Qual., Quant. & the Limits of Disclosure

There are known knowns. There are things we know we know. We also know there are known unknowns. That is to say we know there are some things we do not know. But there are also unknown unknowns, the ones we don't know we don't know.

D. H. Rumsfeld

I like the Globe & Mail. It has such a level, reasonable tone. Last week, it published an interesting article about the Caisse de dépôt et placement du Québec, Quebec's pension fund, investing 8.5% of its assets (thirteen billion dollars!) in a single asset-backed commercial paper deal (ABCP, that's what they call asset-backed securities in Canada).

It is a familiar tale: because it was backed by subprime mortgages, the ABCP investment became unsellable at the same time that the Caisse had to come up with cash to pay margin calls on foreign stock exchange futures. The Caisse ran out of cash and had to sell good assets at fire-sale prices. More than anything, this story reminded me of the bankruptcy of Orange County way back in 1994. For them that don't remember, 1994 was like a mini version of 2008. A number of large, reasonably sophisticated, investors (Orange County, Procter & Gamble, Barings Bank) experienced, as the SEC put it, "significant, and sometimes unexpected, losses in market risk sensitive instruments." (release 33-7250, 1995 WL 774656) In other words: derivatives.

Warren Buffet famously described derivatives as "weapons of financial mass destruction." A small movement in the price of the underlying asset can generate huge losses. As Linda Quinn and Ottilie Jarmel note over-the-counter derivatives are even more sensitive to risk than exchange-traded derivatives. They aren't as liquid because they "may not be settled for long periods." They are subject to scarier margin calls because they are "usually not collateralized." Finally, they have no safety net because they don't "have the benefit of an exchange that serves as a creditworthy counterparty." (DISCLOSURE OF DERIVATIVES TRANSACTIONS AND MARKET RISK OF FINANCIAL INSTRUMENTS, 1065 PLI/Corp 165 (1998))

That's why, in the wake of 1994, the SEC proposed changes to Regulations S-K and S-X designed to improve disclosure about the risk associated with derivatives. The proposing release noted, with characteristic understatement, that "public disclosure about these instruments has emerged as an important issue." In 1996 the SEC extended the PSLRA's forward-looking safe harbor to the proposed rules (33-7280, 1996 WL 163906) and in 1997, the rules became final (33-7386, 1997 WL 39324).

The changes to S-X required more disclosure about derivatives risk management in the "accounting policy" footnote and added item 305, "Quantitative and Qualitative Disclosures about Market Risk," to Regulation S-K. They also require "a discussion of limitations that may keep the quantitative information from fully reflecting net market risk." (DISCLOSURE OF DERIVATIVES TRANSACTIONS AND MARKET RISK OF FINANCIAL INSTRUMENTS, 1065 PLI/Corp 165 (1998))

The SEC rules, of course, aren't designed to protect companies from making bad investment decisions - they're designed to protect investors. These were supposed to give investors a window into the potential downside of opaque financial instruments. But, did this really happen? Was it even possible?

This is where Value at Risk (VaR) comes into the picture. VaR is an accounting tool for predicting the future behavior of a portfolio based on how the portfolio has behaved in the past. As the Globe & Mail article noted, "The statistical models, it turned out, had a fatal flaw. Using data series going back only a few years - a period of low volatility in asset prices - they failed to capture the risk of a market meltdown." The flaw was hardly occult - Lehman Brothers described it this way in their 2007 10-K: "VaR is not intended to capture worst case scenario losses and we could incur losses greater than the VaR amounts reported."

Would that send you running to your broker? Me neither. Maybe it needs a black box around it like the anti-smoking thingy on cigarettes. Predicting the future is, obviously, more art than science. Would it have even been possible to gather enough data about the past to create VaR that was useful? What about for financial instruments that no one understood? What about a housing market that was unlike any that existed before? It seems to me that even with all the available data worked in VaR wouldn't be much more useful than earthquake prediction.

My sister suggests instead of qual vs. quant we borrow a risk tool from rock climbing: subjective risk vs objective risk. Total collapse of housing prices? Objective. Getting your money out in time? Subjective.