Haroon Mumtaz and Paolo Surico, "Policy Uncertainty and Aggregate
Fluctuations", Journal of Applied Econometrics, Vol. 33, No. 3,
2018, pp. 319-331.

The dataset is quarterly and runs from 1970Q1 to 2015 Q4 for the
United States. It is provided in a "tab-delimited" text file called
ms-data.txt. This ASCII file is in DOS format. It is zipped in the
file ms-data.zip. Unix/Linux users should use "unzip -a".

The file ms-code.zip contains Gauss and Matlab computer programs to
produce the estimates reported in the paper. All program files are
ASCII files in DOS format. Unix/Linux users should use "unzip -a". 

The file mumtaz-surico-appendix.pdf is the technical appendix to the
paper and provided details on the Gibbs Sampling algorithm

Description of data

The file ms-data.txt has 11 columns. The first column specifies dates.
The second column is the log of Government Spending per capita. The
third column is the log of investment per capita. The fourth column is
the log of consumption per capita. The fifth column is the log of GDP
per capita. The sixth column is the annual CPI inflation rate. The
seventh column is the log of per-capita net taxes. The eighth column
is the debt to GDP ratio. The ninth column holds the three month
treasury bill rate. The tenth column holds the Business Confidence
Index while the eleventh column holds the Consumer Confidence index.


Description of Code in ms-code.zip

(1) Estimate.src: This Gauss code estimates the benchmark VAR model
with stochastic volatility in mean. Before running this file, the user
must create initial values by running the simpler VAR with stochastic
volatility. This can be done by running the Matlab file estimate1.m in
the sub-folder "start". This will automatically create the starting
values and save them in output files that can be read in by
Estimate.src.

This file contains an option on line 7 to replace the Kalman filter in
Gauss with a faster Fortran alternative. To use this option, the file
Kfilter.f90 in the folder kfilter must be compiled into a dynamic link
library (DLL). In our work, we use Intel Visual Fortran compiler for
this purpose. Alternatively the code folder with the compiled DLL can
be downloaded at
https://www.dropbox.com/s/t8jekmro8ewaizz/MS-CODE_old.zip?dl=0

(2) GETIRF.src: Gauss code to calculate impulse responses to
volatility shocks and Forecast Error Variance Decomposition using the
parameter estimates from Estimate.src.

(3) GETIRFLEVELS.src: Gauss code to calculate impulse responses to
Level shocks using the parameter estimates from Estimate.src.

(4) GETVOLS.src: Gauss code to extract the stochastic volatility of
policy shocks.

(5) GETDIC.src: Gauss code to calculate the Deviance Information
criteria for the VAR with stochastic Volatility in mean.