Noise Trading: An Ad-based Measure

Abstract

This paper proposes a novel measure of noise trading that aims to capture uninformed retail trading. The measure, an indicator of whether the firm placed advertisement(s) in the Wall Street Journal seven calendar days earlier, is motivated by evidence that retail trading spikes seven days after ad days, that firms regularly place ads at weekly intervals, and that weekly ads frequently contain duplicate images. This ad-based measure is positively associated with informed trading and stock price volatility. Collectively, our results provide broad support for the theoretical predictions of Collin-Dufresne and Fos (2016, Econometrica).