Estimation and Inference of the FEVD for Set-Identified SVARs

Replication package for "Estimation and Inference of the Forecast Error Variance Decomposition for Set-Identified SVARs" by Francesco Fusari, Joe Marlow and Alessio Volpicella.

Structure

The repository consists of three subfolders:

• Data: it contains the datasets data.mat and data_levels.mat used in the empirical applications (generated using dataset_creation.m).

• Main: it contains nine run files, each generating some of the results presented in the paper. See below for a more detailed description.

• Results: mat contains the output of the run files; while create_figures.m allows to generate all the figures contained in the paper.

Description

• main_file.m: main empirical application (Figure 3, Table 2) and main Monte Carlo (Figure 6, Table 3, Figure F.14, Figure F.16, Figure F.17).
• main_file_large_R.m: empirical application with large number of possible combinations for r(\phi).
• main_file_proxy.m: Proxy SVAR empirical application (Figure 4, Figure 5).
• main_file_bayes.m: confidence intervals when inference is performed by using Giacomini and Kitagawa's (2021) approach (Figure E.11).
• main_file_delta.m: Monte Carlo for narrow identified sets (Figure 7, Table 4 and Figure F.15).
• main_file_irf.m: Monte Carlo for the IRFs (Figure D.9 and Figure D.10), and FEVD plug-in estimated bounds (Figure E.12).
• main_file_levels.m: empirical application in levels (Figure F.18 and Figure F.19).
• main_file_multiple.m: Monte Carlo under multiple optima (Figure F.20 and Figure F.21).
• main_file_point.m: Monte Carlo under point identification (Figure F.22).
• main_file_gk.m: robust Bayesian inference in GK21 and standard Bayesian inference (Figure E.13).