Buying from a friend via Online Platforms? A cautionary tale of using friendship to support transactions

While observational studies have long suggested a positive correlation between social relationships and online transactions, surprisingly little research demonstrates a causal link. Effects identified in observational data generally conflate the causal Information Effect with the non-causal Homophily/environment Effect. Against this background, we conducted a large-scale randomized field experiment on a major Chinese P2P platform, in which we manipulate buyer and seller's awareness of their pre-existing friendship ties. We provide the first empirical evidence that the effect of revealing friendship information between transaction parties is statistically insignificant. To explain this counterintuitive result, we develop a theoretical model of “relationship risk,” proposing that in high-friction settings, the fear of a transaction-related misunderstanding damaging a valuable social bond can counteract the benefits of trust. We also demonstrate that reliance on observational estimates of the “Total Effect” of friendship significantly overstates the benefits of providing friendship information. Our findings contribute to a better understanding of online anonymous P2P transactions, highlight the potential fallacy of relying on observational data in business studies, and introduces the “relationship risk” model to explain why social ties may fail to facilitate, or even hinder, commerce in high-friction settings.