to the SIVs, they almost certainly would’ve lost a good deal of business from those institutional clients, and suffered a big reputational hit.



So what did Citi do? Naturally, it chose to rescue its SIVs — providing liquidity facilities for five of its seven SIVs, which had $58bn in total assets — rather than take a big reputational hit. Per a December 13, 2007 press release: (emphasis mine)

Citi announced today that it has committed to provide a support facility that will resolve uncertainties regarding senior debt repayment currently facing the Citi-advised Structured Investment Vehicles (“SIVs”).



This action is a response to the recently announced ratings review for possible downgrade by Moody’s and S&P of the outstanding senior debt of the SIVs, and the continued reduction of liquidity in the SIV related asset-backed commercial paper and medium-term note markets. These markets are the traditional funding sources for the SIVs. Citi’s actions today are designed to support the current ratings of the SIVs’ senior debt and to allow the SIVs to continue to pursue their current orderly asset reduction plan. As a result of this commitment, Citi will consolidate the SIVs’ assets and liabilities onto its balance sheet under applicable accounting rules.

And this wasn’t limited to SIVs: Citi later rescued four of its hedge funds (including the Vikram Pandit-run Old Lane Partners), none of which Citi had been under any explicit contractual obligation to support. The hedge funds had a combined $15bn in assets, which Citi had to eventually bring back on its balance sheet.



There’s no doubt that this was a big, sudden drain on Citi’s liquidity. The SIVs alone probably issued between $10–$15 billion of commercial paper per month, which Citi had suddenly committed itself to buying on an ongoing basis. And then there were the SIVs’ medium-term notes, which Citi also had to roll over.



This is the kind of situation that “the Citigroup rule” is designed to protect against. The cash outflows are reasonably foreseeable — everyone knew that Citi was eventually going to rescue its SIVs — but not contractually obligated. Forcing banks to account for these cash outflows its their liquidity pools is critical because in a real crisis, banks really will end up providing this kind of liquidity support (no matter how vehemently they insist otherwise). Another benefit is that the rule will help prevent the abuse of off-balance-sheet accounting, where banks move assets into facilities that are technically off-balance-sheet, but that everyone knows will come back on the banks’ balance sheets if things go sour.



Obviously, what qualifies as a “non-contractual funding obligation” will have to be determined on a case-by-case basis, and will necessarily be at a regulator’s discretion. But determining when a bank is likely to choose to provide liquidity support shouldn’t be as tricky as it sounds. Think about it: for the bank’s failure to provide liquidity support to cause “material reputational damage,” it would have to be obvious to pretty much everyone that the bank should have provided liquidity support — because if it’s not obvious to pretty much everyone, then would it really cause material reputational damage?



As to the run-off rate for these “non-contractual funding obligations,” I’d advocate a very high run-off rate. This kind of non-contractual liquidity support is, almost by definition, required precisely when the market is least liquid (and thus liquidity is the most expensive for the bank). After all, if the market was still liquid, then there would be no need for the bank to provide the liquidity support in the first place. Also, if the regulator determines that the bank would step in and buy an entity’s commercial paper, or would buy back a bond that it just underwrote, then there’s no reason to think that the bank would only buy some of the commercial paper, or repurchase some of the bond issue. If the bank is providing liquidity support specifically to maintain its reputation, then it’s only natural to assume that the bank will go the whole 9 yards.