Information advantage of narratives to consumers' inflation forecasts

This study examines the information content of narratives to inflation forecasts. Although text analysis of news articles, central bank policy statements, and economic outlooks have increasingly gained attention, few studies consider layperson narratives. Using novel text data from Japan, we find that the sentiment indexes constructed from narratives contain additional information on consumers' inflation forecasts. Furthermore, a sentiment index constructed using naive Bayes provides additional information on survey-based sentiment. These findings suggest that narratives provide useful information regarding monetary policies.