When Can the Market Identify Old News

We document an "aggregation effect" in reactions to stale news. Using a novel dataset of news transmission through the Bloomberg terminal, we differentiate stale news into two types: "duplicate" stories that directly restate single previous articles and "aggregate" stories that combine content from several previous articles. We conjecture that duplicate stories are more immediately identifiable as stale than aggregate stories. The empirical results confirm this hypothesis: absolute abnormal returns for a firm are higher when it has more aggregate stories relative to duplicate ones, and these returns are more likely to reverse over the following week. Time trends in the estimated coefficients indicate that reactions to stale stories have decreased over time, but the differential reactions to aggregate versus duplicate stories have risen. Altogether, the results point to investors' increased sophistication in identifying stale information in duplicate stories, but continuing susceptibility to old news in the form of aggregate stories.