Friday, November 16, 2007

ABCP Write Downs By Banks Follow A Pattern

Statistics 101

Thursday, November 15, 2007 Harry Koza


Print this article

TORONTO (GlobeinvestorGOLD) - Let’s brush up on Statistics 101.

Market rocket scientists and quants use the term sigma (geek-speak for standard deviation) to measure how likely an event is to occur. Using the normal bell curve so beloved of university statistics professors, about 68 per cent of events occur within 1-sigma of the mean, or plus or minus one standard deviation. Let’s say that our event is a one-day move in the stock market. Thus, 68 per cent of the time, the daily move of, say, the S&P/TSX composite index, will be within one standard deviation. You can look up the mathematical definition of standard deviation, but for our purposes, suffice to say it measures the volatility of results.

About 95 per cent of daily market moves would fall within two standard deviations of the mean. A 3-sigma range encompasses about 99 per cent of all daily moves, 4-sigma about 99.99 per cent, and 5-sigma, 99.99994 per cent. So, the odds of a one-day market move outside of a 5-sigma range happening is about as likely as you being abducted by aliens.

And a 10-sigma event, well, that would be about as likely as the nutbars who genuinely believe they have been abducted by aliens (and there are a lot of them out there, Mulder) summoning the Mother-ship by holding hands and humming Duelling Banjos.”

Yet, oddly enough, improbable market events seem to happen with alarming regularity.

The Crash of ’87 was a 20-sigma event, or one that only happens once every 100,000 years or so. Two years later, in 1989, the Dow Jones industrial average dropped 8 per cent in one day, a 7-sigma event, and in 2003, natural gas prices shot up 42 per cent in one day, a 12-sigma event.

Years ago, someone told me that when they design offshore oil rigs, they build them to withstand a 100-year wave, one so big that it only occurs once every hundred years. Then they tow the rig out into the North Atlantic, and it gets hit by half a dozen hundred-year waves the first week.

We’re seeing a similar thing in the credit markets these days. Subprime mortgages, the securitizations and collateralized debt obligations (CDOs) they’ve been packaged into, the special investment vehicles (SIVs) that held those securitizations, the investment banks that created those SIVs – they’re all getting whacked. Banks and dealers are taking big writedowns almost every day. A billion here, a few more billion there – next thing you know you’re talking real money.

No mea culpa from the wizards of Wall Street, though. They just shrugged, threw up their hands and explained, as did Goldman Sachs, “We’ve just been hit by a 25-sigma event.” Who could have predicted that?

The Wall Street wizards used sophisticated mathematical models to manage the risk of their creations, and the models, as it turns out, were useless. Models used to price the risk of subprime mortgage securitizations, for instance, were based on assumptions about default risk that were based on historical data, data from the days when people actually needed a down payment, a job, and a good credit history in order to get a mortgage loan.

Then, when they diced and sliced the subprime loans into tranches of variously rated CDO paper, they assumed that the models used to price these tranches were immune to the vagaries of the real world, where fear and greed, not equations, rule.

And so, as defaults and foreclosures rose, and spreads widened, investors suddenly woke up to the fact that the models were flawed, and started wondering just what was in those securitizations they owned. Those who already owned the stuff – whether CDO, asset-backed commercial paper (ABCP) or one of the many other sporty variants doesn’t really matter, they’re all equally sweaty these days – started looking for a bid. Alas, those who didn’t already own the stuff weren’t looking to buy a pig in a poke, not at the asking price, anyway, so buyers and sellers all of a sudden could not get together and transactions didn’t happen. Liquidity vanished.

The ABCP market froze up. The CDO market dried up. Chief executives at big Wall Street firms were suddenly washed up.

Still, every time a bank or investment dealer announces a big writeoff related to their CDO or subprime or SIV exposure, their stock goes up, seemingly on the notion that all the bad news is out now, so it must be time to buy. I dunno, though. I have a feeling that this whole credit crunch thing is just getting rolling, and is going to get a lot uglier before it is through, and there will be plenty more big writedowns in the months ahead.

This week, General Electric announced that investors in its $5-billion (U.S.) Asset Management Enhanced Cash Trust would only be able to get their money out at 96 cents on the dollar. Surprise, what everyone thought was a safe, boring investment yielding a few beeps more than Treasury bills turns out to have been (as of June 30) one-third invested in home-equity securitizations and a quarter invested in residential mortgage-backed securities. Gee, what were the odds of that happening?




Copyright © 2007 CTVglobemedia Publishing

Search The Web