A0162
Title: Forecasting after the start of a trend break
Authors: David Hendry - University of Oxford (United Kingdom) [presenting]
Jennifer L Castle - Oxford University (United Kingdom)
Jurgen Doornik - Oxford University (United Kingdom)
Abstract: A sequence of large same-sign 1-step-ahead forecast errors occurring as the forecast origin advances signal a sudden location or trend shift. To test which, impulse indicators are included as intercept corrections to offset each forecast error, then replaced after 2 or 3 significant outcomes by a broken linear or log-linear trend, tested against each other and a step indicator. The approach is illustrated by a rapid increase in global mean sea level after 2011 quickly detected, providing accurate forecasts a decade ahead and accurately forecasting the UK's inflation over 2021--24 as compared with Bank of England projections. Simulations and analyses confirm the general feasibility of detecting sudden trend changes after just 2 or 3 periods and then forecasting reasonably accurately.
