Brennan, Michael J. and Huh, Sahn-Wook and Subrahmanyam, Avanidhar, High-Frequency Measures of Information Risk (March 12, 2015). Available for download at SSRN: http://ssrn.com/abstract=2578168

“We estimate each day for each stock the conditional probabilities of informed trading on good and bad news in the spirit of Easley et al. (1996), and provide new evidence that these variables capture informed trading. The estimated daily conditional probability measures increase before earnings announcements and remain high over the days following the announcements. Consistent with informed trading accelerating the adjustment of prices to new information, informed trading prior to earnings announcements are associated with smaller price reactions to earnings surprises on the announcement dates. Our results also show that in general higher probabilities of informed trading on good (bad) news forecast higher (lower) stock returns up to three (two) days ahead.“