Bank Scandal and Customer Sentiment

Abstract

In this paper, we investigate the impact of Wells Fargo’s 2016 "Cross-Selling" scandal by using publicly available Consumer Financial Protection Bureau (CFPB) complaint data and farm level risk and sentiment scores. We use natural language processing (NLP) to construct negative sentiment scores (NSS) based on the complaint texts, and employ the synthetic control method (SCM) to estimate how the sentiment score changed as a result of the scandal. We find that, contrary to popular belief, negative sentiment towards Wells Fargo (WF) decreased following the crisis. The result is also statistically significant and robust to alternative specifications. The findings suggest that effective internal practices aimed at improving customer satisfaction could help with more than just minimizing the negative effects of business scandals.

Keywords: Customer Sentiment, NLP, SCM, CFPB

JEL Classification: G20, G21, G29, G40, G41