Home > Papers

Are Algorithms as Risk-Averse as We Are? - Risk-Taking under Accountability for Oneself and Others

  • Co-Author:Prof. Jason Dana (Yale University)
  • Keywords: Experiment, Lottery Choice, Risk Aversion, Responsibility, Accountability
  • Download: Paper (working version)

Project Details

We examine whether decisions made under accountability differ for self and others using lottery choice tasks that contain only positive amounts (gaining lottery), positive and negative amounts (mixed lottery) and mainly negative amounts (losing lottery). Accountability is ensured by letting participants hold up a sign with their decision after the experiment. We find that when being accountable towards the public, participants are more risk averse for others than for themselves in the gaining, mixed and loosing lottery. However, the difference is only statistically significant in the gaining lottery. We also find that participants who decide for another person’s outcome to be significantly less risk averse when being accountable than when being not accountable to the person they decided for in the gaining lottery but find no significant difference in the mixed and losing lottery.

"People choose significantly less risk-averse in the gaining domain but choose quite as risk-averse as people who decide for their outcome in the losing as well as in the mixed domain. A possible reason might be that people who decide for another person start to gamble when facing gains more than people who decide for their payoff."

Presented at: Decision Making for Others 2018, Economic Science Association World Meeting 2018