The relationship between Twitter uncertainty measures and stock market returns during the COVID-19 crisis

Purpose This paper aims to explore how uncertainty captured through social media, particularly Twitter, affects global stock markets during the COVID-19 crisis. It focuses on whether the Twitter economic uncertainty (TEU) and Twitter Market Uncertainty (TMU) indices contain information that helps explain market behaviour in periods of stress. Design/methodology/approach The analysis relies on daily data from January 2007 to May 2023. This paper combines linear and nonlinear Granger causality tests (Diks and Panchenko, 2006) with the BDS nonlinearity test, and a Dynamic Conditional Correlation-Generalized Autoregressive Conditional Heteroskedasticity (DCC-GARCH) connectedness framework is used (Gabauer, 2020) to capture the dynamic transmission of volatility between Twitter-based sentiment indicators and 13 international financial markets. Findings The results highlight strong nonlinear causal effects from Twitter-based uncertainty measures to market returns. TEU and TMU indices appear to drive fluctuations in stock returns, particularly during episodes of heightened uncertainty. The DCC-GARCH analysis shows that TEU-ENG and TEU-USA act as net transmitters of volatility to markets such as Japan, Ukraine, China, Germany and even Bitcoin. Research limitations/implications As the study focuses mainly on the COVID-19 crisis, future research could extend the analysis to other stress periods or explore alternative sentiment measures such as the volatility index. Practical implications The findings suggest that social media sentiment can serve as a timely indicator for risk managers and investors, complementing traditional uncertainty measures. Social implications The results highlight how collective behaviour on social networks can amplify uncertainty and market reactions, underscoring the need for responsible communication and better public understanding of financial information online. Originality/value This study offers fresh evidence on how Twitter-based uncertainty indices interact with market dynamics, combining behavioral–finance insights with nonlinear econometric techniques.