Essays on Inefficient Information Flow and Biased Beliefs

Abstract

This dissertation contains three chapters. Using a combination of field experiments and lab experiments, it examines the barriers that prevent the efficient flow of information and decision markers’ biased beliefs that might contribute to the inefficiency. In the first chapter, which is joint work with Yihong Huang and Ziqi Lu, we examine whether people are biased by repetitive information via correlation neglect bias on social media. We ran a field experiment on Weibo, a Chinese social media platform. In our experiment, we recruit potential civil service applicants and present them with correlated posts about civil service careers written by current workers. Participants in the treatment group are exposed to (obviously) identical and redundant copies of the same information, and these participants develop significantly more negative views on civil service jobs, compared to the control group. Treated participants are also over 6% less willing to sign up for the upcoming civil service exam. In the second chapter, which is joint work with Fei Yuan, we examine whether parents’ investment decisions are affected by their beliefs about their peers’ investment levels. Using a field experiment in China, we demonstrate that parents hold biased beliefs about their peers’ investment levels at baseline. When provided information about the true distribution of tutoring hours invested among their reference groups, parents in the treated group significantly update their beliefs and adjust their investments accordingly. The effect is mainly driven by those with low baseline investment levels. In the third chapter, which is joint work with Yihong Huang, we explore whether and by how much blame concerns deter advice-giving and isolate the mechanisms behind it. We designed a lab experiment where an advisor has private information which could help a potential advisee make a risky investment decision. We first document that both external concerns (punishment) and internal concerns (guilt) deter advice-giving. We also find that allowing advisors to send messages to advisees reduces the perceived probability of punishment. This significantly alleviates external blame concerns and leads to higher advising rates for male advisors but not female advisors.