Sunday, July 22, 2007

Leverage, Ratings and Forced Unwind

by Tanta on 7/22/2007 11:10:00 AM

Honestly, I don't know if this is an UberNerd post or an UnterNerd post. I am not an expert in CDOs or financial "gearing," and I'm happy to be educated by people who know more than I do about high finance. I'm writing this just because our discussions of The Big News Story of the last few weeks are mystifying a lot of folks who are not familiar with the leverage issue. The following is both lengthy and, inevitably, over-simplified in spots, which is such a weird combination. (You'd think for that many words you could get Definitive Answers to Really Complex Questions.) I can only observe that this is why what you read in the papers is, nearly exclusively, so worthless or misleading on this subject. It takes three or four times the length of your average feature article just to clear up the basics, let alone get to the hot new developments in the story.

Given that, if you're still puzzled about how, exactly, all this leverage and marking to market and downgrading of securities can come together to wipe out entire large hedge funds, here's an attempt to explain the basic mechanisms. Those of you who are well beyond basic mechanisms should go play outside today.

Let us imagine a hedge fund operating on a maximum 20% margin, or 4x leverage. That means that for every $100 invested, $20 is equity (investor funds) and $80 is borrowed (from the "prime broker"). The exact mechanism of that borrowing can be more or less complicated (like the dreaded "collateralized repo facility"), but underneath it all it's a margin loan.

That $100 investment is subject to periodic marking to market. What that means, in simple terms, is that the fund must establish what the current observable market price is for the assets in question, and "mark" its holdings accordingly. With frequently-traded things like shares of (and options on) stock and vanilla bonds (Treasuries, GSE MBS), one simply checks the Big Board (or the Bloomberg terminal), finds the closing price for the period in question, and there's your mark. With things that are not frequently traded, like a lot of low-rated tranches of certain complex private-issue ABS (Where the Wild Subprimes Are), it can be a challenge to find an observable market price.

That's what causes this "mark-to-model" phenomenon: the marked price is a construct, a kind of theoretical rather than observed price, based on some assumptions, some adjustments for differences in quality or yield between recently-traded issues and the ones you happen to hold, and so on. Mainstream reporters, for reasons best known to themselves, seem to think such a process is either unheard-of or necessarily Enronified accounting. In reality, just about everyone over the age of 12 has heard of a residential appraisal for a refinance transaction, which is a classic example of "mark to model." The appraised value of your home is not derived from going to the MLS and finding an observed price of your house. The appraiser collects several "comps" and then adjusts them for different quality of the house, age of the last sale, the current cost of financing, and so on. Problems crop up when models are opaque, when assumptions are faulty, and when data is corrupt, in appraisals and elsewhere. But these innocents who just discovered model-based pricing or infrequently-traded assets as if they were A Recent Invention of the Devil are unclear on the concept in a bad way.

However one arrives at a mark, it is meant to be the price the fund would take if it had to liquidate the asset or unwind a position today. It is not a net present value calculation of an asset that is intended to be held to maturity. It is a snapshot in time, and I need hardly mention that it can result in gains as well as losses. It's simply that mark-to-market gains don't tend to cause problems with leverage. We all used to love our appraisals when they were generating the "wealth effect."

What happens if our example investment gets a 5% mark-to-market loss? We now have a $95 investment of which $15 is equity and $80 is borrowed, or leverage of 5.3x. This results in a "margin call" from the lender, who demands that the 20%/4x relationship be reestablished. (Here's where margin lending is very different from mortgage lending, because mortgages are not callable. You live by the original loan terms, you die by the original loan terms. That's why we used to be accused of being so conservative.)

Therefore, the fund sells $20 in assets and pays the lender with the proceeds. That results in a $75 investment ($95 minus $20), $15 of which is equity and $60 of which is borrowed, bringing the leverage back to 4x. What is important here is that the original mark-to-market loss on the investment does not, in and of itself, trigger the sale of fund assets. An unleveraged fund can still hold the assets to maturity, if it chooses and its prospectus allows, waiting patiently and hopefully for the day that market prices recover, the assets can be marked to a gain, and the sea gives up its dead. A mark-to-market loss requires liquidation of assets when there is a lender involved who demands that leverage stay at an agreed-upon limit. You see here how leverage magnifies losses on the downside: a 5% drop in value of the investment results in a 20% 25% drop in the equity value of the fund.

That's how it works in the textbook case. However, a couple of things tend to complicate these "forced unwinds." One is that selling off assets to meet margin calls in an environment when the price is already dropping (that's why we got the mark-to-market adjustment in the first place) creates even more downward pressure on asset prices, forcing another mark, another call, another sale, etc.

Another is that lenders who get spooked enough can raise the margin requirement. Imagine the scenario above--the $95 mark of the investment--with the lender raising the margin limit to 25% or 3x leverage. The fund would have to sell $35 in assets instead of $20, to achieve 3x leverage ($60 investment, $15 equity, $45 loan). Having to sell that much more of the fund's investment creates even more downward price pressure, triggering yet another unwind. We call this a "credit crunch."

Furthermore, investor redemptions can have the same effect, which is why we've seen some funds "halting" redemptions (not allowing investors to liquidate their holdings in the funds) lately. This is what is known as the "rush to the exits" problem: more people can get crushed in the stampede than get burnt by the original fire. By halting redemptions, fund managers are attempting to keep people in their seats while the fire is extinguished. Campers become unhappy.

Finally, a major problem in talking about leverage ratios with hedge funds is that this basic example is just the "financial leverage." Because hedge funds invest in instruments that are themselves leveraged, such as credit default swaps, synthetic or debt-funded CDOs, leveraged loans, and so on, their actual or "effective" leverage can be much, much larger than 4x. When you hear people talk about some high-roller funds being levered at 10x or 20x, they're talking about "effective leverage," not just simple financial leverage.

You should bear in mind that hedge funds being hedge funds--largely unregulated and opaque to the rest of us--public discussions of a fund's leverage are estimates and assumptions, informed or not. Nobody exactly knows except the fund managers themselves, and we have our doubts at times about them. (I for one am ready to be confident that fund managers can fairly accurately calculate their financial leverage; whether the calculations of effective leverage are anything to which the term "accurate" could be applied is less certain. That gets us back to our "mark to model" problem.)

For our purposes, it's important to keep in mind that leverage is the key to understanding how things that look like modest declines turn into major losses. Start with a handful of homeowners who bought a house with 100% financing. A drop in home values of just a few percent puts these borrowers underwater and removes a lot of the incentive to keep making mortgage payments. The lender forecloses and takes, at this point, a fairly modest loss to the asset-backed securities that own these loans. But the credit enhancement of the ABS is fairly thin, and so the lowest-rated tranches of the securities get downgraded. The lower rating means a lower bid price on resale of the securities. That forces a mark-to-market (or "mark-to-model") loss to the holder of them. A CDO of ABS may well issue AAA-rated tranches, but that means that the AAA-piece of the CDO is no more than a senior lien on the payments generated by a large portfolio of BBB-to-B rated tranches of underlying ABS. (You could call that "ratings leverage.")

Perhaps we need to look at a picture again. Here's a chart from Pershing Capital Management:

A CDO of ABS can own 100 BB, BBB, and BBB- tranches from any number of ABS deals. The top tranche of the CDO gets its AAA rating not because any of the ABS it owns are necessarily AAA-rated, but because it does what structured finance instruments do: it carves up the cash-flow and payment priority of the underlying assets, and so provides "credit enhancement" to the top tranches.

Think of it this way: the originator of the mortgage loans we started with "levers" its investment by selling bonds worth 96% of the face value of the loans (leaving it 4% equity, as in the chart above). The CDO buys up tranches of this (and other) securities, and then "levers" its investment by selling bonds worth 95% of the face value of the ABS tranches it owns. By the time you get to the equity portion of the CDO, you have very high yield (the highest-yielding tranche of a CDO composed of high-yielding tranches of ABS), and also enormous risk, since there is no credit enhancement below the equity: it takes all losses first.

But what happens if the BBB- tranches of the ABS start taking major write-downs? The collateral of the CDO can simply disappear at an alarming rate. In the example above, if the original ABS took a 5.5% write-down, that would reduce the face value of the BBB- tranche by half (it would eliminate 100% of the 4% equity, 100% of the 1% BB tranche, and 50% of the 1% BBB- tranche). If a quarter of the bonds in the CDO were BBB- tranches and 10% were BB tranches that all took that kind of loss, then the collateral value of the CDO would drop by 22.5% (25% times 50% plus 10% times 100%). That would take write-downs all the way into the CDO's AAA tranche, which in our example "breaks" at >20%. If the AAA tranche of the CDO is held by a hedge fund using 4x financial leverage, you get a nasty, nasty unwind problem.

I need to stress that CDOs, on the whole, are a lot more complex than this example might make you think. They have short positions as well as long positions, and a lot of credit default swaps and other derivative holdings and bond insurance as well as--one hopes--a fair amount of diversification both within the mortgage category and within the larger category of ABS, which includes other kinds of debt besides mortgages. (Whether that "diversification" is helping any, insofar as some of it involves commercial RE and LBO notes and other perfectly "uncorrelated" risk-free no-brainers, is another question. But we are, at least for purposes of illustration, following the received wisdom that this is all about subprime mortages and nothing else.) It does no one any good to think of "CDOs" as all alike, or as just another kind of CMO or ABS. The fact that the mainstream business press is encouraging you to think that CDOs are just another kind of mortgage-backed security is why we get so pissy with them here at Chez Risk.

I'm not, therefore, trying to suggest that the Bear Stearns hedge funds or anyone else lost their money in this exact way. My point is that it is perfectly plausible that some funds holding CDOs of ABS took huge hits to A-rated tranches even when the underlying ABS collateral has, so far, only taken hits to B-rated tranches. That's the "effective leverage" problem.

We should be aware that downgrades of the lowest-rated underlying collateral tranches can, at least in judicious doses, be a good thing for the holders of the highest-rated tranches. Why? Because downgrades of the lowest tranches will prevent the deals from "stepping down," or redirecting the cash-flow of principal to the lowest-rated tranches. By keeping the principal payments directed to the topmost tiers, those AAA and AA classes prepay while the prepayin' is good.

Another reason that the mainstream press reports mislead everyone about this is that they continue to think of structured securities as "chopping up loans" instead of structuring cash flows from loans. These cash-flow structures mean that the top tranches have 1) first dibs on money coming in from the underlying loans and therefore 2) much shorter durations than the supporting tranches. They're set up with certain "triggers" such that, if and when it starts to look like there won't be enough money for everybody--that's basically what a downgrade means--the flow of cash goes primarily or even exclusively to the top tranches until they're retired (amortized or paid off completely), and only then do the subordinated tranches get any leftovers. Depending on how a given deal is structured (if they were all the same, remember, we wouldn't have such "mark to model" problems), a ratings downgrade to the lower tranches can seriously boost the cash flow up above.

(To those of you who keep asking in the comments why the rating agencies aren't downgrading the A-rated parts of these deals much: remember that downgrading the subs can, in and of itself, improve the loss probability for the senior notes by accelerating their payoff, and some of them in the older deals have already amortized so much that time is very much on their side. The senior notes are a lot less likely to get crushed in a rush to the exit if they're already half-way out the door.)

Remember all the outrage over loan modifications in these securities? If you're holding one of those subordinate tranches, modifications will prevent (in better case) or delay (in worse case) actual realized losses, which is good for you, at least in the near term. However, if you're holding one of those "super seniors," modifications can extend the duration of your investment by slowing prepayments (a modified loan stays in the security, while a refinanced or foreclosed loan is a liquidation or prepayment) and by helping the deal pass the step-down triggers that allow some cash to be shared with the lower-rated tranches.

My own view of the matter is that a lot of this anger and contempt and disgust and general ill-feeling with the rating agencies coming from certain quarters of the investment community is not really a matter of holders of the toxic tranches complaining that their BBBs are "really" BBB-. (This, you might say, is the part we're not "surprised" about.) It's coming from AAA holders who want to see the subordinate tranches downgraded as fast as possible so that the AAAs pay off as fast as possible: from this angle, the problem is not the downgrades but the timing of them. (This is the "why did you wait until there were actual losses?" part.) Class warfare is an ugly thing, and the rating agencies (and mortgage servicers dealing with modification problems) are in a tough position if they think they can make everyone happy. As they are in a tough position which they happily participated in creating, you can as far as I'm concerned leave your lace hankies in your lingerie drawer. This isn't about sympathy; it's about recognizing complexity.

If you require a moral to your story, I'll suggest this one: we just spent the last several years being told that securitization of mortgage debt was a reliable way of minimizing (in some quarters, downright "eliminating") the risk of making very high-risk mortgage loans, by moving the risk off the books of the lenders and onto the books of sophisticated investors who knew how to hedge it. Evidently some sophisticated investors are now telling their lawyers that nobody actually knew what this stuff was or how it was supposed to be hedged or that those juicy returns were being generated by incredible amounts of leverage. To these folks, it was so credible that investing in subprime mortgage loans could easily throw off 20% annual returns forever that they had every reason to believe that their fund managers knew how to do this without losing any of their principal ever, cross our hearts and hope to shout.

So the great innovation of securitizing mortgage financing in support of a wild housing boom involves a chain that starts with a sucker on one end who overleverages a real estate investment with a pseudo-callable loan (i.e., an ARM that reprices to market) and ends with a sucker on the other end who overleverages a hedge fund investment with an explicitly callable loan. In the middle are a bunch of bright lights who kept it all going long enough to extract a lot of fees and bonuses. On the sidelines are a lot of people who leveraged conservatively but who get the same mark-to-market adjustment everyone else does. To judge from the whining from certain parts of the hedge fund world, those "qualified investors" were just part of the great "unqualified buyer auction" that so sadly played out in the housing market, bidding up the value of these CDOs and ABS until it seemed that no amount of leverage was too much and no position would ever have to unwind badly. Scratch the surface of the complaints over "mark to model" adjustments, and you get to the question of which model everybody used to establish the original price of these securities from which perspective the current mark is looking ugly.

Unless, of course, these "qualified investors" were in fact perfectly qualified buyers in the auction, in which case the appropriate response to the Great Bear Stearns Pile On would be something along the lines of "tough breaks." In any case, this is where acid flashbacks to the bad trip of "investor liability" for predatory loans comes in. Mercy, we have been told, how unfair that is, because investors can't be expected to actually investigate whether Bear is preying on Bubba before investing the big bucks. Yeah, well, that's so last week. This week's theme is how Bear preyed on Buffy, and that's serious, dude.