Non-Linear Time Series Models in Empirical Finance
Philip Hans Franses and Dick van Dijk
PUBLISHED BY CAMBRIDGE UNIVERSITY PRESS 2000
297PP，PDF，3.39M




Although many of the models commonly used in empirical
finance are linear, the nature of financial data suggests that
nonlinear models are more appropriate for forecasting and
accurately describing returns and volatility. The enormous
number of nonlinear time series models appropriate for modelling
and forecasting economic time series models makes
choosing the best model for a particular application daunting.
This classroom-tested advanced undergraduate and graduate
textbook – the most up-to-date and accessible guide
available – provides a rigorous treatment of recently developed
nonlinear models, including regime-switching models
and artificial neural networks. The focus is on the potential
applicability for describing and forecasting financial asset
returns and their associated volatility. The models are analysed
in detail and are not treated as ‘black boxes’ and are
illustrated using a wide range of financial data, drawn from
sources including the financial markets of Tokyo, London
and Frankfurt.
p h i l i p hans franses is based at Erasmus University,
Rotterdam. He has published widely in journals, and his
books include Time Series Models for Business and Economic
Forecasting (Cambridge University Press, 1998).
dick van d i j k is based at Erasmus University,
Rotterdam. He is the author of several journal articles on
econometrics.

Contents
1 Introduction 1
1.1 Introduction and outline of the book 1
1.2 Typical features of financial time series 5
2 Some concepts in time series analysis 20
2.1 Preliminaries 20
2.2 Empirical specification strategy 27
2.3 Forecasting returns with linear models 44
2.4 Unit roots and seasonality 51
2.5 Aberrant observations 61
3 Regime-switching models for returns 69
3.1 Representation 71
3.2 Estimation 83
3.3 Testing for regime-switching nonlinearity 100
3.4 Diagnostic checking 108
3.5 Forecasting 117
3.6 Impulse response functions 125
3.7 On multivariate regime-switching models 132
4 Regime-switching models for volatility 135
4.1 Representation 136
4.2 Testing for GARCH 157
4.3 Estimation 170
4.4 Diagnostic checking 182
4.5 Forecasting 187
4.6 Impulse response functions 197
4.7 On multivariate GARCH models 200
5 Artificial neural networks for returns 206
5.1 Representation 207
5.2 Estimation 215
5.3 Model evaluation and model selection 222
5.4 Forecasting 234
5.5 ANNs and other regime-switching models 237
5.6 Testing for nonlinearity using ANNs 245
6 Conclusions 251
Bibliography 254
Author index 272
Subject index 277