Forecasting accuracy has been called into question again in 2020. Economists, climate scientists, and epidemiologists (and probably others) have been tarred with the same brush and their expert modelling described as having a history of failure.
The period since mid-2018 has been a challenging one for forecasting accuracy and a review is timely. Volume 1 of my book “Forecasting: the essential skills” was written over the period 2013 to 2018. I found many flaws in the practice of economic forecasting and some holes in the models for forecasting weather and climate. I found that the accuracy of political polling was in decline. There are, of course, some success stories and also improvements in weather forecasting.
Through case studies and reviews of performance I have identified a set of skills which are essential to improving forecasting accuracy.
Volume 2 is now in preparation. Two chapters are now available: a review of recent criticism and an evaluation of forecasting accuracy as perceived by the court of public opinion. A September 2020 survey of the Australian general public measured perceptions of accuracy in the fields of economic, weather, and climate forecasting. Perceptions are reasonably consistent with the reviews I have conducted and not as bad as suggested by the most strident critics – whose views are reflected in a small minority of the general public.
Future chapters will review the poor performance of forecasts over the period 2018 to 2020 and identify reasons for the shortcomings in economic and political forecasts in Australia. There will also be a chapter on the performance of demographic forecasting upon which economic forecasts and infrastructure plans depend.
Volume 1 and the first two chapters of Volume 2 are available at foreseechange.com.au. This is important information for both decision makers who rely on forecasts and forecasters who wish to improve their skills.