A Particularly Choice Citigroup Disclosure

Various analysts and reporters took keen interest in Citigroup SEC filing Tuesday that revealed that the bank had deployed $7.6 billion of a total $10 billion liquidity facility to assist its floundering structured investment vehciles (SIVs), but stated it does not believe it has to consolidate the SIVs.

CreditSights estimated that Citi’s losses on asset backed bonds could be as high at $13.7 billion, arguing that it needs to write off an $2.7 billion of subprime securities (the bank had announced its writedowns would be $8 to $11 billion this weekend.

However, Bloomberg mentioned another disclosure in the filing:

The New York-based bank also said in the SEC filing that the amount of securities it owns that are considered hardest to value, known as Level 3 assets, rose 42 percent in the third quarter to $135 billion.

Why is God’s name did the bank add $40 billion in arcane assets when the credit markets were in bad shape? This doesn’t simply make no sense; it was reckless verging on suicidal (unless the increase in Level 3 assets is accounting hocus pocus to avoid taking losses on former Level 2 assets by reclassifying them as Level 3 and applying different valuation methodologies, but our layperson’s understanding was that FASB 159 has restricted that pretty severely).

Merrill Lynch was doing its best to exit CDO and subprime positions and succeeded in reducing them by nearly half. Why was Cit ballooning its holdings of similar assets right after the Bear Stearns hedge fund meltdown revealed how unsalable even smaller amounts (mere single digit billions) of mortgage paper were? And you can’t blame this on not-fully-sold underwritings, for that business had pretty much shut down by then.

Update, 11/7, 8:30 AM: A reader pointed us to FT Alphaville, which fills in a piece of the puzzle:

Most CDOs, of course, don’t normally issue CP. Traditionally, CDOs issue straight tranches of rated bonds. But Citi made CDOs issuing CP good business pre-crunch. You might say that commercial paper CDOs are a scion of the SIV world – using CP issuance to supplement their normal debt issues and create a more dynamic, flexible portfolio, benefiting from low yielding CP.

But also just like SIVs, CDOs need to keep refinancing that CP to pay off upcoming redemptions. But where they differ is that CP issuing CDOs mitigate that rollover risks by using their arranging banks’ as underwriters on all new CP issues. Whatever CP they can’t sell, agreements are in place as backup that ensure banks will buy.

And just like with SIV CP investors, over the summer, CDO CP buyers have dried up.

Money market funds – formerly among the biggest players in the CP markets – are loathe to touch anything containing MBS.
So Citi – unable to place CP on subprime CDOs it arranged as far back as 2005, – is having to buy it instead. Right when it can least afford to do so.

Crucially we should make clear that Citi isn’t necessarily being “forced” into buying that debt: not in the most literal sense of the word. The backstop “agreements” it has in place are not set in stone. It could have said no. But had it done so it may have seen CDOs default, or else a rush to sell assets to meet amortizing CP. In the event, that was evidently too ugly an option to countenance.

And Citi may only now be ruing that decision. Commercial paper is classed as “super senior” debt in CDOs, and had until October, held out as a secure tranche. But the contagion has spread right up the tree, and the rating agencies have shown now mercy for even the highest grades of debt. Super senior debt is far from secure.

This is a new revelation. While I have been endeavoring to keep abreast of information on CDOs, I haven’t seen this device before, that CDOs would use CP as part of their structure (query whether this would be “passive” CDOs, which have an identified pool of assets before they sell the tranches of debt against them, or “active,” which are essentially blind pools and trade the CDO assets).

Citi’s practice stinks to high heaven. The regular returns on creating and issuing CDOs wasn’t good enough, so they had to “improve” the structure by using commercial paper as a source of funding, but then putting themselves in the position of having to step to the plate if there was a problem. This may also have been a function of increased difficulty in selling AAA CDO tranches (the bulk of the value of the deal) in 2006 and 2007, when reportedly US buyers were becoming resistant (or had simply bought as much as they intended to purchase) and more paper was being placed overseas. However, Citi was apparently resorting to this structure as early as 2005, so the main motivation was apparently increasing the profits on each deal.

So when Citi’s CDOs started having trouble rolling their CP, Citi stepped up and purchased it instead. On a small scale, this might not have been a terrible idea, but to the tune of a $40 billion increase in one quarter? This was madness. And by the third quarter, when these purchases were taking place, anyone with an operating brain cell should have known that this supposed AAA paper was vulnerable to begin downgraded.

But this explains something I had read over the summer that had mystified me, that of money market funds refusing to buy commercial paper due to subprime exposure. At first I had thought it was reluctance to fund SIVs, but they on average hold only 2% in subprime paper, according to Fitch. Even with SIVs have only 1% equity, the value of the subprime paper would have to fall below 50% for that to put the CP at risk, which didn’t add up (the impression I’ve had, which may not be correct, is that the subprimes in the SIVs are the mortgages themselves, rather than CDOs holding subprimes. We have now seen that CDOs that have been downgraded are pretty much unsaleable, so who knows what their value might be).

But while CDOs have a variety of types of assets in them (collateralized loan obligations, tranches of RMBS, often the mezzanine, whole loans), so the composition varies tremendously, a good deal of paper subprime paper went into CDOs, almost certainly more than 2% of total assets for the product. That, plus a CDOs much greater embedded leverage (their value falls faster when they experience defaults in their underlying assets than regular RMBS), makes the CP buyers’ reluctance now seem sensible.

Bottom line: this is still a very bad sign of the state of affairs at Citi, and in the CDO market as well.

Print Friendly, PDF & Email


  1. Anonymous

    whats interesting about citi level 3 assets is that they now amount to 1.5 times total stockholder equity up from 0.5 times stockholder equity in the 2007 q1.

    Incidentally I am willing to bet that for all the big wall street banks, level 3 assets are now well above 100% of total stockholder equity (or for that matter tier 1 capital).


  2. Anonymous

    The level 2 assets (the reading of their values is as interpretative as level 3) are to be shored up to level 3 for two major reasons the mortgage rates interest rates reset are going higher at the same pace as the long term interest rates are deemed to be lifted (balance of payment deficit, larger USD depositors are fleeing the USD, the USA still need 2 billion USD a day to finance the current account deficit)
    Many banks have conduits inclusive of European banks the liquidity is drying up, there is not market for these assets.
    When aggregating level 2 level 3 all investment banks capital is wiped out, this is not inclusive of the LT LBO’s financing which have a discounted value when on sale.
    Overall the banking world looks pathetic!

  3. Anonymous

    They transferred around 11bn of the 40bn, leaving about 29bn new Level 3 trading assets for Q3, with these most likely;

    -Structured credit products—Structured credit products include synthetic CDOs and other complex derivatives. Synthetic CDOs are not collateralized by a physical portfolio of assets like bonds or loans. Instead, synthetic CDOs gain credit exposure to a portfolio of fixed income assets through the use of credit default swaps. Only structured credit products where key inputs, such as correlation, are not observable are classified as Level 3.

    or-Asset-backed commercial paper (ABCP)—ABCP is short-term debt issued by an SPE. The SPE issuer uses proceeds from the issuance to purchase asset-backed securities. Repayment of ABCP is dependent on collections from these investments or the issuance of new ABCP. This inventory is classified as Level 3 due to illiquidity in the ABCP market during the 2007 third quarter.

    They have additionally placed 15.1bn of ‘loans held-for-sale’ in Level 3.

  4. bob

    At first I had thought it was reluctance to fund SIVs, but they on average hold only 2% in subprime paper, according to Fitch.

    Yes but when they count subprime paper, are they including “non-subprime” Alt-A and Option ARMs? It seems many do not. (See the IMF graph of ARM resets/recasts that has been making the rounds in the past weeks.) As if the magic of a high FICO score (the only way I can see a piece of toxic crap like an Option ARM being considered “non subprime”) will allow people to cope with doubling or tripling payments that or increasing payments far beyond their “stated income”.

    Also, does Fitch even know what is in the stuff they rate? The rating companies seem amazingly clueless.

  5. Anonymous

    Incidentally I am willing to bet that for all the big wall street banks, level 3 assets are now well above 100% of total stockholder equity (or for that matter tier 1 capital).

    These figures are given in an FT Alphaville post.

    Caution: the source is a commenter on Nouriel Roubini’s blog. But if the numbers are verified, then the only institution with a Level 3 to equity ratio less than 100% is… Merrill Lynch.

  6. Yves Smith


    Those are very good points, but I was making a very specific argument: that this summer (starting in August) buyers started repudiating commercial paper, and the concern they gave was possible subprime exposure. No mention at the time of concern about Alt-A or option ARM.

    Remember, the proximate cause had been the failure of a fund related to German bank IKB due to subprime exposure, and then the halting of redemptions from three funds run by Paribas because they had roughly 30% subprime exposure and according to them, could not fairly value the assets to allow for redemptions in a way that would be fair to the investors who wanted their money back and the remaining investors.

    So the news that caused the panic was about subprimes, not about the other sorts of mortgage exposures, And that’s what investor comments to the press stressed. Recall also that US money market fund investors were redeeming out fear of subprime exposure, adding to pressure in the market.

    As for Fitch, while it’s hard to know what they know, SIVs are easier to evaluate than CDOs and even regular structured mortgage deals, which is where their ratings have become suspect.

    This paper by Fitch goes into mind-numbing detail about the legal and financial structure of SIVs, and the characteristics of the SIVs that Fitch rates. It is worth noting that SIVs are not as complex financially as CDOs or even traditional asset backed bonds. You have a pool of assets and three elements in the capital structure: commercial paper, medium term notes, and equity (some have a small subordinated layer between the MTN and equity.

    However, another reason the rating agency grades are now viewed with suspicion is that they downgrade late. They tend to do that anyhow, even with corporate securities, and the tendency is worse with asset-backed securities. So that is a fair reason to worry about SIV ratings.

    A paper by Joseph Mason and Joshua Rosner goes in gory detail through what is wrong with CDO ratings. Most of the problems they cited to not apply to CDOs (which isn’t to say that there may not be problems with SIV ratings too, but that as much as everyone likes to blame the rating agencies these days, the flaws in the process with CDOs doesn’t establish that their SIV ratings are screwed up, although given their track record, it is certainly a reasonable suspicion).

Comments are closed.