Estimating future uncertainty discussed at the Bank of England
25 July 2016
Speaking at one of the Bank of England’s weekly seminars on monetary and macroeconomics matters, Professor Michael Clements explained how a simple change of approach could improve macro variable forecasting models.
Invited by the Bank of England to deliver a ‘One Bank Research Seminar’, Michael Clements, Professor of Econometrics, used the opportunity to present his research on estimating the uncertainty of macroeconomic forecasts.
The One Bank Research Seminars are one of the ways in which the Bank receives academic input and ideas to inform its own activities. At the seminar on 18 May 2016, Michael spoke about the ways in which data revisions may affect forecasts of macro aggregates, such as output growth and inflation, before discussing ways in which their adverse effects may be mitigated.
Most macro variables are subject to revision after the first estimates have been published, as more detailed information on the state of the economy becomes available. However, the traditional approach to macroeconomic modelling and forecasting effectively ignores such data revisions. Michael presented evidence to suggest that this may be particularly misleading if the focus of interest is in forecasting the uncertainty surrounding the future values of these macro variables.
Based on his research, Michael showed how a simple way of creating forecasts based on early-vintage estimates of the history of the series generated accurate estimates of future uncertainty.