Converging risk, part II: the human element. (End Analysis).
There are a number of reasons for this, the most prominent being a string of embarrassing disasters that served as reminders for financial institutions that operational risk management is just as important as financial risk management. (See "Earning from Mistakes," RM, October 2000). Modelers are also starting to accept that forecasting the behavior of financial markets and companies is not like forecasting the weather: The weather does not change because its audience hopes it will; the same cannot be said of the financial markets.
Modeling the Human Mind
In finance, individual human actions dissolve into statistical insignificance much of the time; individual investors usually cannot make much difference to the direction of the stock market; and there is no serious threat to a bank if an individual borrower runs out on a debt. But collectively, humans can change the nature of the financial game dramatically; and sometimes, as illustrated most terribly by the events of September 11, even the actions of a few individuals can make an enormous, and enormously unpredictable, difference.
"When you propose a financial model, you're pretending you can guess another person's mind," observes Emanuel Derman, a quantitative researcher at Goldman Sachs, in a January 2001 working paper. That guesswork can fail, and so, therefore, do the model's predictions.
The most infamous example is the summer 1998 demise of Long Term Capital Management (LTCM), the hedge fund that was founded by legendary Salomon Brothers trader John Meriweather, with Nobel Laureates Myron Scholes and Robert Merton on its payroll.
LTCM represented the zenith of models-based finance, using state-of-the-art analytical tools to find valuable and supposedly low-risk trading opportunities. But it came undone in spectacular style when a succession of market upsets led to rapid decline in its huge, highly leveraged portfolio. So large and networked was the fund that its collapse brought the global financial system to the brink of disaster before a consortium of its creditors bailed it out.
Not only was LTCM's demise prompted by one unpredictable human action--Russia's largely political decision to default on some of its debt--it was also exacerbated by another--the herding behavior of investors. In a mass exodus, investors fled any even vaguely troubled market, causing extreme price movements. What is more, critics have since made controversial claims that the widespread use of quantitative, risk models actually made market conditions worse in 1998 by sending independent, but simultaneous, "sell" signals to many investors at the same time.
People Control the Model
It could be argued that it was not the models themselves that were at fault, but their application. In fact, similar arguments have been made for many other model-related fiascos. Models are most effective when applied very narrowly, and with very keen attention paid to changes in circumstance that might push them beyond their limits.
When faced with an inviting opportunity or a dauntingly complex problem, however, it is frequently hard to resist the temptation to stretch models. Another response is to hope that the problem has been misstated, and that integrating what is known of market behavior and human behavior will develop better theories that are harder to outguess. This "behavioral finance" may help to explain such phenomena as the dot-com bubble and the 1998 liquidity crisis.
But behavioral finance is in its infancy, and while it has had its academic triumphs--with a couple of Nobel Laureates to its credit--its practical application remains only a distant hope.
Next month's column will look at manually reconciling intuition with mathematical models and pricing theories.
Sumit Paul-Choudhury is editor in chief at ERisk.
Sumit Paul-Choudhury ("End Analysis," p. 48) is a frequent contributor to RM and editor in chief at ERisk in London.
|Printer friendly Cite/link Email Feedback|
|Comment:||Converging risk, part II: the human element. (End Analysis).(Brief Article)|
|Article Type:||Brief Article|
|Date:||May 1, 2002|
|Previous Article:||Conferences. (Datebook).|
|Next Article:||For sweeping liabilities. (Coverage).|