一本时间序列分析的教材，引文版的~~~PREFACE TO THE SECOND EDITION

Since the publication of the first edition, this book has been used by many researchers and universities worldwide.
I am very grateful for the numerous encouraging letters and comments that I have received from researchers, instructors, and students.
Although the original chapters in the book still form the necessary foundation for time series analysis, many new theories and methods have been developed during the past decade, and the time has come to incorporate these new developments into a more comprehensive view of the field.
In the process of updating this book, I also took the opportunity to clarify certain concepts and correct previous errors.

In time series analysis, we often encounter nonstationary time series, and a formal testing procedure for a unit root has now become the standard routine in time series modeling. To address this procedure, Chapter 9 on unit root tests for both nonseasonal and seasonal models has been added.

Regression analysis is the most commonly used statistical method, and time series data are widely used in regression modeling, particularly in business and economic research. The standard assumptions of uncorrelated errors and constant variance are often violated when time series variables are used in the model.
In a separate new chapter, Chapter 15, I discuss the use of time series variables in regression analysis. In particular, this chapter introduces models with autocorrelated errors and ARCH/GARCH models for heteroscedasticity that are useful in many economic and financial studies.
Although the basic procedures of model building between univariate time series and vector time series are the same, there are some important phenomena unique to vector time series. After an introduction to various vector time series models in Chapter 16, I go on to cover cointegration, partial processes, and equivalent representations of a vector time series model in the new Chapter 17. They are useful in understanding and analyzing relationships of time series variables.

Many time series exhibit characteristics that cannot be described by linear models.
Therefore, I have included Chapter 19 on long memory processes and nonlinear time series models that are useful in describing these long memory and nonlinear phenomena.
To aid understanding, I have also added supplements, Appendix 16A on multivariate linear regression models and Appendix 18A on canonical correlations. In the chapter on aggregation, I include some new results on the effects of aggregation on testing for linearity, normality, and unit roots. In this revision, I follow the fundamental theme of the first edition and balance the emphasis between both theory and applications. Methodologies are introduced with proper theoretical justifications and illustrated with empirical data sets that may be down loaded from the web site: http://www.sbm.temple.edu/~wwei/. As with the first edition, exercise problems are included at the end of each chapter to enhance the reader’s understanding of the subject. The book should be useful for graduate and advanced undergraduate students who have proper backgrounds and are interested in learning the subject. It should also be helpful as a reference for researchers who encounter time series data in their studies.
As indicated in the first edition, the book was developed from a one-year course given in the Department of Statistics at Temple University. Topics of univariate time series analysis from Chapters 1 through 13 were covered during the first semester, and the remaining chapters related to multivariate time series plus supplemental journal articles were discussed in the second semester. With the proper selection of topics, the book can be used for a variety of one- or two-semester courses in time series analysis, model building, and forecasting.
I wish to thank Dr. Olcay Akman of the College of Charleston, Dr. Mukhtar Ali of the University of Kentucky, Dr. H.K. Hsieh of the University of Massachusetts, Dr. Robert Miller of the University of Wisconsin, Dr. Mohsen Pourahamadi of Northern Illinois University, Dr. David Quigg of Bradley University, and Dr. Tom Short of Indiana University of Pennsylvania for their numerous suggestions and comments that have improved this revision.
I am grateful to Ceylan Yozgatligil for her help in preparing some of the updated figures and tables. Finally, I would like to thank Ms. Deirdre Lynch, Senior Editor of Statistics, Addison Wesley for her continuing interest and assistance with this project as well as Ms. Kathleen Manley, Mr. Jim McLaughlin, and the staff at Progressive Publishing Alternatives who provide wonderful assistance in the production of the book.

William W. S. Wei

Department of Statistics

The Fox School of Business and Management

Temple University

Philadelphia, Pennsylvania, USA

April 2005

Contents

Preface

CHAPTER 1
Overview
CHPATER 2
Fundamental Concepts
CHAPTER 3
Stationary Time Series Models
CHAPTER 4
Nonstationary Time Series Models
CHAPTER 5
Forecasting
CHAPTER 6
Model Identification
CHAPTER 7
Parameter Estimation, Diagnostic Checking and Model Selection
CHAPTER 8
Seasonal Time Series Models
CHAPTER 9
Testing for Unit Roots
CHAPTER 10
Intervention Analysis and Outlier Detection
CHAPTER 11
Fourier Analysis
CHAPTER 12
Spectral Theory of Stationary Processes
CHAPTER 13
Estimation of Spectrum
CHAPTER 14
Transfer Function Models
CHAPTER 15
Time Series Regression and GARCH Models
CHAPTER 16
Vector Time Series Models
CHAPTER 17
More on Vector Time Series
CHAPTER 18
State Space Models and the Kalman Filter
CHAPTER 19
Long Memory and Nonlinear Processes
CHAPTER 20
Aggregation and Systematic Sampling in Time Series

References


Appendix
            Time Series Data Used for illustrations
            Statistical Tables


Author Index


Subject Index