10.14产业经济学系列讲座第120期

时间： 14:00-15:30, Oct. 14th (Tuesday)

地点：Siyuan Building, Room 326

主持人：Dr. Binglin Gong

主题：Can Learning Eliminate Over-confidence? Evidence from a Field Experiment with Feedback

演讲嘉宾：Kingking Li, City University of Hong Kong

嘉宾简历：LI King King is an experimental economist whose work use laboratory and field experiments, combining the methodology of economics, psychology, and neuroscience to understand how people make choices. He has worked on topics such as how extrinsic motivations crowd out intrinsic motivations in the labor market, how people recall their memory and its effect on strategic interactions, and how does the sense of control affects people’s risk attitude.

简介：We conduct a field experiment to test two competing theories on why people are over-confident, namely whether it is due to imperfect information about one’s ability or psychological factor of self-image concern. Previous study by Burcks et al. (2013, Review of Economic Studies) has rejected both theories. In the current experiment, we use a repeated play design with feedback, where subjects make forecasts twice and received feedback after the first forecast, which allows us to track how over-confidence evolves over time and offer a more direct test to the imperfect information theory. The main findings are as follows. First, we found support for both theories. Second, individuals are more likely to be over-confident when they are asked to rank their performance with others (ranking forecast) than only on their own performance (marks forecast). Third, repeated play helps to lower the magnitude of the forecast error but did not eliminate over-confidence/under-confidence. Four, we found that males are more over-confident in the first forecast, a result consistent with existing literature. However, after feedback on first forecast, there is no longer significant gender difference in the second forecast. Overall, our results suggest that both imperfect knowledge and psychological factors such as self-image concerns play a role in over-confidence. Because of the later factor, learning is unlikely to eliminate over-confidence.

产业经济学系

2014-10-14