Katarzyna Budnik and Gerhard Ruenstler, "Identifying Structural VARs
from Sparse Narrative Instruments" Journal of Applied Econometrics,
Vol. 38, No. 2, 2023, pp. 186-201.

The data and used in this paper and the Matlab replication codes to
run the Monte Carlo simulations of section 3 and the estimates of
section 4 are contained in the zip file br-files.zip.  


----------------------------
DATA DESCRIPTION AND SOURCES
----------------------------

The data used for the estimates presented in section 4 are taken from
public sources, i.e. the St Louis Fed database (FRED) and the Bank for
International Settlements (BIS). Narrative indicators and the banking
deregulation index are based on own calculations, which are described
in sections A and B.4, respctively, of the Online Suppplement.
 
The data are in folder Data. Data.csv is an ASCII file in DOS format,
which contains the processed data, as they are uploaded by the
estimation routines. The EXCEL workbook Data.xls contains further 
information, including the original data from the various sources and
the transformations used to obtain the processed data.

Here is an overview of the original data and their sources:

Code         Description                                           Source

GDP          Real gross domestic product                           FRED St Louis Fed
PGDP         Gross Domestic Product: Implicit Price Deflator       FRED St Louis Fed
CPI          Consumer price index: all items                       FRED St Louis Fed
FFR          Effective Federal Funds Rate                          FRED St Louis Fed
BAA_10Y      Moody's Baa Corporate Bond Yield Relative to 10-Year  FRED St Louis Fed
BIS-TC-CRP   Total credit to non-financial corporate sector        BIS Long Credit Series
BIS-TC-HH    Total credit to private households                    BIS Long Credit Series
RPPR         Shiller house price index                             Robert Shiller & FRED

Based on own calculations

DeregGDP     Banking deregulation index (see Online Supplement B.4)
Cap_EXO      Narrative instrument for U.S. policy measures on capital requirements
UWM_EXO      Narrative instrument for U.S. policy measures on underwriting standards

-----------------
REPLICATION CODES
-----------------

The main programs to run the estimates and Monte Carlo simulations and
to produce figures and tables are contained in the root folder of the
ZIP file. 

The VAR estimates of section 4 can be replicated using:

Estimate_DSC_Main     Estimates of all Bayesian VARs
Estimate_Proxy_VAR    Estimates of frequentist proxy VARs

The two programs run loops to estimate all relevant models, including
the robustness checks discussed in section 4.3 and shown in Online
Supplement C. The results are stored in folders Results_DSC and
Results_pV. To save space, these folders are kept empty in the ZIP
file. However, the results files can be provided upon request.

Based on the results files, run Make_Table_2, Make_Table_3,
Make_Table_4, and Make_Figure_23 to make the tables and plots shown in
the paper. Similarly, run the code for producing figures C2 to C6
shown in Online Supplement C.

The simulation results shown in section 3 can be replicated using

Sim_Main_1_10  Produces the simulation results of section 3 for m=10
Sim_Main_1_20  Produces the simulation results of section 3 for m=20
Sim_Main_2     Produces the simulation results shown in Online Supplement B.3

Sim_Main_1_10 and Sim_Main_1_20 run the results for m=10 and m=20,
respectively. The simulations are better run individually, as a single
simulation takes several hours on a laptop. Choose parameters i and j
to select the appropriate simulations (see the comments in the
programs). The results are stored in folder Sim_Results_1.

Thereafter, run Make_Table_1 to process the simulation results and
obtain the statistics shown in Table 1 and figures C.1 shown in the
Online Supplement. Again, the statistics are produced separately for
each simulation. Set parameters sim & hor1 to choose the appropriate
results.

In a similar fashion, Sim_Main_2 and Make_Table_B1 contain the code
and results related to the simulation results shown in Online
Supplement B3.  

Folder Functions contains the functions used by the above programs.
Please see the comments in the individual files for details.  

Estimation      Functions for the various BVAR
Code_Mertens    Functions for the frequentist proxy VAR
Simulations     Functions for the simulations
Plots_Tables    Miscellanenous functions dor plots and tables
Other           Miscellanenous functions for date handling and data transformations


Please address any questions to:

Gerhard Ruenstler
European Central Bank
Sonnemannstrasse 20
D-60396 Frankfurt am Main
Germany
gerhard.ruenstler [AT] ecb.europa.eu