We introduce the tardy news hypothesis as a new channel that complements information leakage to explain why volatility rises before public news. Through a novel information processing framework, we show that both channels can lead to bias and inconsistency in the estimated relationship between news and volatility, and we establish bounds for the true relationship. Using a dataset that combines realized volatility withRavenPack news analytics for S&P 500 constituents, we find that the type of events anews item covers is related to the likelihood of a rise in pre-news volatility, and the sizeof the firm is related to the magnitude of the rise. This heterogeneity from event and firm types is consistent with the tardy news hypothesis.<p></p>