James Mitchell and Kenneth F. Wallis, "Evaluating Density Forecasts: 
Forecast Combinations, Model mixtures, Calibration and Sharpness", 
Journal of Applied Econometrics, Vol. 26, No. 6, 2011, pp. 1023-1040.

The data used in this paper are simulated.  This file describes the steps to
replicate the reported simulation results.

There are three Gauss files, which are zipped in the file mw-files.zip. They
are ASCII files in DOS format. Unix/Linux users should use "unzip -a". All
files are executable in Gauss 9.

There are two simulation experiments in the paper: 

(i) GBR's IID example;
(ii) Forecasting an autoregressive (AR) process.

*****

GBR's experiment:  Fig1.g and Fig1_tests.g

Fig1.g: Re-creates Figure 1 in MW.
Fig1_tests.g: Takes the probability integral transforms (PITS) from Figure 1
(GBR's IID case) for each of the 4 forecasters and uses the evaluation tests
in MW to try and discriminate between them.

*****

Forecasting an AR:  Tables.g

Simulation to generate the information needed for Tables 2-6 in MW. 
Returns tables in LaTeX form. 
Also returns PIT histograms used for Figure 2.


Each Gauss file also provides brief instructions to the user.

Please address any questions about the code to j.mitchell [AT] niesr.ac.uk