Geography, Diversity, and Accuracy of Crowdsourced Earnings Forecasts
61 Pages Posted: 18 Mar 2015 Last revised: 6 Oct 2016
Date Written: October 5, 2016
Using a novel dataset containing the forecasts of both buy-side and sell-side analysts, and individual investors, we find that crowdsourced earnings forecasts are more accurate than expert forecasts of sell-side analysts. Examining the economic mechanisms that generate superior crowd forecasts, we find that the diversity of contributors and their geographical proximity to firm locations improve forecast accuracy. The crowdsourced consensus is a better measure of the market’s true earnings expectations as earnings surprise based on this consensus generates stronger market reactions. A trading strategy based on the difference between the two consensus estimates yields an abnormal 10-day return of 0.465-1.975%.
Keywords: Crowdsourcing, crowd diversity, local bias, analysts, earnings forecasts, forecast accuracy
JEL Classification: G14, G24
Suggested Citation: Suggested Citation