Practical Business Forecasting

by Evans, Michael K.

• ISBN13: 9780631220657
• ISBN10: 0631220658
• Edition: 1st
• Format: Hardcover
• Copyright: 2002-05-06
• Publisher: Wiley-Blackwell

Summary

Stressing the concrete applications of economic forecasting, Practical Business Forecasting is accessible to a wide-range of readers, requiring only a familiarity with basic statistics. The text focuses on the use of models in forecasting, explaining how to build practical forecasting models that produce optimal results. In a clear and detailed format, the text covers estimating and forecasting with single and multi- equation models, univariate time-series modeling, and determining forecasting accuracy. Additionally, case studies throughout the book illustrate how the models are actually estimated and adjusted to generate accurate forecasts. After reading this text, students and readers should have a clearer idea of the reasoning and choices involved in building models, and a deeper foundation in estimating econometric models used in practical business forecasting.

Author Biography

Dr. Michael K. Evans formerly taught at the Kellogg School at Northwestern University. Since 1981 has headed Evans, Carroll & Associates (formerly Evans Economics), and has generated thousands of forecasts at the macroeconomic, financial, industry, and individual company level. He was awarded the Annual Blue Chip Economic Forecasting Award in 1999 for the most accurate macroeconomic forecasts over the past four years

Table of Contents

Acknowledgments	xv
Preface	xvi
Part I	1	(64)
Choosing the Right Type of Forecasting Model	3	(26)
Introduction	3	(1)
Statistics, Econometrics, and Forecasting	4	(1)
The Concept of Forecasting Accuracy: Compared to What?	5	(13)
Structural Shifts in Parameters	8	(1)
Model Misspecification	9	(1)
Missing, Smoothed, Preliminary, or Inaccurate Data	10	(3)
Changing Expectations by Economic Agents	13	(1)
Policy Shifts	14	(1)
Unexpected Changes in Exogenous Variables	15	(2)
Incorrect Assumptions about Exogeneity	17	(1)
Error Buildup in Multi-period Forecasts	17	(1)
Alternative Types of Forecasts	18	(4)
Point or Interval	18	(1)
Absolute or Conditional	19	(1)
Alternative Scenarios Weighed by Probabilities	19	(1)
Asymmetric Gains and Losses	20	(1)
Single-period or Multi-period	21	(1)
Short Run or Long Range	21	(1)
Forecasting Single or Multiple Variables	22	(1)
Some Common Pitfalls in Building Forecasting Equations	22	(7)
Problems and Questions	25	(4)
Useful Tools for Practical Business Forecasting	29	(36)
Introduction	29	(1)
Types and Sources of Data	30	(7)
Time-series, Cross-section, and Panel Data	30	(2)
Basic Sources of US Government Data	32	(2)
Major Sources of International Government Data	34	(1)
Principal Sources of Key Private Sector Data	35	(2)
Collecting Data from the Internet	37	(1)
Forecasting Under Uncertainty	38	(2)
Mean and Variance	40	(2)
Goodness-of-Fit Statistics	42	(4)
Covariance and Correlation Coefficients	42	(1)
Standard Errors and t-ratios	43	(1)
F-ratios and Adjusted R-squared	44	(2)
Using the EViews Statistical Package	46	(3)
Utilizing Graphs and Charts	49	(6)
Checklist Before Analyzing Data	55	(3)
Adjusting for Seasonal Factors	56	(1)
Checking for Outlying Values	56	(2)
Using Logarithms and Elasticities	58	(7)
Problems and Questions	60	(5)
Part II	65	(120)
The General Linear Regression Model	67	(33)
Introduction	67	(1)
The General Linear Model	68	(2)
The Bivariate Case	68	(1)
Desirable Properties of Estimators	69	(1)
Expanding to the Multivariate Case	70	(1)
Uses and Misuses of R2	70	(5)
Differences Between R2 and R2	71	(1)
Pitfalls in Trying to Maximize R2	71	(1)
An Example: The Simple Consumption Function	72	(3)
Measuring and Understanding Partial Correlation	75	(2)
Covariance and the Correlation Matrix	75	(1)
Partial Correlation Coefficients	76	(1)
Pitfalls of Stepwise Regression	77	(1)
Testing and Adjusting for Autocorrelation	77	(7)
Why Autocorrelation Occurs and What it Means	78	(1)
Durbin-Watson Statistic to Measure Autocorrelation	79	(1)
Autocorrelation Adjustments: Cochrane-Orcutt and Hildreth-Lu	80	(1)
Higher-order Autocorrelation	81	(1)
Overstatement of t-ratios when Autocorrelation is Present	81	(1)
Pitfalls of Using the Lagged Dependent Variable	82	(2)
Testing and Adjusting for Heteroscedasticity	84	(4)
Causes of Heteroscedasticity in Cross-section and Time-series Data	84	(1)
Measuring and Testing for Heteroscedasticity	85	(3)
Getting Started: An Example in EViews	88	(12)
Case Study 1: Predicting Retail Sales for Hardware Stores	90	(1)
Case Study 2: German Short-term Interest Rates	91	(2)
Case Study 3: Lumber Prices	93	(3)
Problems and Questions	96	(4)
Additional Topics for Single-equation Regression Models	100	(45)
Introduction	100	(1)
Problems Caused by Multicollinearity	100	(3)
Eliminating or Reducing Spurious Trends	103	(8)
Case Study 4: Demand for Airline Travel	103	(1)
Log-linear Transformation	104	(2)
Percentage First Differences	106	(1)
Ratios	107	(1)
Deviations Around Trends	108	(1)
Weighted Least Squares	109	(1)
Summary and Comparison of Methods	110	(1)
Distributed Lags	111	(6)
General Discussion of Distributed Lags	111	(2)
Polynomial Distributed Lags	113	(3)
General Guidelines for Using PDLs	116	(1)
Treatment of Outliers and Issues of Data Adequacy	117	(5)
Outliers	117	(3)
Missing Observations	120	(1)
General Comments on Data Adequacy	121	(1)
Uses and Misuses of Dummy Variables	122	(5)
Single-event Dummy Variables	124	(1)
Changes in Dummy Variables for Institutional Structure	125	(1)
Changes in Slope Coefficients	126	(1)
Nonlinear Regressions	127	(6)
Log-linear Equations	127	(3)
Quadratic and Other Powers, Including Inverse	130	(1)
Ceilings, Floors, and Kronecker Deltas: Linearizing with Dummy Variables	131	(2)
General Steps for Formulating a Multivariate Regression Equation	133	(12)
Case Study 5: The Consumption Function	134	(3)
Case Study 6: Capital Spending	137	(5)
Problems and Questions	142	(3)
Forecasting with a Single-equation Regression Model	145	(40)
Introduction	145	(1)
Checking for Normally Distributed Residuals	145	(4)
Higher-order Tests for Autocorrelation	146	(2)
Tests for Heteroscedasticity	148	(1)
Testing for Equation Stability and Robustness	149	(10)
Chow Test for Equation Stability	149	(1)
Ramsey RESET Test to Detect Misspecification	150	(2)
Recursive Least Squares- Testing Outside The Sample Period	152	(1)
Additional Comments on Multicollinearity	153	(1)
Case Study 7: Demand for Motor Vehicles	154	(5)
Evaluating Forecast Accuracy	159	(3)
The Effect of Forecasting Errors in the Independent Variables	162	(8)
Case Study 8: Housing Starts	164	(6)
Comparison with Naive Models	170	(5)
Same Level or Percentage Change	170	(4)
Naive Models Using Lagged Values of the Dependent Variables	174	(1)
Buildup of Forecast Error Outside the Sample Period	175	(10)
Increased Distance from the Mean Value	175	(1)
Unknown Values of Independent Variables	176	(1)
Error Buildup in Multi-period Forecasting	177	(1)
Case Study 9: The Yen/Dollar Cross-rate	177	(3)
Problems and Questions	180	(5)
Part III	185	(86)
Elements of Univariate Time-series Methods	187	(38)
Introduction	187	(1)
The Basic Time-series Decomposition Model	188	(5)
Case Study 10: General Merchandise Sales	189	(1)
Identifying the Trend	190	(1)
Measuring the Seasonal Factor	190	(2)
Separating the Cyclical and Irregular Components	192	(1)
Linear and Nonlinear Trends	193	(4)
Methods of Smoothing Data	197	(8)
Arithmetic Moving Averages	198	(2)
Exponential Moving Averages	200	(1)
Holt-Winters Method for Exponential Smoothing	201	(2)
Hodrick-Prescott Filter	203	(2)
Methods of Seasonal Adjustment	205	(20)
Arithmetic and Multiplicative Fixed Weights	208	(1)
Variable Weights	209	(1)
Treatment of Outlying Observations	210	(2)
Seasonal Adjustment Factors with the Census Bureau X-11 Program	212	(3)
Case Study 11: Manufacturing Inventory Stocks for Textile Mill Products	215	(2)
Case Study 12: Seasonally Adjusted Gasoline Prices	217	(4)
Problems and Questions	221	(4)
Univariate Time-series Modeling and Forecasting	225	(46)
Introduction	225	(1)
The Box-Jenkins Approach to Non-structural Models	226	(1)
Estimating ARMA Models	227	(11)
First-order Autoregressive Models- AR(1)	228	(2)
AR(2) Models	230	(1)
AR(N) Models	231	(2)
Moving-average (MA) Models	233	(3)
ARMA Procedures	236	(2)
Stationary and Integrated Series	238	(8)
Identification	246	(3)
Seasonal Factors in ARMA Modeling	249	(5)
Estimation of ARMA Models	254	(1)
Diagnostic Checking and Forecasting	255	(16)
Case Study 13: New Orders for Machine Tools	258	(3)
Case Study 14: Inventory/Sales (I/S) Ratio for SIC 37 (Transportation Equipment)	261	(1)
Case Study 15: Non-farm Payroll Employment	262	(1)
Summary	263	(2)
Problems and Questions	265	(6)
Part IV	271	(130)
Combining Forecasts	273	(38)
Introduction	273	(2)
Outline of the Theory of Forecast Combination	275	(1)
Major Sources of Forecast Error	276	(4)
Combining Methods of Non-structural Estimation	280	(3)
Combinig structural and Non-Structural Methods	283	(7)
Case Study 16: Purchases of Consumer Durables	285	(5)
The Role of Judgment in Forecasting	290	(2)
Surveys of Sentiment and Buying Plans	290	(1)
Sentiment Index for Prospective Home Buyers	291	(1)
The Role of Consensus Forecasts	292	(5)
Case Study 17: Predicting Interest Rates by Combining Structural and Consensus Forecasts	294	(3)
Adjusting Constant Terms and Slope Coefficients	297	(5)
Advantages and Pitfalls of Adjusting the Constant Term	297	(2)
Estimating Shifting Parameters	299	(3)
Combining Forecasts: Summary	302	(9)
Case Study 18: Improving the Forecasting Record for Inflation	303	(3)
Summary	306	(1)
Problems and Questions	306	(5)
Building and Presenting Short-term Sales Forecasting Models	311	(44)
Introduction	311	(1)
Organizing the Sales Forecasting Procedure	312	(3)
Endogenous and Exogenous Variables in Sales Forecasting	315	(4)
Macroeconomic Variables	316	(1)
Variables Controlled by the Firm	316	(2)
Variables Reflecting Competitive Response	318	(1)
The Role of Judgment	319	(5)
Deflecting Excess Optimism	320	(1)
The Importance of Accurate Macroeconomic Forecasts	321	(1)
Assessing Judgmental Inputs	322	(2)
Presenting Sales Forecasts	324	(31)
Purchases of Construction Equipment	326	(7)
Retail Furniture Sales	333	(5)
Case Study 19: The Demand for Bicycles	338	(5)
Case Study 20: New Orders for Machine Tools	343	(3)
Case Study 21: Purchases of Farm Equipment	346	(6)
Problems and Questions	352	(3)
Methods of Long-term Forecasting	355	(46)
Introduction	355	(2)
Non-parametric Methods of Long-term Forecasting	357	(11)
Survey Methods	357	(2)
Analogy and Precursor Methods	359	(2)
Scenario Analysis	361	(3)
Delphi Analysis	364	(4)
Statistical Methods of Determining Nonlinear Trends: Nonlinear Growth and Decline, Logistics, and Saturation Curves	368	(14)
Nonlinear Growth and Decline Curves	368	(1)
Logistics Curves (S-curves)	369	(8)
Saturation Curves	377	(2)
Case Study 22: Growth in E-commerce	379	(3)
Predicting Trends Where Cyclical Influences are Important	382	(4)
Case Study 23: Sales of Personal Computers	382	(4)
Projecting Long-run Trends in Real Growth	386	(4)
Case Study 24: Projecting Long-term Growth Rates in Japan and Korea	387	(3)
Forecasting Very Long-range Trends: Population and Natural Resource Trends	390	(11)
Predicting Long-term Trends in Population Growth	390	(2)
Predicting Long-term Trends in Natural Resource Prices	392	(4)
Problems and Questions	396	(5)
Part V	401	(84)
Simultaneous-equation Models	403	(40)
Introduction	403	(1)
Simultaneity Bias in a Single Equation	404	(5)
Estimating Simultaneous-equation Models	409	(8)
Case Study 25: Submodel for Prices and Wages	409	(8)
Further Issues in Simultaneous-equation Model Forecasting	417	(21)
Case Study 26: Simultaneous Determination of Inflation, Short-term and Long-term Interest Rates, and Stock Prices	418	(15)
Case Study 27: Simultaneous Determination of Industrial Production, Producers Durable Equipment, Inventory Investment, and Imports	433	(5)
Summary	438	(5)
Problems and Questions	439	(4)
Alternative Methods of Macroeconomic Forecasting	443	(42)
Introduction	443	(1)
Structural versus VAR Models	444	(2)
Solving Structural Macroeconomic Models	446	(4)
Outlining the Equilibrium Structure	447	(1)
Newton-Raphson Method and the Gauss-Seidel Algorithm	448	(1)
The Triangular Structure	449	(1)
A Prototype Macroeconomic Model	450	(5)
Summary of Macroeconomic Model Equations	451	(2)
Treatment of Trends and Autocorrelation	453	(2)
Simulating the Model	455	(7)
Preparing the Model for Forecasting	462	(8)
Forecasting with AR (1) Adjustments	462	(1)
Forecasting with Constant Adjustments	463	(1)
Comparison of Alternative Forecasts	463	(7)
Using the Leading Indicators for Macroeconomic Forecasting	470	(4)
Using Indexes of Consumer and Business Sentiment for Forecasting	474	(4)
Conclusion	478	(7)
Problems and Questions	479	(6)
Index	485