GMM Toolbox for Matlab
GMMTBX is a set of MATLAB® functions that perform GMM estimation and testing of linear/nonlinear
time series and cross section models.
One of its major features is that
it includes a Graphical User Interface that controls all the functions
of the toolbox.
This provides a friendly, yet powerful interface for
In addition, all the functions are stand alone and can be
simply executed from Matlab's command prompt.
The GMMTBX requires the use of the Matlab® Optimization and Statistics Toolboxes. All the functions, apart from the GUI, are compatible with Matlab version 5.0 or higher; the GUI is compatible with Matlab version 6.0 or higher.
The GMM toolbox
UPDATE #1 (Feb/01/2005): All GUI windows are now resizable. Just click/drag on any of the window corners to resize it.
UPDATE #2 (Feb/08/2005): Corrected a problem that was resulting to error messages during the execution of the GUI.
Download a zip file that contains:
(i) A Matlab function for calculating the CBAPM moments and their gradient (cbapm.m)
(ii) A Matlab function that calculates a set of linear restrictions on the CBAPM parameters (cbapmtest.m)
Download the dataset used for the CBAPM estimation
Study all the features of the program using the Consumption Based Asset Pricing Model (Page will load in new window)
How this thing works? Documentation/help files (Page will load in new window)
Graphical User Interface (GUI), for a user friendly environment that can control both estimation and testing
GMM estimation for both unrestricted and restricted models (linear, nonlinear, or "bounded parameters" restrictions)
Continuous Updated GMM
Variance, std. deviation, and 95% confidence intervals of the parameters
J-test for testing the validity of the moment conditions
Wald, LM, and D tests for both linear and nonlinear constraints
Structural stability tests in the case of known break-point
Sup-, AV-, and Exp- structural stability tests in the case of an unknown break-point
Moments' long-run covariance matrix includes : Serially Uncorrelated, HAC with Bartlett weights, and HAC with Parzen weights
Bandwidth's value: Provided by the user or automatically calculated using Newey and West's method of bandwidth selection
Last Update: November 21, 2007