Large Bayesian Vector Autoregressions

Joshua Chan (2020)
In: P. Fuleky (Eds), Macroeconomic Forecasting in the Era of Big Data, 95-125, Springer, Cham
[ Journal Version | Working Paper | Code ]

This code produces point and density forecasts from seven large Bayesian VARs with differnt priors and covariance structures, including the Minnesota prior, the natural conjugate prior, the independent prior and the SSVS prior.

The application is illustrated using a US real-time dataset of 20 macroeconomic variables at quarterly frequency.

The seven large Bayesian VARs are:

1. BVAR-Minn: BVAR with the Minnesota prior
2. BVAR-NCP: BVAR with the natural conjugate prior
3. BVAR-IP: BVAR with the independent normal and inverse-Wishart prior
4. BVAR-SSVS: BVAR with the SSVS prior
5. BVAR-CSV: BVAR with a common stochastic volatility
6. BVAR-CSV-t: BVAR with a common stochastic volatility and t errors
7. BVAR-CSV-t-MA: BVAR with a common stochastic volatility and MA(1) t errors