Forecasting Non-Stationary Economic Time Series
by Michael P. Clements, David F. Hendry, (2001)

Publisher: MIT Press
ISBN: 0-262-531895
Pages: 392 pages
Price: £19.95 + p&p

Contents

Table of Contents
Book Order Form
Table of Contents

1 Economics Forecasting

1.1 Introduction
1.2 Background
1.3 Forecasting Economic Time Series
1.3.1 Predictability and Forecastability
1.3.2 Assessing forecast accuracy
1.3.3 Time-series properties of the variables
1.3.4 ARCH, asymmetric loss and non-linearity
1.3.5 Simulation methods
1.3.6 Forecasting in cointegrated system
1.3.7 Intercept corrections
1.3.8 A taxonomy of forecast errors
1.3.9 Leading indicators
1.3.10 Forecast combination
1.3.11 Multi-step estimation
1.3.12 Parsimony
1.3.13 Testing predictive failure
1.4 Concepts
1.4.1 Unpredictability
1.4.2 Moments
1.4.3 Forecastability
1.4.4 Implications
1.5 Theoretical Framework
1.5.1 The data generation process
1.5.2 I(0) representation
1.5.3 The model class
1.6 Measuring Forecast Accuracy
1.7 Casuals Information in Economic Forecasting
1.7.1 Models coincides with the mechanism
1.7.2 Model does not coincides with the mechanism
1.8 Conclusion

2 Forecast Failure


2.1 Introduction
2.2 An I(0) Taxonomy of Forecast Errors
2.2.1 Attempts to offset structural breaks
2.3 Forecast Failure in Congruent Models
2.4 A VEqCM Forecast-error Taxonomy
2.5 Equilibrium Correction and Error Connection
2.6 Non-congruent Devices Need Not Fail
2.7 Extended Model Constancy
2.7.1 Is ex-ante non-constancy a fatal flaw?
2.8 Conclusion
2.9 Appendix A: Taxonomy Derivations for Table 2.1
2.10 Appendix B: VeqCM Taxonomy Derivations

3 Deterministic Shifts

 3.1 Introduction
3.2 Deterministic Shifts in a Static Regression
3.3 Deterministic Shifts in I(0) Processes
3.3.1 An I(0) Monte Carlo illustration
3.4 Deterministic Shifts in VEqCMs
3.4.1 An I(1) Monte Carlo illustration
3.5 Unconditional and Conditional Forecast-error Biases
3.6 Forecasting Levels and Growth Rates
3.7 Variance Effects after Structural Breaks
3.8 Higher Frequency Data: Breaks in Seasonals
3.9 Conclusion
3.10 Appendix A: Chow-test Derivation
3.11 Appendix B: Conditional Forecast-error Biases
3.12 Appendix C: Unconditional

4 Other Sources

 4.1 Introduction
4.2 Model Mis-specification
4.2.1 Mis-specification of stochastic components
4.2.2 Deterministic mis-specification
4.2.3 Mis-specification in non-stationary processes
4.2.4 Breaks in mis-specified "casual" models
4.2.5 Monte Carlo evidence in an I(0) process
4.3 Estimation Uncertainty
4.3.1 Scalar autoregressive processes
4.3.2 Non-modeled regressors
4.3.3 Collinearity in conditional equations
4.3.4 Changing collinearity in a VEqCM
4.3.5 Lack of parsimony
4.3.6 An I(0) Monte Carlo illustration
4.3.7 Overfitting
4.3.8 Macro-econometric models and simulation
4.4 Model Mis-specification and Estimation Uncertainty
4.5 Forecast Origin Mis-specification
4.6 Conclusion
4.7 Appendix: Approximating Powers of Estimates

5 Differencing

 5.1 Introduction
5.2 Forecasting Models
5.3 Forecasts-error Biases in DVs and DDVs
5.4 Comparing Unconditional Forecast-error Biases
5.5 Variance Effects after Structural Breaks
5.6 Comparing Unconditional Forecast-error Variances
5.7 Correct Empirical Variances
5.8 Post-transition Forecast Errors
5.9 Conclusion
5.10 Appendix I: A Dynamic DV
5.11 Appendix II: DV and DDV Forecast-error Derivations

6 Intercept Correction

 6.1 Introduction
6.2 The Basics of Intercept Corrections
6.3 Intercept Corrections to VEqCMs
6.3.1 Biases
6.3.2 Variances
6.4 Time-series Intercept Corrections
6.4.1 VEqCM and DV forecast errors
6.4.2 Residual-based intercept correction
6.4.3 Time-series based intercept corrections
6.4.4 Updating
6.5 Conclusion

7 Modeling Consumers'

 7.1 Introduction
7.2 The Data Generation Process
7.3 Forecasting Methods
7.4 Empirical Example
7.5 The Sample Mean as a Predictor
7.6 Differencing
7.7 AR(1) Model: 1-step Estimation
7.8 Interception Corrections
7.9 ARMA Predictors
7.10 AR(1) Model- Multi-step Estimation
7.11 Disequilibrium Adjustment
7.12 Conclusion

8 A Small UK Money Model

8.1 Introduction
8.2 A Four-equation VAR
8.3 Cointegration
8.4 The I(0) System
8.5 A Simultaneous-equation Model
8.5.1 Multi-step forecasts
8.6 Forecast Comparisons
8.8 Empirical Forecast-accuracy Comparison
8.9 Conclusion

9 Co-breaking

9.1 Introduction
9.2 Contemporaneous Co-breaking
9.3 Intertemporal Co-breaking
9.4 Co-breaking in a VEqCM
9.5 Cointegration Co-breaking
9.6 Multiple Shifts
9.7 Conditional Models
9.8 Empirical Co-breaking in YK Money Demand
9.9 Conclusion

10 Modeling Shifts

10.1 Introduction
10.2 Regime-switching Models
10.2.1 MS-AR models
10.2.2 SETAR models
10.3 Empirical Models
10.3.1 SETAR models of US GNP
10.3.2 MS-AR models of US GNP
10.3.3 Testing for several regimes: the MS-AR model
10.4 A Monte Carlo Study
10.5 Analysis of the MS-AR Model Forecast Performance
10.6 Conclusion

11 A Wage-Price Model

11.1 Introduction
11.2 Modeling Wages, Prices, and Unemployment
11.3 Univariate versus Multivariate Methods
11.4 Direction-of-change Measure of Forecast Accuracy
11.5 Qualitative Evaluation of System Forecasts
11.7 Time-series Intercept Corrections
11.8 Conclusion

12 Postscript

12.1 Overview
12.2 Methodological Implications
12.3 The Way Ahead
12.3.1 What Influences deterministic terms?
12.3.2 What cause deterministic terms to change?
12.3.3 How to detect changes in deterministic terms?
12.3.4 How to offset changes in deterministic terms?
12.3.5 Open Systems
12.4 Conclusion

References
Glossary
Author Index
Subject Index