Uncertain Kingdom: Nowcasting GDP and its Revisions

Abstract

We propose a Release-Augmented Dynamic Factor Model (RA-DFM) that allows to quantify the role of a country’s data flow in nowcasting both early GDP releases, and subsequent revisions of official estimates. We use the RA-DFM to study UK GDP early revision rounds, and assemble a comprehensive and novel mixed-frequency dataset that features over 10 years of real-time data vintages. The RA-DFM improves over the standard DFM in real-time when forecasting the first release each quarter, and economic and survey data help forecasting the first revision round. Afterwards, the predictive content of the data flow is largely exhausted.