SIER Working Paper Series

126 Can information alleviate overconfidence? A randomized experiment on financial market predictions

Abstract

In this study, we examine how information provision affects the degree of overconfidence using an online experiment. The 4,210 experimental participants engaged in stock market prediction exercises in 2014 were asked to evaluate their absolute and relative performance. We conducted a randomized controlled trial such that randomly selected participants obtained information about their own performance and/or the distribution of others' performances before evaluating their performances. We find that while participants exhibit overconfidence bias, this can be alleviated by information provision and that the effect of the elimination of overconfidence is stronger when only partial information, rather than complete information, is provided. Further, we show that the mere provision of information, even if it is consistent with prior beliefs, decreases the degree of overconfidence.
Keywords: Overconfidence; information provision; randomized controlled trial; online experiment; stock market prediction
JEL classification: C91; D83; D91; G41