Thursday, May 29, 2008

CDOs: The Future of Leveraged Fixed Income

Collateralized Debt Obligations (CDOs) have been the bane of Wall Street recently. Falling CDO prices have been a major player in the write downs at big brokerages and the near insolvency of monoline bond insurers. And yet the CDO structure will not only survive this period, but is likely to become the dominant means of leveraged investing in credits.

First, let's consider what a CDO is, and what it isn't. At its simplest, a CDO starts with a portfolio of credit-risky securities. The purchase of this portfolio is funded by the sale of a series of debt tranches. Payment to debt tranche holders follows a seniority scale, such that the senior-most debt holder gets paid first, then the next-most senior, and so on. Most CDOs have at least 4 or 5 levels of seniority among debt holders. If there is any cash flow left, it is paid to an equity holder. It is common for the equity holder to represent as little as 1-3% of the total structure.

It sounds complicated, but the core concept is quite simple. It is similar to how banks fund themselves: you have depositors, senior debt, subordinated debt, preferred stock, and equity. As long as everything is going well, all debt holders get paid what they are promised, and whatever is left over is what the equity holder owns. If the bank goes under, depositors get paid first, then senior debt, etc. A CDO is basically the same idea, except that whereas the bank would have to go bankrupt before the credit tiering came into play, a CDO will take losses on individual credits over time.

The senior-most tranche of a CDO was usually rated AAA. This piece would usually amount to 60-80% of the total CDO structure, or put another way, 20-40% of the cash flows were subordinate to this senior-most piece. This meant the structure could take on substantial losses before the AAA tranche would suffer any cash flow short-fall.

Several things went wrong for the CDO market over the last 18 months. Many CDOs were constructed from consumer loan-related securities. It was assumed that losses in consumer loans would have a relatively low correlation, that a home equity loan for a teacher in Sacramento would have no correlation with a loan to an auto mechanic in Orlando. That assumption proved disastrously false in 2007 and caused CDOs built from consumer loans to fall apart en masse.

But it wasn't really the failure of CDOs themselves that created so many problems. Take the collapse of the SIVs. In most cases, SIVs collapsed not because they took on too much in cash flow losses, but because no one would buy their commercial paper anymore. In other words, the SIV arbitrage relied on the continued confidence in the SIV portfolio. Once that confidence was gone, regardless of what was actually in the portfolio, the SIV was toast. The very same thing happened to countless hedge funds and other leveraged vehicles in recent months. Any vehicle that relies on short-term funding is banking entirely on the continued support of short-term investors. Once that's gone, the whole structure is destroyed. A CDO doesn't have this problem. Generally speaking, the funding of a CDO is locked in at issuance.

Let's compare and contrast for a moment. A CDO equity investor, the one who only gets whatever is left over after all debt holders have been paid, is making a highly leveraged bet on credit losses within the CDO's portfolio. But that's the only bet being made. An investor in a hedge fund relying on repo financing is making a bet on both the portfolio and continued access to financing. Why should investors make two bets when they could make only one?

In addition, if properly funded, CDOs are safer for the system compared with other types of leverage. A CDO is a closed loop. If CDO portfolio losses are greater than initially assumed, investors in that CDO will suffer, but there is no contagion effect. We don't find pockets of lenders to the CDO hidden here and there. Of course, if CDO debt tranche purchases were funded by borrowing, then that's a different story. But such borrowing is not inherent to the creation of CDOs.

CDOs also rely on a very well known and time-tested arbitrage, assuming the right kind of collateral is being used. Credit investments have highly skewed return distributions: either you'll get paid the promised rate, or the bond will default and you'll take a huge loss (this is termed negative skew). It is well established that investors detest large losses more than they lust after large gains. Therefore credit spreads almost always price in more interest than is warranted by default expectations alone. A CDO can therefore buy a portfolio of credit instruments, and if the correlation of defaults is relatively low, make arbitrage profits on the investors disdain for negative skew.

I'm not suggesting the CDO market will soon return to its salad days of 2006. There were a lot of overly complicated structures and even more poorly conceived collateral portfolios. But the CDO market is starting to make a comeback, with 26 CDO deals pricing in April and May. As long as there is demand for leveraged credit, there will be, and should be, a CDO market.


seanDC said...

Great points about contagion and the fact that the leverage is generally permanent.

I believe one of the primary reasons for the CDO issues over the past year is that the mortgage CDOs were constructed not of underlying loans, but rather of near-equity (BBB) tranches of ABS. So, you're not getting diversified exposure to loans--you're getting exposure to the second wave of losses on packages of loans. And while it may have seemed like a reasonable assumption (not accurate in retrospect due to pervasively bad underwriting, but perhaps reasonable at the time) that the loans of Judy in Sacramento and Dave in Tampa would have little in common, it would never have been reasonable to believe that the second-loss tranches of ten diversified batches of subprime home loans would be uncorrelated.

If you write put options on 10 stocks in different industries, that might be a relatively uncorrelated series of bets. But if you write put options on 10 different large-cap stock indexes/ETFs, you are making the exact same bet ten times. It's a digital bet--you're right on stocks writ large, or you're toast.

I think the CDO structure can still be useful if there is an economic reason for the underlying credits to be diversified. But if the underlying credits are multiple broadly-diversified baskets (of loans/bonds/whatever), then they will again turn out to be highly correlated.

Anonymous said...

What about credit enhancement insurance to bring the CDO up to AAA status? Isn't that a source of potential contagion? There's tens of billions of insurance out there.

Having said that, with proper underwriting and risk management there's nothing wrong with the structure per se. It's the pervasive abuse of it that has caused the problems. Of course, with proper oversight and underwriting the structure might not be profitable/necessary either.

Accrued Interest said...

SeanDC, I completely agree. I wrote about that in September here:

Anon: Sure, but again, that's not inherint to the CDO structure. In other words, you don't need monoline insurance to make the CDO market work.

In many cases, the monoline insurance was purchased to improve the accounting treatment of already AAA CDOs for banks. Just speaks to how the stupidity of banking regs leads to others problems.

Anonymous said...

What a fantastic post, AI! Thank you.

proton said...

CDOs could come back but not in the form that was popular. Until recently, CDO investments were based on:

- blind reliance on ratings
- rule based structures, independent of information market tells us through prices
- underestimated correlation behavior

These premises turned out to be seriously misleading.

Correlation assumptions are particularly important. CDO investments are not only bets on the overall level of losses but on the correlations of defaults as well.

If the market is pricing in an overall systemic meltdown, the prices of the CDO tranches could change immensely. We have witnessed this already in a period with almost non-existent defaults in corporate sector.

ziggurat said...

One aspect of the CDO concept is the trade off between pooling and transparency. There is also the question of whether the whole is worth more then the parts. Until the meltdown, it was obvious that pooling and structuring were seen as adding value. Right now, I think the opposite is more likely, at least with RMBS backed CDS's.

There is a lot of work needed to rebuild the trust that has been destroyed regarding valuation of the underlying assets and rating models. Maybe too obvious to mention.

Compare and contrast a tricked out CDO with a stone age credit structure like closed end fund Auction Rate Securities. The auction mechanism failed spectacularly, but the securities are still AAA and will be redeemed at par (likely, imo, in the next 12 months). Instead of 80% AAA, they had 200% coverage ratio (50% subordination), market prices on the underlying credits, and a mechanism to restructure under duress (deleveraging). The coverage ratio wasn't set using computer models but once again a stone age rule of thumb embedded in the Investment Company Act of 1940. Maybe this isn't a good analogy, but it fits with my Neo Luddite world view.

As far as your overall thesis -- CDO's will come back -- I agree to some extent. The underlying concepts are sound. The confidence and trust that facilitated the current problems are not coming back soon.

Accrued Interest said...

Proton and Zig: Clearly the CDOs of 2008 and 2009 are going to be far simpler and probably with greater subordination than those of 2006 and 2007. I can't imagine anyone doing one with consumer credits forever. Or as long as forever is on Wall Street.

Your CEF example is a good one. Imagine a CDO based on HY bonds where the senior-most tranche was $300 million of a $500 million deal. The senior has 1.67x coverage in terms of debt/assets doesn't it? Then given that the HY bonds are probably yielding like L+400 and the AAA is owed like L+100, the interest coverage is even better.

$300MM x 2.68 (3M LIBOR) +1.00 = debt service of $11.04MM.

The portfolio pays a coupon of ...

$500MM x 2.68 + 4.00 = 33.4MM

That's a coverage ratio of 3x. Again, not bad.

I think that speaks to how a CDO could be build on good old fashioned common sense coupled with better constructed computer models that take into account dynamic correlation.

Dave M. said...

AI said.."Let's compare and contrast for a moment. A CDO equity investor, the one who only gets whatever is left over after all debt holders have been paid, is making a highly leveraged bet on credit losses within the CDO's portfolio. But that's the only bet being made. An investor in a hedge fund relying on repo financing is making a bet on both the portfolio and continued access to financing. Why should investors make two bets when they could make only one.""

And the investors in Annaly and Capstead are also making two bets? Even though it is agency paper, you still increase the risk profile? Thornburg was even worse? No agency paper there? How would you compare this risk with the single exposure of CDO equity? Probably a dumb question, cause I'm still learning. Thanks!

Accrued Interest said...

Good question. Add Bear Stearns to your list. Any company who has short-dated liabilities and long-dated assets is making a bet that they will have continued access to capital. NLY in particular is little more than a public hedge fund. I used to really like NLY's strategy, but I think the fact that they aren't moving away from repo is just plain dumb. Right or wrong, they may get away with it for years. But its still dumb.

Dave M. said...

AI said.."I used to really like NLY's strategy, but I think the fact that they aren't moving away from repo is just plain dumb. Right or wrong, they may get away with it for years. But its still dumb.""

That's why I'm keeping an eye on Chimera, CIM. It's run by the Annaly guys but has a hunting license to go anywhere, IIRC. Know anything about it?

matt m said...

How much of this renewed CDO activity is being done to create collateral for the Fed or the ECB? I see Lehman just put together a $4.5B CDO intending to give the $3.4B senior tranche to the ECB and get Treasuries.

Anonymous said...

I agree with a lot of your logic but am left wondering whether the fashion for leveraged investing in credit is not the thing that will go away, hence wiping out demand for CDOs. I have been wondering throughout this crisis at the way that investors who have seen how much trouble leverage can cause on company balance sheets rushed themselves to embrace that leverage. And if you want equity-like returns, isn't an easier way to make "just one bet" simply to buy equities?

Accrued Interest said...

Matt M: Have you heard something to that effect or are you just asking? I haven't heard anything like that, but I can't say its impossible. We haven't seen a large increase in utilization of the TAF, so it wouldn't seem like that's happening.

Anon: Bear Stearns sure wasn't a single bet! Although now with the TSLF and TAF, that kind of run-on-the-bank is a lot tougher to have happen.

I disagree with those who argue that we're going to see investors permanently eschew leverage. I think banks and brokers are going to have to cut leverage. Hedge funds too. But ultimately the greed/fear cycle will turn. I'd say its turned somewhat already.

jult52 said...

Aren't CDOs a transfer of the duty to research credit quality of the underlying assets from the buy side (the asset manager, who assembles a group of assets in a fund) to the sell side (the packager of the CDO, who doesn't hold the assets in the long term)? If so, why wouldn't that expose the structure to the usual agency problems? (And opaque costs?)

Accrued Interest said...

That depends on the type of CDO you buy. If its a static pool, then I'd say that its a quant bet that the static pool will suffer losses within a certain range. Doesn't really relinquish credit analysis so much as it changes it to a portfolio level analysis rather than a security level analysis.

With a managed CDO, then I would agree that the buyer of the CDO tranches has in effect hired an investment manager to run a credit portfolio.