For richer and for poorer? : An Experimental Study on Human and Robo-Advisors in Gain and Loss Frames

Sammanfattning

Despite offering superior cost-efficiency and objective performance, robo-advisors still account for a mere fraction of total assets under management. This study investigates how the framing of a financial decision (gain vs. loss) affects the propensity to delegate the decision to a robo-advisor compared to a human advisor. An experimental survey was conducted via Prolific (N = 200). While an initial logistic regression showed no significant overall effects, the inclusion of an interaction term and control variables indicates a suggestive trend of frame-dependent algorithm aversion; however this effect did not reach statistical significance. Specifically, in the gain frame, the probability of delegation to the robo-advisor decreased by 16.1 percentage points compared to the human advisor (p = 0.100). This effect was neutralized in the loss frame, where no significant difference between advisor types was found. These results, which remained consistent throughout robustness checks, suggest that algorithm aversion may not be a static trait but conditional on the framing of the decision.