TSM v4.40 is a comprehensive package for linear and nonlinear time series modelling. It will estimate and forecast ARIMA and ARFIMA models, several GARCH, FIGARCH, APARCH and EGARCH variants, bilinear models, Markov-switching and smooth transition models. Most model features can be freely combined, and a flexible and intuitive formula-coding feature allows virtually any model to be set up. Alternatively, user-supplied Ox code can be integrated with the other features of the package. Dynamic equations systems can be specified and estimated easily, options including VARs, simultaneous systems, error correction systems, multivariate GARCH and regime switching. Panel data models can be estimated by least squares and GMM with either fixed or random effects.
A comprehensive set of diagnostic tools is implemented, including Q tests, LM tests, moment tests and parameter stability tests. Other features include nonparametric regression, log-periodogram regression, recursive and rolling estimation, automated model selection, Monte Carlo forecasting of general nonlinear time series models, and parametric bootstrap tests and confidence intervals. Cointegration analysis options include Johansen's rank tests, and MINIMAL analysis of restrictions on the cointegrating space. Any model that can be estimated can also be stochastically simulated, and Monte Carlo experiments can be performed with any combination of models for simulation and estimation. Data can be input from suitably formatted text files and spreadsheet formats such as .xls, .csv, .wks or .in7 (OxMetrics format), and transformed and edited interactively.
The program comes with a platform-independent graphical user interface (GUI) — see the screenshots — and comprehensive graphics capabilities.It has special features designed to optimize the package for teaching, such as a simplified interface option, and the ability to distribute data and model specifications for classroom exercises in a single settings file - see the sample class exercises now available for download. Alternatively, TSM can be called from a user's Ox program and return the results to the console or a text file.
Compared to the leading commercial econometrics packages, TSM is a bargain. It is very easy to use, has all the facilities one would expect of a standard regression package, yet also offers a unique range of time series models, techniques and special features.
The current release of TSM (v4.40 or later) runs under Jurgen Doornik's Ox 7 programming system in Microsoft Windows (95, 98, Me, XP, Vista, 7 and 8) ; see http://www.doornik.com/. Earlier versions require Ox 6.21.
Knowledge of the Ox language is not a requirement, and its function is transparent to the user in normal use. Ox Console is freeware for academic teaching and research purposes, but not to non-academic users — in particular, businesses and government agencies. A TSM licence will be supplied to such users only on the explicit understanding that it will be used with Ox Professional , not the free console version. Purchasing information for Ox is obtainable from Timberlake Consultants Ltd at http://www.timberlake.co.uk/.
The 64 bit version of TSM runs with Ox 7 Professional on 64-bit Windows 7 or Windows 8 computers. Ox Professional is not freeware but can be purchased from Timberlake Consultants, see above.
The Windows graphical user interfaces are provided by the Java Runtime Environment (JRE), which must be separately installed on the user's system (32 or 64-bit, as appropriate). The JRE is a free download from Sun Microsystems at http://java.com/. Even if your system has Java pre-installed, it's recommended to upgrade to the latest version.
TSM uses the public domain package Gnuplot to supply graphical capabilities. Although a compact older version of the Gnuplot executable is bundled with TSM (32-bit version only), install the Gnuplot package separately from http://www.gnuplot.info to enjoy the latest features. If found, this will be used by TSM automatically.
Install Ox and the JRE first. Then simply run the setup file tsmod4.??-??-??-??_setup.exe to create directories, copy files and set up the shortcuts needed to run the program. Here, ??-??-?? denotes the version number. Uninstall TSM using the "Add-Remove Programs" applet in the Windows control panel, or the Start Menu shortcut.
Note: no Linux version is available for Version 4.39 and later. Use a Windows emulator such as Wine or Crossover to install TSM in Linux or OS-X.
TSM can be downloaded freely to run in demonstration mode, with a restricted capacity (number of observations x number of parameters < 500). A personal licence code to unlock the fully functional version can be purchased at the price of £65 (GBP). A personal licence entitles the holder to run the fully functional software, including all future upgrades of TSM Version 4, on one or more machines. The name of the licence holder is displayed in the results window at start-up.
Click the Buy Now button to make a PayPal or credit card payment. At the same time, please send me an email to let me know to expect your payment. When payment is received, a personal licence code will be sent to the email address you supply. Your order will be processed as quickly as possible, which in most cases means within 24 hours.
By default, the registered licence holder's name will be the name appearing in the PayPal billing details. To register a different name (which must be the person's full name, not a nickname or initials) the email accompanying your order should give the purchaser's name and desired registered name.
PayPal accounts cannot be created from some countries. In this case contact me to arrange another method e.g. electronic bank transfer or personal cheque.
An institutional multi-user licence for network or classroom use (minimum 10 users) can be purchased for £65 plus £20 for each additional simultaneous user. The name of the department/institution, and the usage limit, are displayed at start-up.
An institutional limited term licence permits any number of users for periods of six months (price £100) or one year (price £160). This licencing mode allows students to run the software on their personal computers. The name of the department/institution and the expiry date are displayed at start-up. The program reverts to demonstration mode after the expiry date.
* Limited term licences are offered to permanent site licence holders at a 60% discount (i.e. £40 for 6 months and £64 for 1 year).
* Limited term licence-holders purchasing a permanent multi-user licence within 3 months of expiry receive a refund of 90% of the original purchase price. Please contact me for further information about site licences.
TSM is still under constant development. New features and refinements to the interface are added regularly, and bugs are fixed as soon as they are discovered or reported. This means that upgrades are posted to this site quite frequently. The names of download files include the version number, so it is easy to check if you are up to date. Once you have installed the main package, upgrading is a quick and easy process. Under Windows, simply run install without uninstalling. Your folder specifications will then be retained. Previous versions remain posted in case of any teething problem with the latest release.
OxJapi (Java application programming interface for Ox) is the system used to create the TSM user interface. All the coding is done in the Ox language, which makes this an especially attractive way of creating free-standing applications using the Ox statistical engine and matrix computing capabilities. Viewed as an alternative to the 'Modelbase' class distributed with OxMetrics, OxJapi will require somewhat more programming effort by the user to achieve comparable results, but then allows much more flexibility in GUI design. TSM is, hopefully, a good illustration of what can be done with it.
The original JAPI due to Merten Joost, designed to interface with C, Fortran and other scientific languages, was ported to Ox by Christine Choirat and Rafaello Seri. The present version, OxJapi 2 for Ox 7 developed by James Davidson and Tim Miller, works with Java Swing and has a number of refinements and enhancements of Joost's original design, including tables and drag-and-drop capabilities.
Click here to download OxJapi 2 for 32-bit and 64-bit Windows. The zip file includes the .dll and header files, an OxJapi class containing some basic functions developed for TSM, documentation, and a range of sample programs.
TSM Discussion List ...
To sign up to the new list, go to http://www.jiscmail.ac.uk/. The list is called TSM-USERS.
The file TSM_Exercises.zip contains a set of exported ".tsm" settings files and accompanying exercise sheets. These materials illustrate how TSM can be set up for use in the classroom. Data, models, and preferred program settings can be distributed to students in a single file. Simply double-clicking on the file in Windows Explorer starts TSM and creates the data file and ancillary files in the student's work space automatically. The examples include regression exercises with the CAPM, and simple Monte Carlo experiments to illustrate econometric concepts such as bias, efficiency and test size and power. Download the file here.
Supplementary Coded Tests
TSM now features improved methods for running user-coded Ox modules. making it easy for researchers to share experimental code with other TSM users. Simply load a specially exported settings file into TSM to un-bundle and load the code. The posting below demonstrates the new technology.
The TSM settings file Subsample_CointegrationTests.tsm contains bundled Ox code for the tests described in the paper "Tests for cointegration with structural breaks based on subsamples" by James Davidson and Andrea Monticini, Computational Statistics and Data Analysis 54,11 (2010) 2498-2511. Models to run the tests, and also to generate cointegrated data with breaks for use in simulation experiments are preinstalled. For programmers who like to do it themselves and can provide their own code for the basic test statistics, the Ox code file itself is here.
A Markov Switching
This TSM settings file contains Hamilton's (Econometrica 1989) model and data set. Drag this file into TSM and click the "Run" button to estimate Hamilton's model. Check out the graphics, especially the smoothed regime probabilities for comparison with the NBER recession indicator series, which is included with the data set. Download hamilton.tsm here.
Data Format: The Grunfeld Data
Here are two files to illustrate the TSM format for panel data sets. They contain the well-known Grunfeld investment data set; check out http://mpra.ub.uni-muenchen.de/20841/1/Grunfeld.pdf for the background.
The data themselves (10 US companies over 20 years, 1935-54) are contained in grunfeld_panel.xls. Since there is no way to store the mnemonics for the firms in the Ox spreadsheet format (only row 1 can contain text) these are included in the text file grunfeld_panel.txt (in general, a file with the same name as the Excel file but with ".txt" extension). Copy both these files into your data directory and load the data in the usual way. Loading suitably formatted data automatically puts TSM into panel data mode. A file with the individual names will be used if it exists, otherwise the individuals will be identified by numbers in the output.
Feedback from users is always welcome, whether :-) with praise and suggestions for improvements and new features, or :-( to report crashes and other unexpected behaviour, or errors in the documentation. Perfecting the software is a high priority, and users will be assisted promptly with operational problems. Should a bug be reported, I will try to get the update online within a few days at most.
However, routine support – setting up the program, understanding the methods, explaining the documentation, etc. – cannot be guaranteed. The low price of the software reflects this limitation. Satisfy yourself that TSM functions correctly on your system in "demo" mode, before purchasing a licence. If you have trouble with numerical optimizations, please consult the relevant section of "Hints and Tips" before contacting me.
The email address for all inquiries is tsmail at timeseriesmodelling dot com,
where "at" and "dot" should be replaced by the usual symbols.