P B RACTICAL USINESS F ORECASTING Michael K. Evans Blackwell Publishers PRACTICAL BUSINESS FORECASTING P B RACTICAL USINESS F ORECASTING Michael K. Evans Blackwell Publishers © Michael K.Evans 2003 Editorial Offices: 108 Cowley Road,Oxford OX4 1JF,UK Tel:+44 (0)1865 791100 350 Main Street,Malden,MA 02148-5018,USA Tel:+1 781 388 8250 The right of Michael K.Evans to be identified as the Author of this Work has been asserted in accordance with the UK Copyright,Designs and Patents Act 1988. All rights reserved.No part of this publication may be reproduced,stored in a retrieval system,or transmitted,in any form or by any means,electronic,mechanical, photocopying,recording or otherwise,except as permitted by the UK Copyright, Designs and Patents Act 1988,without the prior permission of the publisher. First published 2002 by Blackwell Publishers Ltd,a Blackwell Publishing company Library of Congress Cataloging-in-Publication Data has been applied for ISBN 0-631-22065-8 (hbk) ISBN 0-631-22066-6 (pbk) A catalogue record for this title is available from the British Library. Set in 10.5 in 12.5 pt Plantin by SNP Best-set Typesetter Ltd.,Hong Kong Printed and bound in MPG Books,Bodmin,Cornwall For further information on Blackwell Publishers,visit our website: www.blackwellpublishers.co.uk To Sophie Evans Brill C ONTENTS Acknowledgments xv Preface xvi PART I 1 1. Choosing the Right Type of Forecasting Model 3 Introduction 3 1.1 Statistics, Econometrics, and Forecasting 4 1.2 The Concept of Forecasting Accuracy: Compared to What? 5 1.2.1 Structural Shifts in Parameters 8 1.2.2 Model Misspecification 9 1.2.3 Missing,Smoothed,Preliminary,or Inaccurate Data 10 1.2.4 Changing Expectations by Economic Agents 13 1.2.5 Policy Shifts 14 1.2.6 Unexpected Changes in Exogenous Variables 15 1.2.7 Incorrect Assumptions about Exogeneity 17 1.2.8 Error Buildup in Multi-period Forecasts 17 1.3 Alternative Types of Forecasts 18 1.3.1 Point or Interval 18 1.3.2 Absolute or Conditional 19 1.3.3 Alternative Scenarios Weighed by Probabilities 19 1.3.4 Asymmetric Gains and Losses 20 1.3.5 Single-period or Multi-period 21 1.3.6 Short Run or Long Range 21 1.3.7 Forecasting Single or Multiple Variables 22 1.4 Some Common Pitfalls in Building Forecasting Equations 22 Problems and Questions 25 viii CONTENTS 2. Useful Tools for Practical Business Forecasting 29 Introduction 29 2.1 Types and Sources of Data 30 2.1.1 Time-series,Cross-section,and Panel Data 30 2.1.2 Basic Sources of US Government Data 32 2.1.3 Major Sources of International Government Data 34 2.1.4 Principal Sources of Key Private Sector Data 35 2.2 Collecting Data from the Internet 37 2.3 Forecasting Under Uncertainty 38 2.4 Mean and Variance 40 2.5 Goodness-of-Fit Statistics 42 2.5.1 Covariance and Correlation Coefficients 42 2.5.2 Standard Errors and t-ratios 43 2.5.3 F-ratios and Adjusted R-squared 44 2.6 Using the EViews Statistical Package 46 2.7 Utilizing Graphs and Charts 49 2.8 Checklist Before Analyzing Data 55 2.8.1 Adjusting for Seasonal Factors 56 2.8.2 Checking for Outlying Values 56 2.9 Using Logarithms and Elasticities 58 Problems and Questions 60 PART II 65 3. The General Linear Regression Model 67 Introduction 67 3.1 The General Linear Model 68 3.1.1 The Bivariate Case 68 3.1.2 Desirable Properties of Estimators 69 3.1.3 Expanding to the Multivariate Case 70 3.2 Uses and Misuses of R2 70 3.2.1 Differences Between R2 and R2 71 3.2.2 Pitfalls in Trying to Maximize R2 71 3.2.3 An Example:The Simple Consumption Function 72 3.3 Measuring and Understanding Partial Correlation 75 3.3.1 Covariance and the Correlation Matrix 75 3.3.2 Partial Correlation Coefficients 76 3.3.3 Pitfalls of Stepwise Regression 77 3.4 Testing and Adjusting for Autocorrelation 77 3.4.1 Why Autocorrelation Occurs and What it Means 78 3.4.2 Durbin–Watson Statistic to Measure Autocorrelation 79