Risky Forecasts

By Robert S. Benchley | Summer 2010

Printer Friendly

At Mendoza’s annual CARE conference, the focus was on forecasting extreme events and what it means for financial professionals

Don Kent, one of the country’s first TV weathermen, who died on March 2 at age 92, ended every one of his nearly 40 years’ worth of broadcast prognostications with the same six words: “And that’s my forecast, for now.” The operative words, of course, were “for now.” Kent recognized that meteorological forecasting involved monitoring and analysis of a constantly changing set of variables. At some point, however, you had to make the call and hope you got it right.

That is much the way financial forecasting works, too, says Peter Easton, Notre Dame Alumni Professor of Accountancy and director of the Center for Accounting Research and Education (CARE) at Mendoza. “It’s a key issue for those of us who work on the border of accounting and finance, yet very little research has been done on forecasting profitability.”

The dearth of forecasting research was why Easton and his advisory board saw a benefit in sharing how similar forecasting exercises are performed in very different fields. This year’s CARE conference, held April 9-10 in Coral Gables, Fla., supplemented the lineup of accounting academics and practitioners with outside experts discussing topics as diverse as climate change, human behavior and athletic performance. For the most part, the experts from the fields outside of accountancy and finance spoke of forecasting extreme events in their own fields. The topic, however, dovetailed perfectly with the concerns of financial professionals, who are still wrestling with such issues as the ongoing worldwide financial crisis and whether it could—and should—have been foreseen.

The conference kicked off with a keynote by one of Notre Dame’s own—Keith Sherin (ND ’81), vice chairman and CFO of General Electric—who discussed how the $157 billion (FY 2009) conglomerate, the world’s largest company according to Forbes, goes through its own forecasting exercises. GE has 3,000 people crunching financial data full time and feeding the information, ultimately, to Sherin. It’s a stunningly complex exercise, yet it is overlaid with a philosophical and operational commitment to transparency that means the numbers will be communicated to a wide variety of constituencies. The response to GE’s quarterly earnings estimates—40 to 50 percent of GE’s stock trades clustered around these reports—tends to create “noise” and volatility that has nothing to do with long-term performance, said Sherin.

“Opening the conference with this presentation was extremely valuable,” said Easton, “because it grounded us. It showed us what a very complex corporation is doing in this area—which is often simply the best they can—and helped those of us in academia know if we’re teaching what we should be teaching.”

According to Easton, Sherin’s presentation explicated a widely held misconception among educators that in forecasting, a company seeks to report the good news, not the bad. In reality, the goal is to reveal any information deemed pertinent to investors, which speaks to the pre-eminence placed on thorough, ongoing communication so as not to surprise analysts.

As for forecasting models themselves, the discussion of the models of extreme events from experts outside of accountancy challenged faculty to consider a broader range of possible outcomes. Some highlights include:

• Financial catastrophes, like natural catastrophes, happen quickly and for complex reasons.

George Sugihara, a marine microbiologist at the University of California, San Diego, is credited with being one of the first researchers to apply chaos theory to describe how models used to forecast sardine populations could also be used to predict short-term market fluctuations. Financial analysts tend to expect market change to occur due to a linear series of events, said Sugihara, who worked as a managing director at Deutsche Bank for five years. The truth is that any complex system, be it a fishery or an economy, can fail due to a complex web of non-linear forces acting in unison.

• If our response to hurricanes is any example, we’re not likely to remember the lessons of the current global financial crisis for long.

Despite all our experience with catastrophic weather events, we continue to put ourselves in danger and rack up enormous insured losses. Why do we under-prepare? The reason, said Robert Meyer, a marketing professor at the University of Pennsylvania’s Wharton School, is that either we don’t believe it will happen to us, or we don’t believe it will happen again. This head-in-the-sand approach appears to be hard-wired in humans, said Meyer, and applies as much to investing as to weather. We don’t learn from near-misses, and we under-invest in opportunities that only have long-term benefits. Plus we have selective memories: What erodes with time is not our perception of the likelihood of another event, but the perception of our vulnerability.

• Predicting athletic performance can help predict financial performance.

John Einmahl, a professor of statistics at Tilburg University, told the audience that he had calculated the optimal time at which a man or woman could run 100 meters (9.51 and 10.33 seconds respectively, both faster than current world records)—and that this same exercise could predict everything from what an object might sell for at an auction to the maximum daily negative market dip possible for a specific stock portfolio. Einmahl is a practitioner of a subset of statistics called extreme value theory, which uses extreme deviations from median probability distributions to calculate risk. Other examples he presented included forecasting insurance losses for a hurricane, the performance of the dollar against the euro, and how large a capital buffer a financial institution should maintain.

On the whole, Easton said the discussion of the extreme forecasting models provides academics with new ways to look at old questions. The conference pointed out the need to consider the interdependencies of events. “One thing that has constrained what we have done is that old models assume what is done at one company is independent of another company. We have not thought about the dependencies as a critical part of forecasting,” said Easton.

Other notable speakers at the 2010 CARE Conference included Professor Katherine Schipper, a 1996 Notre Dame Honorary Doctorate Degree recipient, currently at Duke University’s Fuqua School of Business, and Barclays Capital’s Matthew Rothman, who examined the impact of the International Financial Reporting Standards (IFRS) on forecasting models. Former Wall Street analyst Kenneth A. Posner also discussed findings from his recently released book, “Stalking the Black Swan: Research and Decision Making in a World of Extreme Volatility.” Slides and video of the sessions can be found at the CARE Web site.