Nowcasting consumer price inflation using high-frequency scanner data: Evidence from Germany

We study how millions of granular and weekly household scanner data combined with machine learning can help to improve the real-time nowcast of German inflation. Our nowcasting exercise targets three hierarchy levels of inflation: individual products, product groups, and headline inflation. At the individual product level, we construct a large set of weekly scanner-based price indices that closely match their official counterparts, such as butter and coffee beans. Within a mixed-frequency setup, these indices significantly improve inflation nowcasts already after the first seven days of a month. For nowcasting product groups such as processed and unprocessed food, we apply shrinkage estimators to exploit the large set of scanner-based price indices, resulting in substantial predictive gains over autoregressive time series models. Finally, by adding high-frequency information on energy and travel services, we construct competitive nowcasting models for headline inflation that are on par with, or even outperform, survey-based inflation expectations.