“In Design and Humans we Trust“? – Drivers of Trust and Advice Discounting for Robo Advice

Abstract

We compare the acceptance of advice in the context of robo-advised individual portfolio allocation decisions with respect to the impact of certain layout and questionnaire characteristics as well as the involvement of a human. Our data are based on incentivized experiments. The results show that a more emotional design of the advice software leads to a higher level of advice acceptance, whereas a detailed exploration questionnaire reduces the level of acceptance. The presence of a human influences trust levels significantly positive, but leads to a lower acceptance of advice in total. The latter finding is moderated by uncertainty avoidance. We attribute this to the idea that a human involved in the process is seen as an additional source of uncertainty concerning a possible betrayal, leading to “algorithm affinity” in the case of robo advice.

Keywords: robo advice, advice discounting, judge-advisory-system, investment advice, portfolio allocation

JEL Classification: D14, D81, D83, G11, G4