Released on 19/03/2012:
• | YADA now supports the univariate Kalman filter and smoother. The basic algorithms are described by . |
• | Diffuse initialization of the Kalman filter has been added. |
• | Monte Carlo filtering for investigating the sensitivity of the DSGE model solution to the values of different parameters. The approach is discussed in some detail by . |
• | YADA now calculates skewness and kurtosis from the DSGE and DSGE-VAR models when simulating the observed variables. For details, see observed variable correlations for the DSGE models. |
• | Conditional forecasting now supports time-varying number of conditioning assumptions and lag-varying K2 matrices. The former are imposed by checking the conditioning assumptions for NaN values, and the latter through K2 being 3-dimensional. For details, see data construction file. |
• | Sequential marginal likelihood estimation and eigenvalue plots for DSGE-VARs. |
• | The Laplace approximation of the marginal likelihood of the DSGE model can now be recomputed for the full sample from the Sequential marginal likelihood estimation menu-item on the View menu. |
• | It is now possible to restrict the posterior draws of the VAR parameters in DSGE-VARs and BVARs to imply a stationary VAR. This option is available on the lag order selection frame on the Bayesian VAR tab. |
Page url:
http://www.texlips.net/yada/help/index.html?3_10.htm