Designing a Nowcasting Model for GDP Growth: A Practical Approach
Published in De Economist, 2025
🆕 This paper presents a practical guide for developing a nowcasting model for GDP growth. We employ a dynamic factor model to generate backcasts, nowcasts and forecasts of Dutch GDP in pseudo real-time. We evaluate forecast errors across a range of modeling alternatives, including data choice, transformations, outlier correction and model specification. The optimal combination of these alternatives outperforms standard benchmarks. Additionally, we demonstrate how to derive forecast contributions and assess the impact of new data releases, offering policymakers valuable insights for interpreting GDP nowcasts. Finally, we collect GDP nowcasts from professional forecasters and show they were relatively accurate during the COVID crisis.
Download publication here or here from the publisher’s website.
A layman’s explanation of DFROG can be found on the DNB-website here or here as a pdf-file
Recommended citation: Rooijen, van M., Dijk, van D.W. en J.M. de Winter (2025), Designing a Nowcasting Model for GDP Growth: A Practical Approach, De Economist , 173, 665–704.