CDOs explained

NEW YORK (Fortune) -- Say CDO to anyone in the banking industry right now and they'll probably respond with another acronym: SOS.

CDOs, or collateralized debt obligations, have triggered large losses for several banks, including Citigroup (Charts, Fortune 500) and Merrill Lynch (Charts, Fortune 500). But what is a CDO and why have they ended up hurting banks so badly?

And while we're at it, what the heck are these 'liquidity puts' that led Citigroup, as well as other banks, to increase their exposure to CDOs so suddenly?

To start out, it helps to think of CDO structures like you would any fund: They raise money from investors and then buy assets with that money.

A large bank like Citigroup would act as a CDO arranger, which means it would set up a separate trust. To raise money, the trust would sell bonds to investors. Then it would buy assets.

When constructing a CDO, the overriding aim was for the yield on the trust's assets to be greater than the yield on the bonds the trust issued to raise money. Different CDOs bought a different range of assets, but many bought bonds collateralized with subprime mortgages, loans made to homebuyers with sketchy credit who were charged higher interest rates than prime borrowers.

And CDO entities didn't just sell one type of bond to raise money. Instead, they sold bonds that had differing degrees of credit risk, with the most risky generating the highest yield.

The riskiest slice of debt had the most exposure to losses on the trust's assets. As a result, it would get the lowest rating from ratings agencies like Moody's (Charts) and Standard & Poor's (Charts, Fortune 500). The slice of debt that was last in line to take any losses was often dubbed the "super-senior tranche" and rated AAA, the best rating possible.

So what went wrong? In short, all the parties involved massively underestimated the amount of losses that would occur on the CDOs' assets. In particular, with soaring delinquencies on subprime mortgages, it soon became clear that many CDOs wouldn't be able to pay back their own bond holders in full.

Funds who bought the CDO bonds will of course have to book losses, but how did this debacle hurt the banks?

When the credit crunch intensified in the summer, demand for CDOs dried up. Consequently, banks that arranged CDOs got stuck with subprime assets they had intended to eventually place in CDO entities. In addition, as underwriters of the deals, some were also left with large amounts of CDO bonds they could no longer sell. In both cases, the CDO assets and debt were already on a bank's balance sheet.

But in some cases, banks were suddenly forced to take large amounts of fresh CDO debt onto their balance sheet.

Here's how. Some CDO trusts funded themselves by repeatedly selling short-term debt into the markets. When the debt became due after, say, three months, it was supposed to be re-sold to investors, or in market jargon, it would be "rolled over."

In the summer, with the credit crisis deepening, investors had no desire to keep buying any short-term debt issued by CDO trusts. But some banks, in trying to make their CDO deals more attractive to investors, had committed to buy that short-term debt if it couldn't be rolled over. Citigroup called that commitment a 'liquidity put,' when explaining why it ended up taking $25 billion of CDO debt onto its books.

Legally, banks can't get out of commitments like this. Also, if they didn't step in and supply the short-term funding, the CDO trust would be caught short and it would have to sell off its assets. That in turn could cause ratings on the CDOs debt to fall further.

Typically, this short-term debt is the most highly-rated. But those ratings mean little now, because losses on subprime assets are likely to be far higher than anyone expected.

The short-term CDO debt now has to be valued on the bank's balance sheet, and reductions in value get recorded as a loss in the income statement.

Banks claim there is no liquid market for CDO debt in order to get a firm market price. But that's disingenuous.

A bank could always find firm buyers if it dropped its price low enough. But banks are reluctant to do that, often arguing that trades like that would force them to use 'firesale prices' to record losses.

But recent guidance from auditors suggest that banks, in calculating losses, will have to use the low prices at which they could get a firm offer from a potential buyer -- regardless of whether they like those prices or not.

As a result, expect CDO losses to be even higher than the banks' estimates to date.