Evaluating Inflation Forecasts

Dean Croushore

Professor of Economics and Rigsby Fellow, University of Richmond and Visiting Scholar, Federal Reserve Bank of Philadelphia

June 1998

WP 98-14 – In the early 1980s, economists tested inflation forecasts and found that the forecasts were very bad. Either the surveys didn't capture forecasters' expectations, or forecasters didn't have rational expectations.

However, the sample period being examined consisted mostly of data from the volatile 1970s, when forecasting was extremely difficult. The question is: If we run the same types of tests that were performed 15 years ago on an updated sample, will we find the same problems with the forecasts?

This paper finds that much of the empirical work from 15 years ago does not stand the test of time. The forecast errors from the surveys aren't nearly as bad today as they were in the 1970s. However, some problems remain in the forecasts. It appears to be possible to improve inflation forecasts over some sample periods using bias regressions, and the forecasts don't pass all tests for optimality.

View the Full Working Paper