Asset Pricing Implications of Social Networks

Recent empirical studies suggest that social networks, according to which communication takes place, have a significant impact on traders' financial decisions. Motivated by this evidence, we propose an asset pricing model in which agents communicate information according to a social network. In the proposed model, agents initially have imperfect and diverse information about the asset payoff structure. Via communication, agents learn, i.e., gain information, about the asset payoff structure from others in the economy. The social network indicates whom each agent learns from. The social network is exogenous and can be considered to represent geographical proximities as well as social relationships (such as friendships and acquaintanceships). The model generates several novel implications. First, we prove that social influence is a determinant in asset pricing, where one's influence is determined by her connections in the social network. Then we show that proximities between agents in the social network affect agents' asset demand correlations: demands of agents from the same tight-knit social cluster exhibit higher correlations compared to demands of those from disjoint social clusters. Impact of social networks on asset price volatility is also explored. We demonstrate that learning in social networks may account for the observed high volatility ratio of price to fundamentals in the stock markets. Finally, we investigate how different specifications of social networks affect agents' assessments of the asset payoff structure. To that end, we introduce the notions of informational dominance and informational efficiency, which essentially rank social networks according to the precision of information they generate for agents in the economy. We provide partial characterizations of informationally dominant and informationally efficient social networks.