I usually rely on public information, but I’ve had two not-so-public (well, one is public but second-hand) data points converge, and they are consistent with the MSM information on the matter at hand, namely, how Morgan Stanley came to post a $9.4 billion loss on the actions of one trading desk, which in turn led to a first time quarterly loss of $3.85 billion.
The reports in the press are not terribly specific but consistent. For instance, from the Guardian:
The outcome is particularly galling for Morgan Stanley because it spotted early signs of the looming sub-prime mortgage crisis and took steps to hedge its trading positions to protect itself against financial damage.
But poor execution allowed these hedges to fall away – a failure which prompted internal soul-searching and the recent departure of the bank’s co-president, Zoe Cruz.
The Wall Street Journal, in a page one story “Loss Pressures Morgan Stanley CEO,” offers a mere sentence;
Mr. Mack said on the call that the losses resulted from “an error of judgment that occurred on one desk, in our fixed-income area, and a failure to manage that risk appropriately.”
So by all accounts, the losses took place in one trading unit. Reader Steve provided more detail from the conference call:
During the conference call, Mack said the losses were attributable to just one prop desk (and none of the analysts said, `Oh, just like Amaranth, right?’). The position was described as long $14B super senior, short $2B mezz. Wow, nice desk limit (was that you, Ms. Cruz?). And Kelleher actually said they went long the $14B `to cover the cost of negative carry’ on the short position! Priceless.
It will be curious to learn whether anyone can reconcile that disclosure with what I heard tonight from a Morgan Stanley employee. Unfortunately, circumstances were such that I couldn’t grill him further, but he said the losses were in proprietary trading group and resulted from error, pure and simple. Someone made a mistake and put a BBB hedge on an AAA subprime position. Not only did no one notice that the hedge was incorrect, but Kelleher’s comment suggests the long position might have been increased. The FT’s Lex comments are consistent:
A dozen or so traders laid on a position to offset the cost of shorting subprime. Had things worked out, the short could have netted the bank at most about $2bn. Instead, it cost the bank more than $7bn, as the traders’ correct hunch was overwhelmed by a deteriorating long position in top-rated collateralised debt obligation securities.
Mind you, I only got the sketchiest of outlines of what transpired, but the part that was very clear was “mistake” and specifically that someone, somehow misread BBB for AAA (or vice versa). The fact that short mezzanine was described as a hedge for long AAA whas been confirmed in the press. I’m sure real traders who have access to real data can interpolate better.
Then we get to the more interesting question of what sort of managerial failure this represents. It could have originated with bad management information systems, ones that didn’t closely or clearly show how long positions and supposed hedges relate to each other. Per Steve’s comments, it also suggests overly large desk limits, at least relative to how risks were reviewed, This may in turn suggest over-indulgence of big swinging dick proprietary traders. And it also indicates failings in firm-wide risk management processes Even if the trade was put on incorrectly, how in God’s name was it allowed to keep going south until the losses reached $9.4 billion? The cardinal rule of trading is “cut your losses, let your gains run.”
The other question is why did Mack sell a stake to the Chinese? Morgan Stanley will still show a profit for the year, unlike Merrill Lynch, which has given up a whole year of earnings to its writeoffs. If this were a one-off due to an embarrassing screw-up, it doesn’t merit running to get foreign capital. Is this merely storing up reserves for what may be a long, dry period, or is other bad news in the offing?
17 comments
Comments are closed.
Hi Yves,
You ask `why did Mack sell a stake to the Chinese? Morgan Stanley will show a profit for the year…’.
The probable answer is in the 8-K: `While the Firm continues to maintain total capital levels which significantly exceed regulatory capital requirements, at quarter end because of the loss for the quarter and the increase in capital assigned to the Institutional Securities segment, the Firm’s unallocated economic capital was a negative $4.1 billion.’
Economic capital is an internal risk metric. What they (and the RA’s) are worried about is the ability to withstand, at a very high probability, any unexpected losses. So if their target is the usual 99.97%, they are saying they are $4.1B shy of that level. And that implies a (possibly dramatically) lower probability of surviving unexpected losses as defined by their firm-wide risk model. Bottom line is they need more capital to `safely’ cushion their highly volatile P&L, for some management-defined (or RA-defined) value of `safely’.
Steve,
Aha, thanks for doing the spade work. But that begs another question: does MS need more capital because they don’t want to shrink their balance sheet, or it would be too costly in the current environment to reduce their risk exposure? I suspect the latter.
Great observation. I think you’re right.
The problem right now with doing economic capital calculations is the sudden emergence of ugly, unexpected correlations that raise estimates of simultaneous loss probabilities. Previously independent series start moving in lockstep and the VaR model says the cushion is too small. If you decide to shrink, it can be expensive to exit a marginal business quickly, impossible to get rid of a lemon, and foolish to drop a profitable business just because it has a volatile daily P&L. So you `dial down’ your bets and pay whatever it takes for immediate capital…before the well runs dry.
I heard that they had a correlation delta neutral position of Longer super senior/short junior mezz which would tally with the 14long/2short position you report. Sadly for them in a crisis correlation tends to 1 and they were long correlation on the 2/short on the 14, if markets were liquid they could have attempted to dynamically hedge this (a trade they propped around hedge funds, so don’t be surprised if other people have the same trade) with a smaller loss, sadly it wasn’t and they were burned.
Yves,
It seems the longAAA/shortBBB trade was less a mistake than a “soft landing” trade.
Most observers, even into September, believed that only a “Great Depression” event could cause losses for the subprime AAA tranche. They likely put the trade on when the AAA abx was trading at close to par, aiming to pocket the credit spread on that tranche to pay the high cost of credit protection on the BBB.
So basically, the bet was for housing to be bad enough to wipe out BBB but leave AAA unscathed.
Doesn’t seem like a mistake — more of a failure of regression models in predicting six-sigma probabilities. Those same models are still in use to predict Value at Risk all over the street. Notice that Goldman’s VAR has literally exploded in the past two quarters, even with models that understate VAR. I think you’re asking the right question: Is this because they want to increase risk, or because they can’t unwind bets?
David,
The person who spoke to me specifically said the trade executed by mistake, with someone misreading the rating on the original position. Now the error may have gone uncaught for the reasons you mention, that someone looking at it attributed logic to it that wasn’t there.
The demise of the Enron Corporation occurred relatively swiftly. Over a 16-month period, Enron’s stock price declined from a high of $90 to a low of under $1 before the giant energy-trading company declared bankruptcy.
Yves and Steve
Perhaps another aspect of the sale may lie in the fact that should further losses arise, CIC could simply make further injections of capital/increase the stake, after all SWFs tend to have fewer irate shareholders to answer to should they decide to go bargain hunting.
Hi foesskewered,
Sure, SWF’s can act quickly and in most cases their stakes amount to a small percentage of their funds. In the case of CIC further increasing its MS stake, there may be political impediments down the road…or a change in their view of MS. The shareholder issue is on the other side; witness what’s happening with UBS shareholders objecting to the dilutive effects of the recent recapitalization. So far, with MS at least, the dilution hasn’t been reflected in the share price. Odd, but as somebody much smarter than me once said, brains gives you no competitive edge in equities.
The dunce in the class has now raised his hand (me).
aiming to pocket the credit spread on that tranche to pay the high cost of credit protection on the BBB.
I don’t understand why one would buy protection for a short position. Isn’t the whole idea that the BBB tranche will decline?
I guess I need help with hedging-101.
Buying protection on a tranche you do not own is the short. While you’re sitting waiting for the tranche to meet its maker, you are making periodic payments to the protection writer. If you expect that higher rated tranches will be winners, then the net cashflow from a long position in the higher tranches can cover the cost of the credit protection and a portion of your bonus besides. The problem is the ratio: 7 long to 1 short, ugly when the long side blows up.
Steve,
Thanx for the time – I think I get it. I was thinking about a stock short where you borrow and sell shares. This seems more like a put option.
I have a math degree and some quantitative skill but the world of structered finance is not intuitive until it is explained. It would be helpful to discover a blogger like Tanta at CalculatedRisk.com who has authored an “UberNerd” series of tutorials on mortgage finance.
Steve,
Of course, we will never know unless other moles at Morgan Stanley come forward, but I wonder if both viewpoints could be correct, i.e, someone put BBB hedge mistakenly on AAA position, someone later looks at what BBB is costing and decides to increase AAA to offset the outlay.
nc jim,
I have been tempted to try to get smarter about sturctured finance. I’ve read a few geeky papers, but the best sources are books (I had found a series, but seem unable to unearth them now). This text looks promising.
Having said that, if you are like me, you probably want to read 25 to 40 pages, not 200-300 pages.
I think the reason no blogger has yet surfaced is some of the ones who are familiar with the terrain like jck are so deep into it that they find it tedious to bring non-experts up to speed. And any proper primer in this area requires lots of charts….
nc jim,
I hesitate to mention any texts because they are expensive and may not be what you are looking for. That said, a basic overview book is The Salomon Smith Barney Guide to Mortgage-Backed and Asset-Backed Securities by Lakhbir Hayre. But if you’re not already familiar with basic fixed-income concepts this might be a loser for you. There’s never been an incentive to write a popular guide to complex instruments traded in institutional markets. Maybe there is an incentive now that we’re in the middle of a crisis, but that will probably shade off into human interest anecdotes like popular books on Enron and LTCM. (`It was a quiet foggy morning like so many others in London, when Fred Quantman arrived early to his desk to find that…’) It would be terrific if finance had an expositor like Feynman on QED, but I don’t think finance has any Feymans to begin with.
I heard that they had a correlation delta neutral position of Longer super senior/short junior mezz which would tally with the 14long/2short position you report. Sadly for them in a crisis correlation tends to 1 and they were long correlation on the 2/short on the 14, if markets were liquid they could have attempted to dynamically hedge this (a trade they propped around hedge funds, so don’t be surprised if other people have the same trade) with a smaller loss, sadly it wasn’t and they were burned.
Sure, but at that point why not invert the polarity of the deflector shields, redirect the Berthold rays through the Jeffrey’s tube and flood the hedge with nanytes?
as a mortgage trader i can say with the utmost confidence that there is no possible way this trade was an error in the way you have described. it may have been an “error” to use a 7-to-1 hedge ratio, but there is a no way it could have been the result of someone misreading the letters. most mezz tranches of subprime deals were really small. to get up to 2billion you have to do a lot of transactions. this position was carefully constructed.
also, this trade idea is pretty standard, it can be called a “credit curve steepener”. you don’t have to wait for there to be losses to make money, you just need spreads on the lower rated part of the trade to widen out a lot more than the AAA stuff.
Oh, there’s no question the initial position was carefully constructed (well, thoroughly planned).
The question is: given that the 7:1 ratio is initially correct, and knowing that liquidity has been drying up since February (at the latest; that’s when I noticed it, and I’m not on the front lines), why would you ADD to the LONG position? Isn’t the idea of the trade that the short-term negative carry will be paid for in the end??
(As noted in Yves’s initial post, if the negative carry is a problem, you REDUCE the short position AND the overall exposure–you don’t try to maintain a $14,000,000,000 position if you’re worrying about being short 1/7th of that. At least, not if you don’t want to spend a few hours each day talking with progressively senior members of the Credit Risk Department.)