Monday, September 22, 2008

Fallacy of Composition In Risk Management

In this post from last Friday, I argued that allowing the worst risk-managed firms on Wall Street to exit the sector was not a bad thing. In that piece, I wrote,

"You may argue, as Zachary Karabell did in yesterday's Wall Street Journal, that this is all due to the Enron-era Sarbanes-Oxley legislation which mandates 'mark to market' of thinly-traded, poorly-understood structured financial securities.

So be it. But the whiz kids at these Wall Street houses, and AIG, all knew the rules, or should have. What they apparently didn't know, in reality, was what could really happen to complex instruments, and the values of their firms, which held so much of this paper, when markets for such instruments simply vanished.

Ironically, we saw the same miscalculations when the same firms used 'portfolio insurance' approaches to risk management in the crash of 1987. It didn't work.

Wall Street is, in truth, littered with past crises and market crashes during which the predominant risk management solutions of the day failed to anticipate the next risky environment.

Part of the reason, of course, is the concept of the 'fallacy of composition.' I first learned of it in Paul Samuelson's classic text, "Economics," in my undergraduate economics courses.

What the typically-young, inexperienced risk management model-builders on Wall Street usually fail to anticipate is that, when a market is composed of firms operating similarly-run risk management functions, then they will all behave similarly in the face of the same price information. What happens next falls under the fallacy of composition. All desks try to sell at once, and prices plummet even further, triggering even larger sales of even more instruments."

This is a topic on which I have seen almost no ink, whether physical or electronic, spilled. The tendency of major financial service trading houses to employ similarly-vintaged and -featured risk management systems virtually guarantees the fallacy of composition with respect to sudden windshears in markets.

When one model sees a need to dump a security, due to excessive risk, they all do. And what the models don't account for is other desks, using similar risk metrics, piling on with similar actions, all of which rapidly accentuate the pricing moves that first triggered the sales of risky instruments.

What people outside the financial services industry fail to understand is that, like most industries, vendors supply products and services- data, software, information, risk management methodologies- to as many industry competitors as possible. Further, there is a constant flow of people back and forth between vendors, consultants and financial services firms. For example, a former investment management business partner and one-time colleague at Oliver, Wyman & Co., recently left a senior position at Mercer Management Consulting, which acquired Oliver Wyman a few years ago, to join some of his former OWC colleagues at a risk management consulting firm in California. This will speed the dissemination of his observations on industry risk management practices, and weaknesses, to another major sector consultant.

Thus, similar to what occurs in many industry sectors, the same cutting-edge technologies for risk management diffuse throughout competitors in the industry faster than one might expect. And, because risk management systems are the backbone governing position sizes and limits, as well as triggering sales of positions, and they interface each other in the markets, market behavior becomes clone-like. Similarly-built and -functioning risk management systems managing risk for similar instruments cause multiple parties to begin making the same moves nearly in unison, which cascades prices, when risk is judged to be excessive, suddenly lower, with seemingly no bottom in sight.

Just how many of these financial services competitors using similar risk management systems, trading the same instruments, do we really need? Not all that many. Not when there are a myriad of hedge funds, private equity trading desks, fund management companies and other private investment groups all buying and selling the same securities as the few, visible, well-known investment and commercial banks.

If anything, these days, save for perhaps Goldman Sachs, the best-paid financial service executives and traders work for privately-owned firms which the average American would not know if he read their names in a newspaper.

And now, with last night's news that Morgan Stanley and Goldman Sachs are becoming Federally chartered commercial bank holding companies, we see the exit from the field of the last publicly-held investment banks which, for so long, relied on proprietary and customer trading for so much of their profits.

With so much asset trading capacity using essentially the same risk management models, there was overcapacity in that activity for at least a decade. And the phenomenon will still exist, as the same risk management models are traveling with exiting firms. Only capital allocations and, perhaps, numbers of traders devoted to the activities will change.

But the herd-like reactions by financial institutions traders and investors to market prices, as guided by similar risk management models and functions, will remain with us for some time to come.

No comments: