2
Book reviews / International Journal of Forecasting 19 (2003) 751 761 753 International Marketing Forecasts (2002), London: from a wide range of sources including national Euromonitor Books, 606 pages. ISBN 0 84264-152- statistical offices, international bodies like the UN 2, Paperback, $1250, £625, »1250. and OECD, individual company reports and media coverage, together with the views of independent This book contains tables of detailed forecasts of analysts. Also included in the Introduction are some growth for the markets of over 330 consumer goods broad assumptions which underlie the forecasts. For in 26 non-European countries up to 2006. Forecasts example over the next 14 years, they envisage a of over 100 socio-economic variables for these fundamental shift in market share for consumer countries up to the year 2014 are also included. It is goods in general from Western Europe, North one of a series of books produced at regular intervals America and Japan to South East Asia and (to a by Euromonitor. Other titles include European Mar- lesser extent) Latin America and Eastern Europe. keting Forecasts and World Economic Prospects. Other more specific, but interesting, assumptions are The intention is ‘to produce some broad guidance on that sales of snacks will see their greatest growth in markets and economies to help business planners’. Latin America (due to the US influence), while the The forecasts are divided into two sections. The trend of children growing out of toys at earlier ages socio-economic section covers forecasts in ten main will continue. areas, including demographic and economic in- How useful are these forecasts as a basis for dicators and forecasts of income, earnings, household planning and decision making? This is difficult to characteristics and media access. The level of detail judge. The book does not provide anything more is impressive and the variations in the forecasts than an outline of the underlying forecasting process. intriguing. For example, while 20.6 per 100 house- We have no details of the statistical models or how holds in Japan are expected to possess a piano in the expert analysts interacted with these models. For 2014 the equivalent figure for the USA is only 6.9. example, were their judgmental adjustments formally Similarly, while the number of book sellers and structured (Sanders and Ritzman, 2001, p. 409) and stationers in China is expected to increase by over were efforts made to avoid double counting of 28% from 2000 to 2006 (to 1,023,142), Mexico and effects already partly included in the statistical model Japan are expected to see overall falls over the same (Bunn and Salo, 1996)? That said, most research period of 9.2% and 14.0%, respectively. does suggest that expert judgmental intervention This extensive detail is also evident in the second (even informal intervention) based on domain knowl- section, which contains the forecasts of consumer edge, tends to improve the accuracy of forecasts of markets for each year to 2006, together with tables this type (e.g. McNees, 1990, Donihue, 1993). of international comparisons for 2000 and 2006. A second limitation is that only point forecasts are These are subdivided into 14 main areas ranging presented – forecasts expressed as prediction inter- from food to eyewear. Within these sections there are vals would have allowed an assessment of risk to be forecasts of the size of national markets for in- made (Armstrong, 2001, p. 723), though admittedly dividual products ranging from air extractors and their inclusion would have required an extensive cooker hoods to watches and from automatic de- lengthening of the book. In some cases market sizes tergents to writing instruments. Despite the detail, would appear to be heavily dependent on a specific the tables are clearly presented and easy to find. set of events (e.g. whether or not particular Eastern How were the forecasts derived? A short section European countries will join the European Union) in the Introduction suggests that initial forecasts were and in such cases it might have been possible to obtained from statistical models (which include present alternative forecasts for the different factors like income and price elasticities), but these scenarios. were subsequently modified on the basis of expert Finally, this is the fifth edition of this volume and opinion which took into account qualitative forces, a decision maker using the book would probably like a country’s expected policy orientation. The data want to see some data on the accuracy of past underlying the forecasts was apparently gathered forecasts. In a dynamic world such data would not

International Marketing Forecasts (2002),: London: Euromonitor Books, 606 pages. ISBN 0 84264-152-2, Paperback, $1250, £625, €1250

Embed Size (px)

Citation preview

Book reviews / International Journal of Forecasting 19 (2003) 751–761 753

International Marketing Forecasts (2002), London: from a wide range of sources including nationalEuromonitor Books, 606 pages. ISBN 0 84264-152- statistical offices, international bodies like the UN2, Paperback, $1250, £625,»1250. and OECD, individual company reports and media

coverage, together with the views of independentThis book contains tables of detailed forecasts of analysts. Also included in the Introduction are some

growth for the markets of over 330 consumer goods broad assumptions which underlie the forecasts. Forin 26 non-European countries up to 2006. Forecasts example over the next 14 years, they envisage aof over 100 socio-economic variables for these fundamental shift in market share for consumercountries up to the year 2014 are also included. It is goods in general from Western Europe, Northone of a series of books produced at regular intervals America and Japan to South East Asia and (to aby Euromonitor. Other titles includeEuropean Mar- lesser extent) Latin America and Eastern Europe.keting Forecasts and World Economic Prospects. Other more specific, but interesting, assumptions areThe intention is ‘to produce some broad guidance on that sales of snacks will see their greatest growth inmarkets and economies to help business planners’. Latin America (due to the US influence), while the

The forecasts are divided into two sections. The trend of children growing out of toys at earlier agessocio-economic section covers forecasts in ten main will continue.areas, including demographic and economic in- How useful are these forecasts as a basis fordicators and forecasts of income, earnings, household planning and decision making? This is difficult tocharacteristics and media access. The level of detail judge. The book does not provide anything moreis impressive and the variations in the forecasts than an outline of the underlying forecasting process.intriguing. For example, while 20.6 per 100 house- We have no details of the statistical models or howholds in Japan are expected to possess a piano in the expert analysts interacted with these models. For2014 the equivalent figure for the USA is only 6.9. example, were their judgmental adjustments formallySimilarly, while the number of book sellers and structured (Sanders and Ritzman, 2001, p. 409) andstationers in China is expected to increase by over were efforts made to avoid double counting of28% from 2000 to 2006 (to 1,023,142), Mexico and effects already partly included in the statistical modelJapan are expected to see overall falls over the same (Bunn and Salo, 1996)? That said, most researchperiod of 9.2% and 14.0%, respectively. does suggest that expert judgmental intervention

This extensive detail is also evident in the second (even informal intervention) based on domain knowl-section, which contains the forecasts of consumer edge, tends to improve the accuracy of forecasts ofmarkets for each year to 2006, together with tables this type (e.g. McNees, 1990, Donihue, 1993).of international comparisons for 2000 and 2006. A second limitation is that only point forecasts areThese are subdivided into 14 main areas ranging presented – forecasts expressed as prediction inter-from food to eyewear. Within these sections there are vals would have allowed an assessment of risk to beforecasts of the size of national markets for in- made (Armstrong, 2001, p. 723), though admittedlydividual products ranging from air extractors and their inclusion would have required an extensivecooker hoods to watches and from automatic de- lengthening of the book. In some cases market sizestergents to writing instruments. Despite the detail, would appear to be heavily dependent on a specificthe tables are clearly presented and easy to find. set of events (e.g. whether or not particular Eastern

How were the forecasts derived? A short section European countries will join the European Union)in the Introduction suggests that initial forecasts were and in such cases it might have been possible toobtained from statistical models (which include present alternative forecasts for the differentfactors like income and price elasticities), but these scenarios.were subsequently modified on the basis of expert Finally, this is the fifth edition of this volume andopinion which took into account qualitative forces, a decision maker using the book would probablylike a country’s expected policy orientation. The data want to see some data on the accuracy of pastunderlying the forecasts was apparently gathered forecasts. In a dynamic world such data would not

754 Book reviews / International Journal of Forecasting 19 (2003) 751–761

provide perfect guidance on the reliability of the practical problems first and mathematical niceties,forecasts, but in the absence of a transparent fore- second. In my own experience the latter group oftencasting process, it is unclear how else a decision take the view that: ‘‘ If it works don’t ask why.’’maker could assess the credibility of the forecasts This dilemma pursued me in my Ph.D. yearspresented here. where I faced the two major types of forecasters:

those who will not use a model unless its underpin-nings are theoretically rigorous and well understoodand those who, in the name of forecasting accuracy,R eferencesare prepared to break all, or many, of the theoreticalrules. As a person brought up to love pure mathe-A rmstrong, J. S. (2001).Principles of Forecasting: A Handbook

for Researchers and Practitioners. Boston: Kluwer Academic matics, but who later trained as an engineer, I believePublishers. that I am somewhere in the middle.

B unn, D. W., & Salo, A. A. (1996). Adjustment of forecasts with The author of this book, Chris Chatfield, clearlymodel consistent expectations.International Journal of Fore-

falls into the mathematically rigorous camp. All hiscasting, 12, 163–170.previous four books, including his bestseller, ‘‘TheD onihue, M. R. (1993). Evaluating the role that judgment plays in

forecasting accuracy.Journal of Forecasting, 12, 81–92. Analysis of Time Series: An Introduction’’ (Chat-M cNees, S. K. (1990). The role of judgment in macroeconomic field, 1996) are consistent with this perspective.

forecasting accuracy.International Journal of Forecasting, 6, Indeed, in the light of this, I was surprised when I287–299.

heard that Chris had authored a ‘‘Time-SeriesFore-S anders, N. R. & Ritzman. L. P. (2001). Judgmental adjustment ofcasting’’ book. However, I was relieved when Istatistical forecasts. In: Armstrong, J. S. (Ed.),Principles of

Forecasting: A Handbook for Researcher and Practitioners actually read this book, since – thank goodness – his(pp. 405–416). Boston: Kluwer Academic Publishers. mathematical rigor is still very much to the fore.

This book begins with a summary of basic time-Paul Goodwin series analysis and modelling procedures, followed

The Management School by a fairly brief catalogue of the many differentUniversity of Bath time-series forecasting methods. These range from

UK exponential smoothing, through ARIMA and state-space modelling to multivariate methods. Recent

doi:10.1016/S0169-2070(03)00055-4 arrivals, such as GARCH models, neural networks,and cointegrated models are also covered. The moreimportant methods are compared in terms of theirtheoretical inter-relationships and their practicalmerits including empirical accuracy. This particularchapter, which deals with forecasting competitions,Time-Series Forecasting, Chris Chatfield, Chapmanis very clearly written and tries to answer the key& Hall /CRC, London, 2001, Hardcover, 280 pages.question ‘‘What is the best method of forecasting’’.ISBN: 1-58488-063-5, $74.95.A new set of the ‘‘Ten commandments’’ is thesummary of this chapter and it is a very practicalArguably, there are two kinds of forecasting book:guide for successful business forecasting. The bookTime Series Analysis (TSA) books (such as thealso considers two other general forecasting topicsbooks written by Hamilton (1994) and Box et al.that have been somewhat neglected in the literature,(1994)) and Time Series Forecasting (TSF) booksnamely the computation of prediction intervals and(such as those by Makridakis et al. (1998) andthe effect of model uncertainty on forecast accuracy.Diebold (2000)) – usually these have attractive

Can this book beat the TSF competition andcovers, are relatively easy to read, and are generallybecome a classic ‘‘graduate course’’ textbook? Ifree of mathematical jargon. To some extent thedoubt it. Advanced mathematics, only seven pagesdifferences between these books reflects the distinc-devoted to ad hoc methods and the absence oftion between pure mathematicians and people, likejudgmental forecasting make this monograph un-engineers, who are interested in solving difficult