The Effect of Image Choice on Airbnb Reservations: A Combination of Deep Learning and Econometric Analysis

In a digital sharing economy, sellers and buyers provide cues to reduce information asymmetry in transactions. Using Information Processing Theory and Signaling Theory as our theoretical lens, we investigate how the signals provided by the hosts using images impact renting decision of guests on \ Airbnb. We conduct an exploratory study using cross-sectional data of Airbnb listings. We use trained Convolutional Neural Network (CNN), a deep learning technique along with manual labeling to reliably code the images as indoor and outdoor images. Our results demonstrate that the hosts presenting outdoor image of their houses as the first image receive lower reservation compared to hosts presenting indoor image as the first image. Moreover, the number of outdoor images presented by the host is negatively associated with the reservation of the listing. Our study provides practical implications for Airbnb hosts and furthers our theoretical understanding of role of cues in digital sharing economy.