The Social Internetwork and Stock Returns

Abstract

We construct the network of U.S. firms implied by social media. We identify connections between firms and use these to construct a network that we call The Social Internetwork. The Social Internetwork describes deep, non-obvious links between firms that subsume product-supplier, industry, or news implied networks – thereby providing a more complete representation of the U.S. economy. Using own and peer stock returns, we show that information in the form of earnings surprises, and noise in the form of transitory stock returns, are transmitted across the network. Stock price changes at the most central firms in the economy ripple outwards to the most peripheral firms. We show that 17% of firms are linked directly and 92% of firms are linked indirectly through one other firm – suggesting that the economy may be more interconnected and potentially riskier and more prone to contagion than initially thought. Our Social Internetwork provides an accurate depiction of the network of firms as seen by a broad cross-section of the public, and may therefore be more accurate, timely, and dynamic than other measures.

Keywords: Networks, Cross-Asset Learning, Peer Firms, Wisdom of the Crowd

JEL Classification: G12, G14, G18