Stock Price Movement Cross-Predictability in Supply Chain Networks

Abstract

In today’s interconnected business world, firms are part of a much larger interconnected supply chain network. Individual firms are not working in isolation but rather depend on, build, and leverage the connected nature of the supply chain network. In fact, an individual firm’s performance measures, such as stock return, can be impacted by other firms in the supply chain network. In this paper, we leverage structure and connections of firms in the supply chain network as well as each firms’ performance to predict a firm’s stock price movement. Using data from real-world supply chain connections, we build the supply chain network of firms in S&P500 and utilize this network to predict a firm’s stock movements by leveraging the performance of its network neighbors. We propose four different approaches to identify network neighbors for a focal firm based on its business partners, its network community, and its role in the supply chain network. Once neighbors of the focal firm are identified, we aggregate their performance with a set of network-based feature representations we developed. Experimental results show that our approach significantly improves the performance of stock movement prediction compared to traditional features proposed in the finance literature. We are also the first to show that the performance of a focal firm is associated not only with its business partners, but also with similar firms located farther away in the network. We then analyze the contribution of upstream suppliers and downstream customers of a firm to the prediction of its stock price movement. Managerial insights from our results can improve investment decision making, as well as help supply chain managers to proactively predict risks in the supply chain network rather than simply reacting to them.