PageRank and Regression as a Two-Step Approach to Analysing a Network of Nasdaq Firms During a Recession: Insights from Minimum Spanning Tree Topology

The presence of focal firms driving entire stock markets has been proven by a series of existing studies that relied on the topological properties of minimum spanning trees. Historically, central firms have been identified primarily based on the degree centrality of nodes. This article proposes an alternative selection method, combining PageRank scores and modularity classes, which does away with the problem of ties in rankings when selecting a specific number of nodes. We use PageRank-based network analysis along with regression analysis to identify focal firms in the Nasdaq-100 index during the three most significant recent recessions in the United States. This approach validates and robustly supports our two-step method, showing that the combination of minimum spanning trees and our selection method explains over 90% of the Nasdaq-100 index’s dynamics. The analysis identified significant topological changes during the global financial crisis (with CSCO emerging as the star firm) and the COVID-19 pandemic (exhibiting strong market co-movements).