This page contains brief review of large number of time series books (mostly textbooks). For each textbook it discusses its strong and weak points and their suitability for different stages of the learning process.
The books areseparated into four categories:
Textbooks that provide general treatment of time series, touching upon most important topics
Textbooks with focus on forecasting; while the focus is on forecasting, most of these textbooks can be used as (often better) introduction to time series models and time series in general
Topical textbooks, which cover particular time series with focus on particular field, class of models or method
Specialized books, which provide detailed treatment of particular topic in time series
For those who are looking for quick recomendation for introductionary textbook and do not want to read through the whole list, I recomend textbook by Gonzales-Rivera, possibly followed by Enders. For those willing to read more than one textbook, before diving into these two I would start by Deibold, who provides very easy introduction to univariate time series. For reader who is already advanced and withses to dive deeper into the topic, Shumway and Stoffer is a good text.
Probably the most famous time series textbook. And also probably the least suitable as introductory textbook of them all, despite often being recommended to students (including me): you must be either genius or insane (not mutually exclusive, obviously) to recommend this textbook to starting students. The textbook is very exhaustive and very rigorous, but this also makes it hard to read for those who are new to the topic. That said, this is the textbook everybody should know about – once you become serious about doing time series analysis (rather than just modelling) you will want to consult this book.
The best introductory textbook in this list. The books is especially strong in other than univariate topics, such as transfer function models, VARs, cointegration and non-linear models. Nevertheless its coverage of univariate models is still better than most. The book’s value comes from focus on intuition rather than technical exposition, extensive use of simple illustrative examples as well as more complicated real-world examples; all of this leaves you understanding when and why are given models used, and how do they work. Yet, despite not being technical, it still provides the right amount of technical material for the reader to see time series models as mathematical constructs they are.
TBA
TBA
TBA
Probably most famous book dedicated to time series, from two pioneers of modelling time series. It should be stressed that their work and book is not solely focused on economics, which is a serious limitation for using this book as introductory textbook. Still, the book has its undisputable value in providing very detailed, and mostly digestible exposition of ARMA models. It should be consulted by those who have basic knowledge of time series but want to get deeper understanding of (mostly) univariate time series models.
This textbook provides very digestible mix of intuition and theory when presenting standard time series models and methods. From the perspective of modern reader the list of models and sequencing of their exposition is somewhat outdated, but for each type of model (ARMA, unobserved components, …) it provides exposition that is illuminating to beginners and advanced readers alike. Still, I would recommend this textbook as something you read after you read more introductory textbook to complement your knowledge by seeing somewhat different treatment.
This textbook is best thought as complementary to ‘Time series models’ by the same author. It goes into the details of estimation techniques of different econometrical models, including the workings of algorithms and underlying statistical theory. That means that for the question of “what&why happens after I click estimate” it is unparalleled resource. In addition the chapters on multivariate single equation time series models provide very useful exposition of these models.
While being introductory textbook for forecasting rather than time series, this book still manages to be the best intuitive introduction to time series modelling (as opposed to analysis – do not search for it there). Diebold has the unique ability to understand what people who don’t understand are likely not to understand. While it likely cannot serve as the sole textbook for time series course, it should be suggested as introductory reading to students – a book they want to read before they want to get serious studying time series. Major drawback is the limited scope of the book, which covers only univariate models.
TBA
TBA
If you want to learn about multivariate single equation models, this is the book. The exposition is very digestible but at the same time provides sufficient technical detail. Moreover, it includes large number of very detailed examples that help reader understand the material.
In depth treatment of all about VAR models (and partly also other multiequation models). Do not let yourself be fooled by the word "introduction" used in the title, it will not serve well somebody who is relatively new to VAR models. Rather, it should be thought of as a good textbook of people who wish to do research in the field of VAR models, or used them at very advanced level. Correspondigly it includes large number of proposition and their proofs, lot of theoretical examples and much smaller number of practical examples.
This is a great introductory textbook with focus on finance applications. The textbook is on the low end of the technical apparatus and as such it reads well. Moreover, it provides ample illustration of the theory, so that the basic concepts sink in well. Overall, it is recommended for courses that avoid the technicalities to focus on the intuition, but as such it cannot be the last textbook one reads before going out in the real world.
This book is sometimes feels like in-between. In most cases it is too technical for most starting students, but at moments it is able to suitably simplify difficult material – for example it contains the most digestible introduction to Kalamn filter mechanics. It should be recommended as textbook for students that have some basic knowledge of time series models and what to get deeper into the topic with focus on financial time series.
This is an in-depth textbook on structural models and Kalman filter. As such it goes further than probably most readers will want to go. However, the introductory chapters are written with the usual great mix of intuitive and technical approach typical of the author. More than recommended for the start of using Kalman filter.
This is probably the book on unit roots and cointegration, but one should be aware how to use this book. The best way to think about this book is as a textbook for advanced reader on relevant topics; but it will not serve well to beginners. Assuming one is knowledgeable enough then reading this book will be extremely beneficial. An especially good features of the book are (1) inclusion of historical narrative which allows the reader to orient himself in the literature, (2) encyclopedical approach to existing statistical tests combined with audacity to evaluate alternative tests, (3) intuitive introduction to Winer process theory (much more digestible than Hamilton) underlying much of the econometrics of integrated processes.
This is not a textbook, but it is a useful source for some specific topics. It can serve as very good advanced introduction to econometrics of integrated processes, including the unit roots. It has great introduction to error-correction models in its multiple representations, which is useful to anybody interacting with multivariate single equation models. And finally, it provides the reconstruction of academic research on co-integration as it was in 1991, eliminating the need to go into the actual papers.