U.S. Banks’ lending behaviour, financial stability, and investor sentiment: A textual analysis

Abstract

We examine the impact of investor sentiment on bank credit and financial stability. We also investigate how loan growth may affect bank stability. We use an unbalanced panel data set of 6,886 U.S. commercial banks over the period 1990-2015, using bank-level data. Investor sentiment is proxied by two novel but alternative measures based on textual analysis. First, we employ the measure constructed by Garcia (2003) based on the fraction of positive and negative words in two columns of financial news from the New York Times. Second, we employ the text-based measure of uncertainty constructed by Manela and Moreira (2017) called News Implied Volatility, which uses front-page articles of the Wall Street Journal. The results show that banks’ lending falls when investor sentiment is low, while this effect is more pronounced when banks hold a higher level of credit risk. These effects are more pronounced during recessions, and in these periods loan growth also responds negatively to the anxiety of investors. Finally, during the 2007-2009 financial crisis the negative effect on bank stability was weaker since any increase in bank lending provoked by investor sentiment was counteracted by the events that took place during and after the crisis. Keywords: U.S. banks; textual sentiment analysis; anxious periods; loan growth; risk taking; bank stability JEL: G10; G14; G21; G28; G39