Abstract: Economic decisions are often made after human interactions. This paper proposes an empirical approach to process and quantify features of micro-level human interactions, documents their connections to financial investment decisions, and investigates the underlying economic mechanisms. Using machine learning (ML)-based algorithms with videos as data input, we quantify human interactions in three-V dimensions—visual, vocal, and verbal—and construct interpretable metrics along these dimensions. We apply our method to videos of entrepreneurs pitching investors for funding. We find that venture investors are more likely to invest in startup teams that show more positivity (i.e., happy, warm, passionate), and the penalty of deviating from these features is larger for women. This relation does not appear to be driven by investors correctly calibrating startup quality using pitch features. Instead, through pitch data analysis and an experiment, we show that interaction features affect economic decisions through both the taste-based channel (18 percent) and the inaccurate beliefs channel (82 percent). Overall, the evidence is consistent with interaction-induced biases.
