Press Coverage of the Economy: Construction of Coverage and Sentiment Indices

Media represents the main source of information regarding economic issues. This article builds comprehensive indices reflecting the media narrative on the overall state of the economy in Italian newspapers using the Lexis Nexis database from 2009 to 2020. For this reason, we employ a machine learning approach, implementing two NLP models with the language model BERT: the first discriminates between articles discussing the economy and those that do not, while the second assigns sentiment to the “in-topic” news. Additionally, we compare these coverage and sentiment indices with traditional indicators of the Italian economy. The results show that the economic trends described in the media are not always aligned with actual economic trends. We also discuss potential applications of these indices in population studies, such as fertility research, highlighting the potential of media-based indices as a complementary tool for understanding social phenomena.