The Bank of England's Monetary Policy Committee (MPC) has been overestimating uncertainty and the risks of higher inflation, according to new research by Professor Ken Wallis of the University of Warwick, presented at the Royal Economic Society's Annual Conference on Tuesday 26 March. His analysis of the 'fan charts' used to represent probability forecasts of future inflation lead him to the following conclusion:
"The demons that the MPC thought they saw did not materialise. These are early days yet - the numbers are small, and the MPC is still learning. But these mistakes have costs, and more accurate probabilities would have made earlier reductions in interest rates more likely."
Our economic prospects are uncertain. In many areas of economic activity, it is important to have forecasts of future developments, but these too are uncertain. Forecasters often overlook this, and present forecasts as if they are exact - "inflation next year will be 2.4%", for example - with no guidance as to their likely accuracy.
In 1996, the National Institute of Economic and Social Research (NIESR) and the Bank of England began to publish information about the uncertainty surrounding their forecasts. This was a welcome development. The Bank presented estimates of the complete probability distribution of possible outcomes for future inflation up to two years ahead. Forecast intervals into which inflation might fall with probabilities 10%, 20%, - , 90% were plotted, and since uncertainty increases and the intervals 'fan out' as the forecasts get further into the future, the plot has become known as the fan chart.
The Monetary Policy Committee (MPC) has adopted this practice since its inauguration in 1997. The fan charts give a full account of the MPC's subjective assessment of inflationary pressures, recognising its imprecision.
How good the forecasts are is a question that is asked wherever forecasts are made. It applies to probability forecasts too. If it turned out that it rained on half the days Michael Fish said there was a 50% chance of rain, he would be doing a good job. Ken Wallis's study develops techniques for answering this question for the fan chart forecasts, and applies them to the first three years of the MPC's forecasts.
He finds that the average forecast of inflation one year ahead overestimates inflation by 0.2%. But the novelty of the fan charts is the dispersion, not the average. Here he finds that the fan charts overestimate uncertainty, and overestimate the upside risks to the future outcome. Taking a central interval into which inflation is forecast to fall half the time, he finds that inflation actually fell there two-thirds of the time - the MPC overestimated uncertainty. And inflation never fell above this range, whereas in an accurate probability forecast this would have occurred one quarter of the time.
Notes for Editors: "Chi-squared Tests of Interval and Density Forecasts, and the Bank of England's Fan Charts' by Kenneth F. Wallis was presented at the Royal Economic Society?s 2002 Annual Conference at the University of Warwick.
Wallis is Emeritus Professor of Econometrics, Department of Economics, University of Warwick, Coventry CV4 7AL.
For Further Information:
Before and after the conference: contact Ken Wallis on 024-7652-3055 (direct line: 024-7652-3026; home: 024-7641-4271; email K.F.Wallis@warwick.ac.uk; website: http://www.warwick.ac.uk/fac/soc/Economics/wallis); or RES Media Consultant Romesh Vaitilingam on 0117-983-9770 or 07768-661095 (email: email@example.com).
During the conference (25-27 March 2002): contact Romesh Vaitilingam on 07768-661095 (firstname.lastname@example.org).