Futures Shock

By Seth Grimes

The meteoric rise and fall of the Pentagon's Policy Analysis Market (PAM) illustrates the limits modern society imposes on technology. The program featured a Web-mediated futures market on Middle East politics and sought to apply established forecasting techniques to predicting terrorism. But instead of sticking to the dry economic and political data found at hundreds of Web sources, a public site mock-up included assassinations and the like as sample events, inviting the perception that the Defense Department would let traders profit from mayhem. Within days of the program's late-July 2003 outing by two U.S. senators, it was dead and its author sent to the showers.

Yet, every intelligent enterprise and most of us in our private lives employ the Pentagon's techniques, most often to create or protect profits and sometimes for security goals similar to the Pentagon's. But it's one thing to bet on the year-ahead price of a bushel of wheat and another to place a monetary value on the overthrow of the Jordanian monarchy.

Futures, Forensics, and Forecasting

A contract or option to buy or sell a farm commodity at some future time for a price fixed now can assure a food processor's supplies and that a producer will be able to recoup production costs. Futures thus help stabilize commodities markets, and they guide investment decisions by letting producers and consumers forecast supply and demand on the basis of contract price.

Organizations often go a step further, selling and purchasing contracts on the future value of indexes that weigh the contributions of several measures in a given business sector to create artificial, composite financial products whose value is computed using formulas incorporating prices, interest rates, unemployment figures, and so on. These derivative instruments are designed to serve as measures of the health of a business sector or economic system.

In essence, purchasing a contract whose value increases as conditions deteriorate can hedge against disruptive events such as economic downturns that subvert plans. The seller profits by ensuring that the price of the contract accounts for the likelihood that the buyer will want to execute it. Conventional futures markets are motivated by the desire to ensure and secure profitability.

Something anomalous happened at the Chicago Board Options Exchange (CBOE) in the weeks leading up to Sept. 11, 2001. The Chicago Sun-Times reported on Sept. 20 that five days before the attacks, U.S. markets recorded four times more daily average trading of options to sell shares in UAL, the parent of United Airlines, and 16 times the average trading in American Airlines' parent. "Most experts said the wide swings and the timing indicate some trades were made with advance knowledge of the attacks. "The trading 'is so striking that it's hard to attribute it to chance,' said University of Chicago finance professor George Constantinides. 'So something is definitely going on.'"

I believe that forensic analysis of CBOE records led only to dead ends, but it surely contributed to the creation of PAM, which would've let traders buy and sell contracts on specific events and derivative indicators. The site administrators would "make a market" by establishing what could be traded. If you owned a contract on an event that occurred, you'd make a profit that depended on what you paid for the contract.

Although General Motors might hold a contract on the 2005 price of a barrel of oil (a steep increase could hit vehicle sales hard) and lobby the government to enact policies that would affect oil prices, it's in no position to oust a Middle East oligarchy. The U.S. government, making a market for political and economic events, can and does act where private-sector traders can't. The issue isn't whether a PAM bidder can make a bundle on the end of the Syrian occupation of Lebanon, it's whether U.S. policy will be made by markets rather than by rational deliberation, political debate, and constitutional processes. Then there's the ethical question: By supporting trading of assassination futures, is the government condoning or even promoting illegal and immoral tactics?

These questions carry weight because of the personalities involved: PAM's government sponsor was the Defense Advanced Research Projects Agency (DARPA) Futures Markets Applied to Prediction (FutureMAP) program. Admiral John Poindexter was a key figure who has been dogged by outcry against the invasiveness of DARPA's Terrorism [formerly Total] Information Awareness program (TIA). (TIA is a research project aimed at developing technologies to mine vast, distributed data collections for evidence of terrorist activity, but Congress recently suspended funding because of privacy concerns.) Poindexter was a Reagan-administration national security advisor whose criminal convictions for his role in the Iran-Contra scandal and for lying to Congress were overturned on a technicality, so his PAM involvement invited suspicion that much had been left unsaid.

Do Try This at Home

Since retrospective analysis of 2001 United and American options trades foreshadowed the impending attacks, you might ask what value a futures market dedicated to terrorism indicators would offer. A first answer is that a dedicated futures market would aggregate diverse data and indicators in a single interface. PAM would serve the role of an analytic portal with a look and feel that would be similar to the dashboard displays that have now become common in the BI world.

Second, PAM would provide a form of collaborative analytics, collecting and weighing input from diverse expert players. But rather than relying on a deterministic, fixed model, PAM would facilitate dynamic electronic market-based decision support. PAM concepts owe much to Delphi Methods, which were similarly developed to synthesize diverse expert opinions in a quantitative manner that would avoid individual bias. To cite a less esoteric comparison, PAM isn't so dissimilar from a market researcher's focus group.

PAM's data collection would have been no more novel than its techniques. According to the PAM Web site, "The Economist Intelligence Unit [(EIU)] ... continuously assesses and forecasts political, economic, and business conditions in 195 countries" via "a global network of over 500 analysts." EIU "is working ... to collect and process the data on which the securities in PAM are based, and then to assess the value of the securities when they mature."

While I can't know if EIU provided unique services to PAM, I will observe that private customers can obtain similar data and analyses from sources ranging from the World Bank to vendors such as EcoWin.

You can also buy extensive data from DoubleClick — Web statistics in this case. DoubleClick's 1999 proposal to merge supposedly anonymous Web-tracking data with direct-marketing information collected offline set off a scandal. So not even public outrage over data-use policies is new with PAM.

What about PAM's goal of quantifying the importance of indicators that could forecast terrorism and the likelihood of terrorist events? Especially in the wake of the Sept. 11 attacks, every organization knows that it should attempt to quantify and mitigate risk by hedging against disruptive events and preparing disaster-recovery plans. Insuring the lives of key executives isn't considered ghoulish or a veiled death threat — it's standard business practice.

And every significant organization concerns itself with external trends. Witness the popularity of analyst reports that prognosticate about emerging technologies and markets and even assign probabilities to possible outcomes.

Coming Up Short

Although PAM was using common techniques, it lacked explanatory power. For example, a high contract value for a weakening of Saudi religious fundamentalism suggests a likely outcome without illuminating the events that would precede and lead to the end result, that is, without regard for the bidders' reasoning. PAM-based predictions would be opaque, contributing little to the goal of preventing terrorism and lending nothing to the construction of predictive models. Last, it appears that PAM, like most of today's business systems, would have been unable to accommodate unexpected events, which wouldn't have been listed on the market or had the precedents necessary for pricing.

There's a simple conclusion to be drawn from PAM's short life and rapid demise: It's important to identify and assess risk, exploiting the wealth of available data and modern, Web-based technologies that can facilitate aggregation, analysis, and visualization. But for goodness sake, keep your head down.

RESOURCES

"Terrorists Trailed at CBOE," Chicago Sun-Times, Sept. 20 2001

Defense Advanced Research Projects Agency (DARPA): www.darpa.mil

Economist Intelligence Unit: www.eiu.com

EcoWin: www.ecowin.com

The HERO e-Delphi system: www.hero.geog.psu.edu/products/Delphi_white_paper.pdf

World Bank Data and Statistics: www.worldbank.org/data