Nowcasting the Finnish economy with a large Bayesian vector autoregressive model

Abstract

Timely and accurate assessment of current macroeconomic activity is crucial for policymakers and other economic agents. Nowcasting aims to forecast the current economic situation ahead of official data releases. We develop and apply a large Bayesian vector autoregressive (BVAR) model to nowcast quarterly GDP growth rate of the Finnish economy. We study the BVAR model’s out-of-sample performance at different forecasting horizons, and compare to various bridge models and a dynamic factor model.