Download This Article

Zornitsa Todorova ORCID logo


Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.


Using methods from graph theory and network analysis, this paper identifies, visualizes and analyzes a correlation network of residual stock returns for more than 5,000 US-based publicly traded firms. Building on prior work by Billio et al. (2012), the paper computes a systemic measure of network centrality using principal components analysis. Two main questions are addressed: 1) What is the empirical relationship between expected stock returns and network centrality? and 2) Does network centrality have predictive power to identify firms, which are most at risk during systemic events? First, the paper finds that network centrality has substantial predictive power in out-of-sample tests related to the recent financial crisis. Second, firms that are more central in the network earn higher returns than firms that are located in the periphery. The paper rationalizes this finding by arguing that central firms are characterized by higher market risk because they are more exposed to idiosyncratic shocks passing through the network. Finally, the paper develops a novel factor-mimicking portfolio, weighted by centrality scores. The investment strategy earns an annualized risk premium of 3.38 % controlling for market beta, size and book-to-market.

Keywords: Correlation Networks, Stock Returns, Idiosyncratic Risk, Network Spillovers

Authors’ individual contribution: the author is responsible for all the contributions to the paper according to CRediT (Contributor Roles Taxonomy) standards.

JEL Classification: G01, G12, C58

Received: 02.08.2019
Accepted: 23.09.2019
Published online: 24.09.2019

How to cite this paper: Todorova, Z. (2019). Firm returns and network centrality. Risk Governance and Control: Financial Markets & Institutions, 9(3), 74-82.