DFROG: A nowcasting model for GDP growth
Published: 20 November 2024
This paper presents DFROG, the “nowcasting” model employed by De Nederland- sche Bank (DNB) to generate near-term forecasts of the quarter-on-quarter growth rate of Gross Domestic Product (GDP). The core concept of DFROG is that the co- movement of a potentially large set of monthly economic indicators can be sum- marized into a few factors, which can then be used to forecast GDP growth. We compare the forecast accuracy of DFROG with several benchmark models and pro- fessional analysts, and conduct a thorough review of the model’s optimal specifica- tion.
Keywords: nowcasting; dynamic factor model
JEL codes C33; C53; E37
Working paper no. 819
Research highlights
- This paper presents the nowcasting model DFROG (Dutch Forecasting Model for Real-time Output Growth) that De Nederlandsche Bank uses for its near-term GDP growth forecasts;
- We find that model specification, the composition and transformation of the dataset of monthly series included in the model, and the estimation period matter for forecast accuracy;
- To insure against model instability the DFROG forecasts are an unweighted average of eighteen model specifications;
- We find that DFROG is a very competitive model within the class of nowcasting models, and also with respect to professional analysts;
- DFROG, and the other mechanical models considered, are unable to beat the near-term forecast of professional analysts during the highly volatile COVID-period.
819 - DFROG: A nowcasting model for GDP growth
819 - DFROG: A nowcasting model for GDP growth
Discover related articles
DNB uses cookies
We use cookies to optimise the user-friendliness of our website.
Read more about the cookies we use and the data they collect in our cookie notice.