The GVAR toolbox was created as an accessible and easy-to-use collection of MATLAB procedures designed for GVAR modelling.
The procedures operate by processing information inputted by the user throughout the course of the program via an Excel-based interface.
A GVAR model, also known as a Global Vector AutoRegressive model can combine individual country vector error-correcting models.
Here are some key features of "GVAR":
· An updated GVAR dataset spanning 1979Q1-2009Q4 with corresponding trade flows
· Descriptive statistics of the variables included in the individual VARX models
· Unit root tests
· Ex-ante aggregation of countries into regions
· Fixed and time-varying weights for computation of the foreign variables and solving the
· GVAR model
· Specification and estimation of VARX models including model selection criteria
· Co-integration tests in the presence of I(1) weakly exogenous regressors
· Estimation of VECMX models
· Diagnostics for individual model equations including: serial correlation tests,
· contemporaneous effect of foreign variables on domestic counterparts, average pairwise
· correlations, structural stability tests together with bootstrapped critical values
· Weak exogeneity testing
· Testing over-identifying restrictions on the cointegrating relations together with bootstrapped
· critical values
· Solution of the GVAR model and resulting eigenvalues
· A shrinkage estimator for the covariance of the GVAR model residuals
· Persistence profiles (PPs)
· Generalized impulse response functions (GIRFs)and forecast error variance decompositions (GFEVDs)
· Structural GIRFs and GFEVDs
· Bootstrap error bounds for the PPs, GIRFs and GFEVDs
· Ex-post aggregation of GIRFs and GFEVDs using country and regional weights
· GVAR model forecasts
Requirements:
· MATLAB
· Microsoft Excel
Find us on Google+
