Supercomputer May Reduce Guesswork in Economic Forecasting
MINNEAPOLIS (AP) _ The Minneapolis area will soon be at the center of an attempt to use one of the world’s fastest supercomputers to dramatically reduce guesswork in economic forecasting.
If successful, the three-year project using a Cray Research Inc. supercomputer could make businesses, consumers and workers less vulnerable to blunders caused by faulty assumptions about the nation’s economy.
In recent years, supercomputers have produced revolutionary solutions to complex problems in engineering, chemistry, physics, biology, meteorology and other disciplines.
Computers have been less successful in economic forecasting, and some economists argue that the problems are so complex that they cannot simply through faster ″number crunching.″
Arthur Rolnick, director of research at the Federal Reserve Bank of Minneapolis, said: ″In the long run, the supercomputer research should mean a smoother-running economy with fewer government mistakes. That’s the objective. That’s the hope.″
The supercomputer project, to be conducted at the University of Minnesota’s Institute for Empirical Macroeconomics, will make Minnesota a magnet for top economic thinkers from around the world.
The institute already has applications from as far away as England and Australia from scholars seeking one of several resident fellowships to work with supercomputers, said David Runkle, the institute’s associate director.
Ten to 15 economic researchers from the Minneapolis Fed and the University of Minnesota will take part in the supercomputer research, a study that will begin sometime this year.
″This is path-breaking work,″ said Daniel Newlon, economics program director at the National Science Foundation, which is reviewing an application to finance a large share of the supercomputer research. The balance of the tab would be picked up by the Minneapolis Fed.
The Minneapolis-based Cray is donating free access to a supercomputer.
Newlon said the project promises to bring together public and private economic forecasters, who make their living predicting the future, with researchers who are on the frontier of using supercomputers to expand economic thinking.
″What could emerge from this is a new generation of policy, analytical and forecasting tools that will dramatically improve our ability to understand and forecast the way the economy performs,″ he said.
In computerized economic forecasting, a computer sifts through millions of pieces of information about the behavior of thousands of businesses and millions of consumers. In this process, the computer tries to simulate changes in the economy in the face of changing circumstances.
Christopher Sims, director of the macroeconomics institute, said the mammoth memory and rapid pace of a supercomputer will vastly increase the number of pieces of information that a computer simulation of the economy could store and manipulate.
If all goes well, Cray and other supercomputer makers could find a whole new market among economists and other social scientists.