Forecasting: of suckers and seers

Press, 3 July 1985, Page 33

Forecasting: of suckers and seers

By

NEILL BIRSS

Play-acting can be an excellent way of forecasting the future. A world authority on forecasting, at present in Christchurch, has successfully put it to the test. Dr Scott Armstrong, who has a chair in marketing at the Wharton School of the University of Pennsylvania, tested the technique during an American football team’s dispute with its owners. Opinion tests on members of the public suggested there would be one chance in 20 of a strike by the footballers. Dr Armstrong used the technique of role-playing, which had been used by the German military since the 1920 s to predict responses during conflict. Using profiles of the players printed in an issue of “Sports Illustrated,” teams of volunteers acted out the conflict. Repeatedly, the simulation ended in a strike. The professor announced his findings to the press, and won celebrity when the forecast strike occurred. Fascination with forecasting goes back beyond Roman times, when chicken entrails were one source of inspiration. Nowadays, the quest is for more scientific forecasting, and Dr Armstrong and many psychologists have been studying the efficiency of various techniques. Those who prefer to rely on their intuition, and perhaps would dismiss experts such as Dr Armstrong out of hand, are up a blind street. Intuition is not the best basis for forecasting, the experts have found. “I am an empiricist. I look for evidence that intuition will provide better answers than qualitative approaches. The answer is no,” says Dr Armstrong. Some other research results may surprise the gut-feeling brigade less ... Groups of experts are not a lot better than average people. Dr Armstrong, in his textbook, “Long Range Forecasting: from Crystal Ball to Computer,” which is just about to come out in its second edition, wittingly debunks the notion that experts know best. He quotes the chairman of IBM in 1943 forecasting that there will be a world market for about five computers. He also quotes research which establishes that expert opinion about stock-market movement is

no better than that of novices. (Inside information is an exception to this, says Dr Armstrong.) An American researcher examined 255 editorials by one Hamilton, an editor of renown of the “Wall Street Journal” who had a reputation for successful forecasting. In the editorials, Hamilton made 90 forecasts about changes in the market. He was right 45 times, and wrong 45 times. “People following Hamilton’s advice would not have done as well as if they had made random investments,” Dr Armstrong writes. The same researcher found also that a sample of 20 insurance companies did slightly worse than market averages from 1928 to 1931 and that 16 financial services and 24 financial publications did slightly worse than the averages. Other researchers have established that expertise is of little value in forecasting, Dr Armstrong adds. One of his conclusions: Do not hire the best expert you can — or even close to the, best. Hire the cheapest. The professor has a term for public willingness to hire experts who will make forecasts about the future: the seer-sucker theory: “No matter how much evidence exists that seers do not exist, seers will find suckers.” He does like markets as forecasters, however. He says the stock market is seldom wrong about what it collectively forecasts. A free market would not have allowed a debacle such as the Government-run Concorde project, he adds. A misconception about forecasting is that the more information, the better. Who hasn’t heard someone jabber: When you can’t make up your mind that means you need more information — collect it. Paralysis by analysis, is how the president of Chrysler, Mr Lee lacocca once put it. Dr Armstrong’s knowledge of the waste of too much information is first hand. He did research for the Central Intelligence Agency (at the time he knew only the commissioner of the project was a Government department) on whether if a nation had more information on important people it would affect the ability to negotiate. The result of the research was that the information would be unlikely to help.

This does not all mean that there is no way of improving forecasting. One of the best tips is to follow the advice of Benjamin Franklin, Dr Armstrong says. Draw a line down the middle of a page, put the pros for a judgment on one side, and the cons on the other. Weight the reasons if you wish. Then take the sums. “It turns out that this is a verj' good approach,” says Dr Armstrong. Paper and pencil give significant advantages over mental calculations. A person can have only a small number of points of information in his or her head at the one time. The advantages are not only with storage, but with the computing, Dr Armstrong says. Better systems than

Franklin’s are on the way, however. Dr Armstrong has high hopes for expert systems — artificial intelligence computer systems which will help people forecast by sifting facts and make decisions by following rules. He became interested in forecasting when he joined Eastman Kodak in Rochester, New York, as in industrial engineer in 1960. Given a task in improving scheduling in colour printing and processing, he searched for information on forecasting, and became hooked. He returned to university, graduating a Ph.D. in marketing from the Massachusetts Institute of Technology. He helped found, and is a former president of, the International Institute of Forecasters. He also helped

found the “International Journal of Forecasting.” (He has edited a number of other academic journals, too.) Dr Armstrong was chairman of an international symposium on forecasting in 1983 attended by 1100 people; 400 papers were presented. Dr Armstrong is well versed in techniques of qualitative forecasting such as regression analysis and “decomposition,” but his main interest is judgmental forecasting. He will give three public lectures at the University of Canterbury on forecasting. These are free. The first will be in room S 5 in the science block at 1.10 p.m. today. The next two will be in the same room at 1.10 p.m. on July 17 and July 24 respectively.