Travail de recherche/Working paper
| Résumé : | We propose a new framework for measuring investor attention in real time at high- frequency trading scales. Our approach relies on the Twitter messages of financial experts around the release of unscheduled news announcements. Using high-frequency data on large-cap U.S. stocks, we examine market reaction to new information both in the presence and absence of investor attention. We find evidence that news events receiving high attention lead to large and persistent changes in abnormal trading activ- ity, volatility and price jumps. When investors pay little attention to news, however, the effects of news on such trading patterns tend to be smaller and vanish quickly. With respect to timing, approximately half of the high-attention news stories arrive first on Twitter before being reported by Bloomberg. This result suggests that pre- announcement effects may not be explained only by private information, but could be related also to timestamp delays. We control such potential biases with attention- adjustment and corrected newswire timestamps, which significantly eliminates pre- announcement effects. When timing trades, market participants should account for investor attention and biases in news releases. |
