Measuring market volatility connectedness to media sentiment

•We build novel sentiment indices for news and Twitter for six selected markets using machine learning.•We examine directional connectedness from media to market volatility.•Market volatility connectedness to media is time-varying.•Sentiment relevant to one market induces volatility in other markets.•Sentiment-driven market volatility appears and disappears following a spiky pattern. We examine directional connectedness patterns from news and social media to financial market volatility using textual analysis and high-frequency data. We find that media sentiment induces market volatility, but the magnitude of that connectedness is time-varying. In addition, news and social media sentiment pertinent to one market transmits volatility to other markets. Finally, we find that sentiment transmits sharp shocks to markets during major events. At other times, there are smaller spillover effects, indicating that the directional connectedness from sentiment to markets follows a spiky pattern over time. We conclude that news and social media play an important (but not constant) role in transmitting volatility across financial markets. This insight explains earlier divergent findings in the literature.