Nikolay Gospodinov and Damba Lkhagvasuren, "A Moment-Matching Method
for Approximating VAR Processes by Finite-State Markov Chains",
Journal of Applied Econometrics, Vol. 29, No. 5, 2014, pp. 843-859.

1. The estimation of the bivariate VAR process in Equation 15 (see the
paper) represents the main empirical part of the paper. The file
gl-data.txt (size 4KB) contains two main variables that are used to
estimate the VAR process. There are two columns in the file. The first
column is the Solow residual constructed by Gomme and Lkhagvasuren
(2013). The second column is the logarithm of the real per-capita
government expenditure in the U.S (also see part 4 below). The two
series are at a quarterly frequency. Specifically, there are 252 rows
in the data file, spanning the period of 1948:Q1-2010:Q4. 

2. The matlab code estimating the VAR process, using a bootstrap
bias-corrected procedure, is provided in the file shocks_estimate.m
(size 2KB).

3. The file gl-appendix.pdf (size 181KB) is the online appendix which
contains further analytical results (incl. Proposition 1) and
numerical details.  

4. We also provide additional data needed to calibrate Gbar and to
construct the real per-capita government expenditure. These are
contained in gl-data-aux.txt (size 20KB). The file contains five
columns.

    column 1: quarter (YYYYQ);
    column 2: nominal private gross domestic product (in billion dollars);
    column 3: nominal government expenditure (in billion dollars);
    column 4: the consumption deflator; and
    column 5: U.S. civilian non-institutionalized population aged 16 and
              over (in thousands).
	
The total size for the four files is approximately 210KB.

All files are ASCII files in DOS format. They are zipped in the file
gl-files.zip. Unix/Linux users should use "unzip -a".

Please address any questions to:

Damba Lkhagvasuren 
Department of Economics, Concordia University
1455 Maisonneuve Blvd. West, H1155.43
Montreal, QC H3G 1M8, Canada