v4.3.0 with example nonlinear estimation code on PHPC 2.8
Posted: Thu Jun 21, 2012 5:23 pm
PelicanHPC (http://pareto.uab.es/mcreel/PelicanHPC) is a live Linux distribution for clustering that can be run on Windows or MacOS using virtualization. The latest release has Dynare 4.3.0, Octave 3.6.2, and it includes some example code for Bayesian estimation using 3rd order solution of the simple RBC model in the Dynare User Guide. If you boot up the image, go to /home/user/Econometrics/MyOctaveFiles/Econometrics/IL/DSGE and execute "./runme", you will see something like the following table after several minutes. The results can be improved by adjusting the script to do more simulations, the code as provided uses a small number of simulations and is only iillustrative.
Doing SBIL estimation using k-nearest neighbors nonparametric regression
100 rows read in
Doing SBIL estimation using k-nearest neighbors nonparametric regression
100 rows read in
- Code: Select all
SBIL estimation results
alpha beta delta gam rho sig eps
true param 0.330 0.990 0.023 1.750 0.950 0.010 10.000
SBIL mean 0.326 0.988 0.034 1.868 0.936 0.015 10.456
prior mean 0.291 0.976 0.033 1.903 0.924 0.023 11.301
SBIL st.dev. 0.014 0.003 0.004 0.086 0.004 0.001 0.091
prior sdev 0.078 0.014 0.015 0.590 0.039 0.011 0.966
SBIL bias -0.004 -0.002 0.011 0.118 -0.014 0.004 0.456
prior bias -0.039 -0.014 0.010 0.153 -0.026 0.013 1.301
SBIL rmse 0.014 0.003 0.011 0.143 0.014 0.004 0.464
prior rmse 0.087 0.020 0.018 0.609 0.047 0.017 1.620
number of Monte Carlo reps: 11