Friday, June 01, 2007

Don't worry, she'll hold together

Like leverage? CDO equity is for you. As discussed in an earlier post titled How do CDOs Work? CDO equity can be very risky. Many times, CDO equity has de facto leverage of 20-100 times. On top of that, CDO collateral is often made up of weak credit quality instruments, such as junk-rated bank loans, CMBS, or sub-prime RMBS.

One would expect the returns of CDO equity deals to be highly volatile, but a recent report from Citigroup might surprise you. They found that CDO equity return volatility is not consistent across deal types. The following table is from a May 30 report titled US CDO Equity Returns - Our Fourth Update.

Mezz ABS generally means asset-backed bonds rated A and lower, but in this case, probably concentrated in BBB and lower stuff.

Bank loans sure seem to carry the day. Most CLO's ("collateralized loan obligation" -- the term used for bank loan CDO's), have average ratings in the B or BB area, so this is dicey stuff. Why are the returns so consistent?

A big part of it can be found in the recovery statistics for bank loans. According to the 2005 Moody's default study, senior secured bank loans had an average recovery rate of 70% from 1982-2005. That means that given a default, the lender recovered 70% of principal on average. For senior unsecured loans, the figure is 58%. Compare that with bonds, where the senior secured recovery was 52% and the unsecured a mere 36%.

There is reason to believe that recovery rates are more consistent over time than default rates. Moody's doesn't publish historical figures for bank loans, but taking a look at default and recovery for bonds paints an interesting picture. See the graph below.

I adjusted the scales such that the percentage variance from the median was approximately equal. Visually, this shows you that among bonds, the recovery rate is much more consistent than the default rate. To put numbers to it, the percentage standard deviation of default rates from 1982 to 2005 was 61% versus only 23% for recovery rates.

Its logical that recovery rates would be more consistent and predictable than default rates. Corporate default rates have historically been a function of the business cycle and degree of credit availability. Classically, credit is widely available when the economy is booming, but towards the end of the boom cycle, banks and bond holders start making poor loans. When the economy turns, the projects funded start to struggle, and withing 2-3 years, default rates spike up. Note that the actual recession may be over (e.g., the spike in 2002) but that doesn't mean that the default spike was not a result of a weakening economy.

On the other hand, the types of assets that used for recovery, such as real estate and equipment, tend to have more consistent value in liquidation. Although the graph above shows that recovery is actually weakest exactly when defaults are high, this is a marginal difference. The lender still winds up with decent recovery.

So back to CDO equity holders. Even though the default rate of a loan portfolio may be highly variable, the fact that recovery is so strong mitigates the losses. So the CDO waterfall is more likely to stay in tact, and therefore pay the equity holder a solid return.

Now you might ask, if its all about recovery, why doesn't the ABS and CRE deals perform just as well?

  • Although all ABS are technically securitized by some sort of cash flow, you have to really think about how secured the cash flow itself is. Many ABS (particularly the mezzanine stuff) is backed by unsecured loans (like credit cards, student loans), second liens (like HELOC), or depreciating assets (like car or boat loans).
  • ABS is also thought to have a higher default correlation compared with bank loans. In other words, particularly with consumer ABS, the defaults tend to occur all at once, which tends to stress a CDO. Think about the current situation in subprime as an example of how this might happen. A HELOC deal might have almost no defaults for a couple years, then suddenly is slammed with consumer balance sheets weaken.
  • Commercial real estate probably has the same correlation problem. When a bank makes an operation cash flow loan with real estate as collateral, the real estate value is secondary. With a CRE deal, the real estate is primary.

Do you hear me baby? Hold together.


Anonymous said...

Great reading - thank you.

Anonymous said...

I'm not sure that I understand the reasoning behind CRE CDO's having a higher variance in returns.

CRE investments are backed by income generating properties, which historically are underwritten at levels where cash flows cover debt service (until very recently). To me, this makes it similar to a bank loan backed by assets. Also, generally CRE recovery rates are high relative to other distressed assets. I wonder if the variation might be due to a smaller population of CRE CDO (fairly new asset class to CDO mkt) and some particularly venomous collateral placed into an early deal or two that is skewing those results.

Any chance I could get a copy of the Citi report?

Accrued Interest said...

The Citi report is on their institutional website. E-mail me (accruedint at

On the CRE thing, I'm admittedly trying to make the logic fit the data. My guess is that CRE deals have a higher asset correlation. Because a CRE project is reliant almost entirely on a strong real estate market. Whereas a bank loan with real estate as collateral has the corporate operations as the first credit, the real estate as the second credit.

Its also possible that some of the differences are explained by vintage.