Measuring tone shifts in media coverage: An empirical application to bank financing conditions in Italy

Abstract

Media narrative is key to track changes in depositors’ and investors’ trust alike towards banks. Previous studies have applied sentiment analysis to news, to monitor changes in liquidity and financing conditions. Our proposal leverages natural language processing to measure shifts in Italian newspapers’ tone at the individual bank level. We show that our set of indicators is effective at tracking and predicting the evolution of several variables related to the financing condition of the banking sector, including their stock prices, trading volumes and sight deposits. These results highlight the benefits of employing natural language processing methods to monitor banks’ funding risks.