Nowcasting and Aggregation: Why Small Euro Area Countries Matter

Abstract

The paper studies the nowcasting of Euro area Gross Domestic Product (GDP) growth using mixed data sampling machine learning panel data regressions with both standard macro releases and daily news data. Using a panel of 19 Euro area countries, we investigate whether directly nowcasting the Euro area aggregate is better than weighted individual country nowcasts. Our results highlight the importance of the information from small-and medium-sized countries, particularly when including the COVID-19 pandemic period. The analysis is supplemented by studying the so-called Big Four-France, Germany, Italy, and Spain-and the value added of news data when official statistics are lagging.

Keywords: hierarchical nowcasting, high-dimensional panels, mixed-frequency data, text data