Social Ties and User Generated Content: Evidence from an Online Social Network

Abstract

We use variation in wind speeds at surfing locations in Switzerland as exogenous shifters of users' propensity to post content about their surfing activity onto an online social network. We exploit this variation to test whether users' online content generation activity has a causal effect on their social ties. Under weak monotonicity assumptions, we also estimate nonparametric bounds on the causal effect of user's social ties in turn on their content generation activity. Economically significant causal effects of the type above can produce positive feedback that generates local network effects in content generation. We find evidence for such network effects. We argue this feedback generates a multiplier effect on interventions that subsidize tie formation. We use our estimates to measure the ROI from such interventions and discuss implications for the site's monetization strategy. Our empirical strategy provides one way to address a significant identification challange with online social network data that the observed network structure is endogenous to the actions taken by agents on the network. Augmenting the model of agent's actions with a model for the network structure requires solving a formidable network formation game. Our approach to this problem is to conduct inference with an incomplete model of network formation under weak assumptions that deliver informative bounds on the causal effects of interest, while avoiding taking a strong stand on a specific model of network formation.

Keywords: Marketing, User-Generated Content, Social Networks, Monotone Treatment Response, Monotone Treatment Selection, Monotone Instrumental Variable, Homophily, Endogenous Group Formation, Correlated Unobservables, Endogeneity