High-flying Internet companies, this electronic crystal ball says, might not be poised for the spectacular collapse Silicon Valley doomsayers predict. But some more traditional technology companies, like Motorola, are fraught with risk. The spectacular recovery in Southeast Asia, meanwhile, may be an apparition. Companies there stand within shouting distance of bankruptcy.
These insights are the work of a miraculous computer model designed by technicians at KMV, a financial engineering firm launched a decade ago by a trio of San Francisco entrepreneurs. With this model, KMV lays odds on which companies and banks -- and, by inference, the countries where they are headquartered -- will flourish, and which will perish. In its 10 years, KMV has emerged as one of the hottest comers in the field of numbers-based risk assessment.
In the age of globalization, there is no theology more powerful than risk assessment. Risk is God.
Pension fund managers, currency speculators, banks, insurance companies, corporate financial officers, bond traders, stockbrokers -- even firemen and nursery-school teachers -- are competing in a fevered contest to move billions of dollars around the world profitably. To do that, they need a way of measuring risk. More than ever, they are turning to computer models like KMV's.
In this modern financial age, a shift in investors' ideas about, say, the relative risk of Third World sovereign debt and Japanese stocks can cause billions of dollars to suddenly sweep across continents, sundering some nations while bringing riches to others. Brazil's unexpected decision to devalue its currency last week, for example, caused traders to move money away from Third World countries and other similarly risky investments, roiling financial markets around the globe.
The short-term winners in this frenetic game are those with nerves of steel, lightning reflexes, and a keen appreciation of mob psychology. But the real sharpies are the ones with the best insight into where on the globe -- at any given moment -- the greatest reward is possible with the least risk.
In this game, computer scientists like those at KMV are the gurus of a new age. To handicap the future, KMV crunches an amalgam of raw numbers: 150,000 corporation-years of industry data; cash flow; stock-price swings; and defaults. Sifted and massaged through KMV's computer instructions, this data produces a comforting, seemingly concrete number called a default-probability rating. This number is supposed to predict the future -- Company A is likely to fail; Company B is destined to sail -- so that investors can make decisions. By lumping together companies in a particular country, a nation's destiny can be similarly divined.
Banks have invested heavily in KMV crystal balls. Sixty-five of them pay up to $500,000 per year to get the company's default-probability ratings. These bankers consider KMV's model a financier's version of the Rosetta stone, deciphering the true meaning behind otherwise mysterious financial instruments. The Bank of Montreal, for example, has $2.5 billion worth of corporate debt funds crafted specifically with KMV's model in mind.
"We only buy companies with publicly traded equity in order to use the KMV model," says Barry Campbell, managing director for leveraged debt management at the Bank of Montreal. Using a computer program, instead of a huge staff of analysts and loan officers, Campbell is able to buy and sell the debt of around 400 companies with the help of only four people. It's simple, he says: "We look at KMV's estimated default-probability ratings, and sort things out based on the rating."
Other traders buy corporate bonds that KMV's model says are low-risk, and place bets against bonds chosen from the list of dicey issuers.
KMV Managing Director Tim Kasta won't reveal the names of customers who use the model for arbitrage -- as the exploitation of such price discrepancies is known. But the Bank of Montreal's Campbell says it's being done by more and more KMV customers.
"We're not doing that, but I know others who are," he says. "It's not completely dissimilar from the kind of arbitrage Long Term Capital was doing."
Long Term Capital, of course, is the Nobel Prize winner-packed investment fund that last fall nearly brought down the world financial system. It is the reigning poster child for the pitfalls of computer modeling.
Long Term Capital's most famous players were economists Robert Merton and Myron Scholes, who won the 1997 Nobel Prize for developing complicated mathematical formulas in 1973 that put a price on risk.
The fund specialized in bond arbitrage, using Merton and Scholes' math formulas to make $100 billion worth of bets on certain types of bonds and stocks. But the calculations were based on historical data. They didn't allow for what would happen when the world's financial markets melted down, as they did on Aug. 17, 1998, when Russia devalued the ruble and defaulted on some of its debt.
Risk trumped intellect. The fund collapsed.
To forestall a worldwide financial collapse, the U.S. government pressured New York banks to pony up $3.5 billion to keep the fund from going bankrupt.
Yet Merton and Scholes' 1973 calculations -- particularly those dealing with the wealth of information inherent in a company's stock price -- beat at the heart of computer models such as KMV's. Like Long Term Capital's calculations, KMV bases its projections on historical data, and on Merton and Scholes' ideas about handicapping risk.
Computer models shouldn't be blamed for putting the world at peril of financial disaster, says Cornell University derivatives professor Robert Jarrow. The math is sound, he says. "A crude analogy is fire. If you don't use it right, it burns things down," he says. "If you use it right, you can cook, heat, and provide comfort."