Greek GDP Forecasting Using Bayesian Multivariate Models

Building on a proper selection of macroeconomic variables for constructing a GDP forecasting multivariate model (Kazanas, 2017), this paper evaluates whether alternative Bayesian model specifications can provide greater forecasting accuracy compared to a standard VECM model. To that end, two Bayesian models are estimated, a Bayesian VAR using Litterman’s prior (1979) and a Bayesian VAR with time-varying parameters (TVP-BVAR) (Primiceri, 2005). The out-of-sample forecasting performance of all three models is then evaluated over a period of 21 quarters (2016:Q1 to 2021:Q1), where the Litterman BVAR is found to greatly outperform both the VECM and the TVP-VAR