Matlab code for a series of exercises on informational sufficiency for SVAR. It applies monetary policy surprise series as in Gertler and Karadi (2015) as an external instrument to identify a monetary policy shock, then proceeds to assess whether the underlying SVAR is fundamental using orthogonality tests as in Forni and Gambetti (2014). Then it compares two versions of an SVAR: an original and an informationally sufficient one by computing the impulse-responses and forecast error variance decompositions.
main.m: Main file to produce most of the results.
\functions: Functions to estimate VAR and IV regressions, compute impulse-responses and other subroutines.
\data: Data for analysis:
- fred_md_012022.csv: series from FRED-MD database used in factor construction.
- GK2015_Data.xlsx: original Gertler and Karadi (2015) data.
- FRED-MD_updated_appendix.csv: an auxiliary file with filenames and transformation codes.