None of the economic indicators often used to forecast inflation -- such as changes in the unemployment rate or in the price of gold -- do a good job of prediction, a new study suggests.
Researchers examined 19 inflation indicators to determine how well each would have worked at predicting the rate of inflation for 13 time periods between 1985 and 1998. They tested how well these indicators did compared to a method that projected future inflation based simply on recent changes in inflation.
"None of the indicators we tested did very well at prediction," said Stephen Cecchetti, co-author of the study and professor of economics at Ohio State University. "There is good reason to be wary of any forecast of inflation that rests on a single indicator."
Cecchetti, former director of research at the Federal Reserve Bank of New York, conducted the study with Rita Chu, a graduate student at New York University, and Charles Steindel, senior vice president in the Research and Market Analysis Group at the Federal Reserve Bank of New York. Their results appeared in the April issue of Current Issues in Economics and Finance, published by the Federal Reserve Bank of New York.
Accurate forecasts are vital to the Federal Reserve's Federal Open Market Committee (FOMC) when it considers what steps it should take to control inflation, Cecchetti said. The FOMC's decisions on interest rates, the primary tool to influence inflation, has a huge impact, influencing everything from car loans to corporate planning. "If you're going to stabilize inflation, you have to know where it's heading. That's why inflation forecasts are so important," he said.
Some of the indicators tested included commodity prices, such as oil and precious metals, financial indicators, such as exchange rates, as well as unemployment rates, average hourly earnings and others.
Cecchetti said that while some indicators did better than others, none was very useful. In fact, 10 of the 19 indicators were consistently worse at forecasting inflation than were projections based on past inflation numbers -- that is, the indicators underperformed the projections in more than half of the 13 time periods.
One of the best indicators was the Journal of Commerce price index for industrial materials, which outperformed the standard projection nine out of 13 times. However, the performance was wildly erratic: in one time period, it produced the worst forecast of all 19 indicators. "Even the best indicators were not very consistent," Cecchetti said. "You can't put much stock into an indicator if it sometimes leads you far astray."
It is possible that some combination of these indicators would provide more consistently useful forecasts, but Cecchetti said such a combination may simply combine the faults of the individual indicators.
For now, policymakers and other forecasters need to rely on a variety of economic indicators and their own judgment to arrive at useful predictions, Cecchetti said. "Clearly, no one should be relying on a single indicator to predict inflation."
Written by Jeff Grabmeier, 614-292-8457; Grabmeier.1@osu.edu