Although most people are not aware of it, forecasting plays a major role in everyday life. From choosing what clothes to wear, to deciding what time to leave for work, we are making a forecast as to what the weather or traffic conditions might be. However, forecasting has a much bigger part to play in helping to better understand and prepare for the challenges that tomorrow will inevitably bring.
Up to now, most writings on the matter are very technical and inaccessible to all but the specialist reader. Jennifer L. Castle, Michael P. Clements and David F. Hendry’s new book, Forecasting: An Essential Introduction, goes against this trend by presenting the subject in a simplified way that is accessible for a wider audience. It is unique for omitting mathematics and technical descriptions, while retaining the subtleties and complexities of the concepts which are key to truly understanding forecasting.
The Motorist and the Economist
The book’s original approach relates forecasting in economics to an example of a motorist trying to forecast her journey time in the face of a myriad of problems. What might otherwise be dismissed as abstract ideas and concepts are related to the everyday and the commonplace. Everyone will at some time or another have deliberated about when they need to leave home for a rendezvous with a friend, or to catch a flight, and so on. Many of the forecasts that people make in their daily lives have parallels in economic forecasting, and the book focuses on real world examples to help the reader better understand the more abstract scientific approach.
The Key Issues
This non-technical explanation of forecasting is coupled with a full and frank discussion of the key issues:
- why do forecasts sometimes go so badly awry and what can be done about it?
- what does it even mean for a forecast to be bad?
- how should readers think about forecast uncertainty?
Two controversial aspects are:
- showing that having a ‘correct model’ of the economy (or the motorist having an accurate map) is not necessary for successful forecasting
- a `bad forecast’ might instead reflect a poor understanding of how well anyone could possibly forecast in the given situation by failing to grasp the inherent uncertainties involved.
Bad Forecasts
Following the 2007 financial crash, The Queen asked why no one saw it coming. How could forecasters have got it so wrong? A key idea in the book is that forecasts will fail badly when the world changes to such an extent that the past is no longer a good guide to the future. This is explored through graphs with figures of shifted distributions, and historical examples of real income growth to name a few. The book considers ways of minimising the deleterious effects of such shifts when they occur, using examples from macroeconomic forecasting as well as climate science, illustrated by the motoring analogy. The examples explain quite difficult concepts in a very easy to understand, and hence accessible, way.
How Can Forecast Uncertainty be Best Communicated?
Like the poor motorist attempting to forecast her arrival time in response to unforeseen problems, forecasters also find themselves having to update their forecasts in light of new information. The `homely’ setting of an everyday activity thereby provides an intuitive understanding. For example, how should one judge forecast accuracy when the costs of over and under-forecasting are very different? Under-estimating the amount of fuel required to safely return a space rocket and crew to the earth after a lunar mission is vastly more costly than landing with some unspent fuel. Or a motorist may wish to guard against the possibility that she arrives too late at the airport for a flight by deliberately over-estimating her expected journey time. Thus, biased forecasts can be sensible.
So how can forecast uncertainty be best communicated? The authors argue the most effective way is through the use of models, which allow for a range of likely outcomes. A forecast might indicate that the economy will grow by a certain percentage, but there could be a 90% chance (say) that the final figure will be lower or higher than the original forecast. By allowing for uncertainty, both the accuracy of the forecast, along with the reported uncertainty, can be evaluated.
Forecasting: An Essential Introduction
With this thorough coverage of the topic, the reader will be able to fully engage with the challenges faced by forecasters and how these can be overcome. Thus, the book does indeed provide an essential introduction to forecasting.
Dr Jennifer L. Castle is a Tutorial Fellow in Economics at Magdalen College, Oxford University, and a Research Fellow at the Institute for New Economic Thinking at the Oxford Martin School. Michael P. Clements is Professor of Econometrics at the ICMA Centre, Henley Business School, University of Reading, and is an Associate Member of the Institute for New Economic Thinking at the Oxford Martin School. Sir David F. Hendry, Kt is a Senior Research Fellow of Nuffield College, Oxford University, and co-director of the Program in Economic Modeling at the Institute for New Economic Thinking and of Climate Econometrics, both at the Oxford Martin School.
