Bayesian Compressed Vector Autoregressions

This code demonstrates the methods in the paper

Koop, Korobilis and Pettenuzzo (2017). Bayesian compressed VARs. Journal of Econometrics

Note that all the functions needed for replication of the results are there, but due to the vast amount of computation needed we have instead set up a default forecasting exercise using medium-sized VARs with point forecasts only (i.e. with no computationally expensive predictive simulation). This can serve as a demonstration for less experienced users in order to understand how BCVAR (and the competing methods) work in practice.

Feel free to use the code any way you want (and don't forget to cite please), however, remember that we do not offer support for this code and that by downloading the code you bear sole responsibility for finding out how BCVAR (or MATLAB, or Bayesian inference in general) works.

Download code here: BCVAR.zip