HuffPo on CDOs: Great Metaphor Marred by Some Incredible Assertions

It’s probably a character defect, but I get wound up when I read something that is directionally correct but then discredits itself by getting important facts wrong.

The latest case in point is a Huffington Post post by Eugene Linden on “The Ecology of Toxic Mortgages.” It’s a more than usually frustrating example because 1) Linden presents himself as an insider 2) he presents factoids on CDOs, an area where concrete information is sorely lacking and any authoritative comments are likely to be treated as gospel;and 3) HuffPo attracts a large readership, many of which I imagine to be intelligent but not necessarily very far down the curve in the operations of financial markets, and so are probably not equipped to give the piece a critical reading.

Now let me stress again, the piece has a lot going in its favor; it’s well written, attention-getting, and accurate in its overall observations. Let’s start at the top:

I lead two lives. Three days a week, I’m employed as chief investment strategist for a hedge fund that specializes in distressed and bankrupt situations. The rest of my time, I do what I’ve done for decades, which is to write about nature and the environment. There is virtually no overlap between these two worlds — with one exception. At a metaphorical level, there are irresistible parallels between a profound flaw in early models of how to deal with pollution, and an almost exactly analogous flaw in financial models for how to deal with the financial universe’s own version of toxics: risk.

My role at the fund is to look at the macro situation, and to help the portfolio managers interpret how larger trends in the economy will interact to the benefit or detriment of our investments and prospective opportunities. In that capacity, I’ve been looking at the unfolding debacle in subprime lending, a slow motion, far-reaching toxic poisoning, whose reach and impacts have been obvious for at least eighteen months to anyone not involved in making money off the origination, sale, and securitization of these subprime loans.

Unfortunately for investors, that aforementioned conflicted group includes virtually everybody in finance, including the mortgage brokers, subprime lenders, Wall Street firms that securitize the loans into mortgage-backed securities, Wall Street firms that then resecuritize slices of these bonds into collateralized debt obligations, and the rating agencies that, for a price, enable all these securitizations and re-securitizations, by blessing these teetering structures with ratings that imply far less risk than is turning out to be the case. There has been a good deal written about the ecology of finance in recent years, but reading about theoretical parallels between the worlds of nature and finance pales in comparison to the thrill of watching a toxics crisis in finance unfold before your eyes almost exactly as it does in the environment. For all our vaunted foresight, it’s interesting to see that when greed and self-interest come into play, collectively we’re no smarter than fruit flies.

In this case, the flawed environmental model for dealing with risk might be summed up by the cute phrase, “dilution is the solution to pollution.” For a number of years, we freely poured toxics into the water and skies under the assumption that pollutants would disperse and become harmless in these vast receptacles. Instead, what we discovered is that these toxics re-accumulate as creatures eat each other and are eaten, a process repeated on up the food chain until the toxics reach deadly concentrations in the top predators and big animals. I remember years ago reading that dead whales washed up in the Saint Lawrence seaway contained such high concentrations of heavy metals and other toxics that in the U.S., they would be declared superfund sites.

That also could be said for some of the big investment banks, hedge funds, and Wall Street firms at the moment. The toxics in this case would be portfolios of various forms of securitization of subprime, alt-a and other loans that, amazingly enough, aren’t performing according to models developed during the greatest run-up of home prices in American history. (I recall attending one conference on securitizations of home equity loans in early 2006 where the quants showed us supposedly reassuring “stress” tests of these bonds under various scenarios of home price appreciation. The most “stress” they envisioned was 3% appreciation, and not the negative price movements we are seeing just a year later). The practical logic behind packaging these risky loans was that most of them were money good, and that so long as defaults did not exceed expectations — say 4-5% of the loans being packaged — the great preponderance of the securitization could be treated as investment grade. And, the philosophy behind this whole process was that risk could be reduced if it was sliced up and efficiently dispersed in the investor ocean.

But, in an exact analogy, to the environmental example, risk did not stay dispersed. Rather it re-aggregated in the whales (hedge funds, investment banks, and pension funds) of the investment community. And now these top dogs are discovering that risk is just as toxic if it’s sliced up and reformulated as if it never was broken up in the first place.

This is great stuff, the confessional intro, the analogy between environmental and Wall Street toxins, the insider detail first about dead whales, then about overoptimistic quants.

But then he tries to prove his thesis and the piece starts to go off the rails….

The analogy does break down ultimately, because in the investment universe version we have an accelerant to the toxicity of risk in the form of leverage. Because so many of these repackaged subprime loans were rated investment grade, the whales could gorge on the stuff using borrowed money. The embedded leverage is astonishing. While each deal is different, and this unregulated market remains opaque to non-participants, an idealized example illustrates this point:

Take a billion dollars in subprime mortgages and package them into a new security. Typically, a model security would rate about 95% of the slices in this new bond as investment grade. Under these high-rated slices are what are called mezzanine tranches, the lowest piece of the investment grade slices, and the lowest of these would be rated BBB-, or just above junk status. Typically, these mezzanine tranches will amount to about 4% of the$1 billion total value. Below the mezz pieces would be the lowest rated tranches, including the equity which absorbs the first losses if borrowers default. In this idealized securitization, the BBB- tranches might represent 1% of the total value of the bond and be buffered from losses by about 5% of equity and junk (which represents a computer model’s estimate of the outer limit of realized losses).

If you are going to do this sort of thing, the math has to work. I read the second paragraph four times, and I read him as saying that the BBB- tranche is 4% and later 1%. If you are going to set up an illustration, it needs to be clear and consistent.

Back to Linden:

So in this case, those buying the BBB- tranche are betting that losses for the entire billion dollars in loans never rise above $50 million over the life of the bond. Fair enough, but if they do rise higher, those holding this tranche lose money in a hurry. Let’s say, losses rise to 8% (some predictions are even higher). In that case, the value of the BBB- tranche would be worthless, and losses would take out all of the BBB tranche and half the BBB+ tranche as well. That’s the price of leverage.

Where is the leverage in this structure? Perhaps I am being too fussy about terminology, but tranching means you have set priority in who gets paid first. The way leverage kills you in an investment is that if you have put down only $20 to buy something that costs $100: a mere 10% decline wipes out half your equity. Here, the chumps who bought the equity tranche may well get wiped out, but not because the structure had any borrowings in it, but because they bought something analogous to a C or D rated bond: high yield, high risk of loss.

And I am curious about this 8% loss figure. I’ve seen estimates of subprime defaults running at a widely reported 13 to 14% (20% is likely more accurate) and foreclosure figures all over the map due to wide differences in definition and reporting. However, the key issue is that defaults do not equal foreclosures, and foreclosures do not equal losses.

Foreclosures lead to three things: an interruption in monthly interest payments, an acceleration of principal when the property is sold (which means bye bye future interest payments), and possibly a loss if the recovery is less than the mortgage balance. (At the foreclosure sale, the mortgage holder or its agent will bid for an amount equal to the mortgage balance, so either a third party beats that bid, paying off the mortgage, or the mortgage owner winds up owning the property). Unless the property was condemned, there will be some recovery. The property is not worth zero.

Consider this estimate from American CoreLogic, which published an extremely detailed break down of mortgages by type (when originated, what initial interest rate, what reset date, what reset level, how much equity in the property at time of mortgage origination). Its conclusion:

The main-line scenario, which can be experimentally varied to investigate different possibilities, estimates the total amount of loans that may default under reset sensitivity combined with a lack of equity. Mortgage payment reset is expected to have its impact into the early years of the next decade. Under the main scenario, the study anticipates 1.1 million foreclosures spread out over a total period of six to seven years – which represents thirteen percent of the adjustable-rate mortgages originated through purchase or refinance from 2004 to 2006 – constituting $326 billion of debt. After foreclosure and resale, it is projected that about $112 billion will be lost to remaining equity, lenders and investors over several years.

Before you protest that 13% is too low (I agree, BTW), note that this study was focusing on all adjustable rate mortgages originated during the 2004 to 2006 period, so it included alt-As, so the defaulting percentage of subprimes is higher (the study assumptions were 32% for teasers, which were heavily skewed towards subprime, and 12% for other subprimes). And while the $112 billion estimate is optimistic ($170 billion seems to be the new middle of the road number, and some, without explaining how they derived it, are bandying about $250 billion) what interests me is the estimate of mortgage value that goes into foreclosure – $326 billion – versus the amount that winds up being lost – $112, or roughly 1/3.

So 8% in losses in consistent with roughly 24% foreclosures (and I don’t mean foreclosures where the owner is able to sell the property on his own, which is often what happens when the foreclosure ball gets rolling. Any owner with equity in his house will sell it rather than have a foreclosure on his credit record). That implies a calamity of massive proportions. You’d see government intervention if things got that bad.

Now perhaps I am being a bit unfair, since I am talking in aggregate, while he is talking about one $1 billion transaction, which could be more doggy than average. Let’s continue:

But it gets worse.

Given the risks of subprime loans, many lenders could not afford to make large volumes of loans if they were forced to keep the loans on their own books since they would tie up too much capital. So they finance the loans with short term borrowing and then sell the mortgages into securitizations. The buyer — the securitizer — then puts together his MBS. To do this, the buyer has to sell the mezzanine tranches (many securitizers keep the equity themselves). These tranches buffer the whole structure from losses, and once they have been sold, it’s easy to sell the higher rated stuff.

In recent years, the money funding these mezzanine tranches has come from a subset of another securitization called collateralized debt obligation or CDO. To form a CDO that invests in subprime mortgages, a securitizer will buy up mezzanine tranches from perhaps 100 different mortgage-backed securities, and then package them in different tranches similar to the way a mortgage backed security was packaged in the first place. Thus, some CDO’s can consist entirely of BBB- tranches of subprime mortgage MBS, but still have 95% of their value rated investment grade.

So far, so good. Only one quibble: “some CDO’s can consist entirely of BBB- tranches of subprime mortgage MBS, but still have 95% of their value rated investment grade.” Yes, but he just told us above that that’s either only 4% or 1% of the deal’s value. And from what I can tell, that isn’t common practice. From what I have read, it seems more typical for CDOs to have a mix of paper (unless they are CDO-squared or cubed, which hold only other CDO paper, I can’t generalize about them), which can include commercial mortgages and LBO debt along with residential mortgages. One of the reasons to buy a CDO rather than an MBS is to get diversification.

Back to Linden:

Here is where leverage is the true killer. While an increase in realized losses from 5% to 8% will wreck havoc on a $1 billion MBS, even a smaller increase from say 5% to 6% losses could utterly destroy a CDO based on BBB- tranches where the leverage is over 100 to one. That additional one percent in losses will not only wipe out the bottom tranches of the CDO, but it will eat through most of the investment-grade slices as well. Bearing in mind that many hedge funds also used leverage (meaning that they borrowed most of the funds to buy a CDO tranche), it becomes obvious that even minor variations from the expected performance of subprime loans can have a huge impact on results.

This is the part that sent me off the deep end. Where does he get this 100 to 1 figure from? It’s not simply that I’ve never seen that sort of figure claimed for CDOs, I’ve never seen that sort of leverage for a single instrument in the financial markets. There are some derivative contracts where the notional amount is a hundred times the cash value, but that isn’t what he is talking about here. LTCM, which was known for leveraging to the max in liquid markets, got to 28 to one leverage at its highest, excluding derivatives.

You can use a repo to create leverage, and I used to know but have completely forgotten the rules, but I am pretty certain that even with Treasury bonds you can’t get 100 to one leverage (although I think you can get close). That’s as good as it gets. Exchange traded futures are cash settled daily (which is tight credit control) and even those you post at least 5% opening margin (meaning only 20 to 1 leverage).

Now I’m not arguing that you can’t eventually get to 100 to one leverage with a CDO once you get further out in the food chain. Perhaps a CDO goes into a second CDO that used derivatives which is bought by a hedge fund that uses leverage and some investors put their money into that hedge fund via a fund of funds, which also was geared. But that isn’t what he said. Using words like “leverage” when you might mean something like “the effect of subordination on first losses” just compounds the general confusion among the lay audience.

It’s critical for those of us who are trying to play one-eyed man in the land of the blind not to go beyond our facts and be as precise as possible in the use of language, even if that means belaboring things a bit. Ultimately, there will be more regulation in this area, and it’s going to be based on what was perceived to have gone wrong. It’s therefore incumbent upon all of us to do what we can to have those perceptions bear some relationship to the underlying reality.

Those of you who follow this blog know I expect this to come out badly. How bad depends on how the various actors behave when real trouble begins. And I suspect that some, perhaps much, of this paper was under water from the get-go, that even using the assumptions under which it was sold, investors overpaid by a big margin.

But Linden’s use of the most extreme estimates of losses and implausible levels of leverage makes it easy to discredit his insight, namely, that the damage of the “toxic waste” tranches will be more pronounced the further you go up the food chain. And it’s the investment banks who are the alpha predators.

Update 7/15: Linden wrote me a polite e-mail, which was very sporting of him under the circumstances, to clarify his discussion of the mezzanine. And in fairness, I may have been taking out on him an odd use of nomenclature I have observed elsewhere in CDO land. The “mezz” tranches for CDOs are the Bs, from BBB to B. The equity tranche is unrated. However, I often read the CDO mezzanine trances described as not being investment grade. That’s not accurate. Any triple B rating, down to BBB-, is investment grade. So some mezzanine paper IS investment grade, some isn’t. And interestingly enough, that notion, that mezzanine paper can be investment grade, is at odds with the way mezzanine is construed in most other segments of the debt markets. Mezzanine is generally junk, meaning non-investment grade.