Algorithmic Influence: Empirical Evidence from Microlending
Consumers often follow prior decisions of other consumers due to observational learning, which is known as "social influence.'' This paper explores whether consumers also follow prior decisions of algorithms, which we call "algorithmic influence.'' Using detailed data from a leading peer-to-peer microlending platform. We find evidence that consumers, conditional on making a manual investment, invest more in a listing when a prior investment made by an algorithm is visible. This effect is driven by the visibility of the "algorithmic'' label that appears next to the algorithmic investment rather than by the nature of the investment. Further, we find that consumers respond more positively to visibility of an algorithmic investment when there is greater uncertainty about whether a listing will become a loan and when the listing's risk level deviates more from consumers‘ risk preferences. These findings are driven mainly by consumers who have limited experience with manual, non-algorithmic investment. With increasing exposure to algorithmic investments, consumers are more likely to diversify the degree of risk in their manual investment portfolio and to obtain higher returns. Our results suggest that it is potentially advantageous to managers to highlight instances when algorithms make decisions to induce individual investors to emulate algorithmic decisions.
Meeting ID: 955 6069 1918