Customer Reviews

Non-Linear Time Series Models in Empirical Finance

5 star:		(6)
4 star:		(1)
3 star:		(0)
2 star:		(0)
1 star:		(0)

 

 

If you are interested in what's up nowadays in the finance modeling, you should have this book. It's a review of some of the more recent, important and promising works of the field. Advanced undergraduate students and graduate students will probably understand the book (although I recommend it mostly for people interested in the field). If you want an easy introduction of most of the topics (but pretty older), then, grab Walter Enders book or, the more complicated, but also more complete book of James D. Hamilton. Reading this manual is easy because it's clear and its style is not boring. If you really love finance econometrics, you'll find this book fun to read. The fields covered by the authors are: 1.-Linear models (pretty brief), unit roots, seasonality and aberrant observations; 2.-Regime-switching models for returns such as TAR (Threshold Autoregressive), SETAR,...; 3.-Regime switching models for volatility (and here you'll have the entire family of ARCH models, with its youngest cousins such as GARCH QGARCH, LSTGARCH, VS-GARCH); 4.-Artificial Neural Network for returns. I'm particularly interested in GARCH-type models, and I can tell this part is particularly well done. At the end of the chapter there is a very illuminating empirical comparison between the models. I cannot say if the "artificial neural networks" is a good chapter since I'm not an expert, but the least I can say is that it's pretty understandable (although quite challenging for an ignorant like myself).

The title of this book caught my attention immediately and it actually contains more interesting topics than I thought. After I bought a copy through Amazon and have a closer read, I'm not disapointed by the two authors' writing, which is probably partially based on the second author's PhD dissertation, and so it is a little narrow-focused. But as the authors stated, they want to produce a book which deals with nonlinear techniques as opposed to Mills's mostly linear methods in fiance time series. They have delivered. With hot topics such as regime switching, ARCH models, and neural network applications in finance, I'm sure this book will find a lot of interested readers and will be a key reference in nonlinear empirical finance.

Like his other books, Franses provides an nice applied treatment of non-linear time series models that are in this case applicable to finance. It includes extensive coverage of regime switching models. It includes data drawn from several financial markets including Tokyo, London and Frankfurt.

If you are looking for a book that expands on financial econometrics beyond "The Econometrics of Financial Markets", the dated but otherwise excellent book of Campbell, Lo, and MacKinlay this is an excellent choice.

The premise is the well-known: while models used in empirical finance are linear, the nature of financial data suggests that non-linear models are more appropriate. It is particularly in forecasting and more accurately describing returns and volatility where the non-linear models offer considerable advantages over linear models.

Since there are considerable candidate non-linear time series models available for the modeler or forecaster of economic time series, selecting the right model from the get-go can be difficult. Of course, if you have had good lecture notes from your grad program, you are set. If not, then this book does help you along the way. It is an up to-date guide and provides a rigorous treatment of non-linear models. I like the regime-switching but the artificial neural networks part leaves me cold.

One of the nice things about the book is that it uses a wide range of financial data, from Tokyo, London and Frankfurt.

1. Introduction;

2. Some concepts in Time Series analysis; (Good review of TS stuff)

3. Regime-switching models for returns; (I like this part; explains everything well and easy to follow. Of course, if you are new to the area, this is hard)

4. Regime-Switching models for Volatility; (This is a tough area and they do a good job)

5. Artificial neural networks for returns;

6. Conclusion.

The GAUSS code is available at the authors' website. This is a nice feature, although I do not use GAUSS.

The major distinction of the book from Granger&Terasvirta's earlier work is its focus on financial applications of regime switching (RS) models and the author's separate treatment of RS in returns(means) and volatilities(variances) by putting them in different chapters. Another welcome feature is the availability of accompanying procedures in Gauss downloadable from the author's website. I would have expected a lengthier treatment of Markov RS models but I guess either the authors leave this to Tsay's new book or quote Hamilton as classical reference source.

The major distinction of the book from Granger&Terasvirta's earlier work is its focus on financial applications of regime switching (RS) models and the author's strategy of separate treatment of RS of returns(means) and volatilities(variances) by putting them in different chapters. Another wellcome feature is the availability of accompanying procedures in Gauss downloadable from the author's website. I would have expected a lengthier treatment of Markov RS models but I guess either the authors leave this to Tsay's new book or quote Hamilton as classical reference source.

Just as in his other books, Franses demonstrates that he is a good communicator. Good structure (follows a logic path), well written and good examples.

Is it a good idea to buy this book? Yes, I would say it is mandatory if you are interested in the subject.

Nevertheless, it misses more indepth treatment of non-linear models (precisely what the book is all about). The authors spent too much time on elaborating a comprehensive chapter on linear models when it was sufficient to cite a few references in case the reader wasn't familiar with the required background.

Some demonstrations and explanations were left uncovered which means that you will have to rely on other sources such as Hamilton (1994) to get the whole picture.

This is not a self-teaching guide but one important source in this field.