Dynamic Topic Networks to Evaluate Systemic Risk in Financial Markets

Abstract

The study proposes a Dynamic Topic Network (DTN) approach to assess systemic risk in financial markets, utilizing a combination of topic modeling and network analysis. By employing Latent Dirichlet Allocation (LDA) to analyze news articles, the study extracts topics that are then used to construct topic similarity networks over time. The results obtained highlight the interconnectedness of topics, allowing for the correlation of abnormal behaviors with volatility in financial markets. Using the 2015–2016 stock market selloff and the COVID-19 pandemic as case studies, the study demonstrates how the DTN approach can provide insights into abnormal movements in the Dow Jones Industrial Average and predict the gradual recovery of the market following such events. From a risk management perspective, the analysis can be conducted on a daily basis with new data to predict real-time systemic risk in financial markets, providing valuable information for decision-makers in managing financial stability and mitigating potential losses