Social Connections and Crowdsourced Earnings Predictions
主 题: Social Connections and Crowdsourced Earnings Predictions
报告人: 王翀 (香港城市大学商公司)
时 间: 2017-03-16 15:00 - 16:00
地 点: 理科一号楼1114
报告人简介：王翀，Assistant professor，香港城市大学商公司
2000级本科，2002年-2004年金融数学系本科学习。此后分别于清华大学和香港科技大学获得金融学硕士和管理学博士学位。2012年起于香港城市大学商公司咨询系统学系任教。王翀博士的研究关注信息技术发展的社会、经济意义。他的研究课题涉及网络信息环境中的决策过程，金融信息技术创新与商业模式研究和平台化商业模式中的机制设计问题。他独立主持多项由香港大学教育资助委员会资助的研究课题，同时主持和参与两项自然科学基金项目的研究。他的研究成果发表于Information SystemsResearch，Journalof Management Information Systems 和 Decision Support Systems等期刊。
【报告摘要】：Crowdsourced(earnings) predictions are gaining popularity. As in many other applicationcontexts, the collective wisdom can make more accurate earnings predictionsthan professional sell-side analysts. However, the ingredients for the wisdomof crowds remain mysterious. In this study, we explore whether online socialconnections, enabled by an analyst follow function, can help to improve theaccuracy of earnings predictions. Using data from a major crowdsourcingplatform for earnings predictions, we leverage on a natural experimentalsetting and find strong evidence that social connections help to enhance theaccuracy of individual predictions. Specifically, when analysts have moresocial exposure (i.e. more followers), they become more prudent and accurate intheir predictions. We do not find evidence that access to information throughthe social connections (i.e. more followees) improves prediction. However, ourbacktesting suggests that betting on more popular analysts does not deliver abetter return. We discuss the different implications of our study for platformmanagers and fund managers.
