[Paper Review] Networks of Economic Market Interdependence and Systemic Risk

Dion Harmon, Blake C. Stacey+2 more|arXiv (Cornell University)|Nov 16, 2010

Complex Systems and Time Series Analysis14 references20 citations

TL;DR

This paper analyzes financial market correlations from 2003 to 2008 to reveal how inter-sectoral dependencies, especially through the financial sector, amplify systemic risk during economic downturns. Using network analysis of daily stock returns, it shows that financial institutions act as cross-sector linkers, increasing cascading failure risks—suggesting that regulatory firewalls between financial services and non-related sectors could reduce systemic risk without hindering growth.

ABSTRACT

The dynamic network of relationships among corporations underlies cascading economic failures including the current economic crisis, and can be inferred from correlations in market value fluctuations. We analyze the time dependence of the network of correlations to reveal the changing relationships among the financial, technology, and basic materials sectors with rising and falling markets and resource constraints. The financial sector links otherwise weakly coupled economic sectors, particularly during economic declines. Such links increase economic risk and the extent of cascading failures. Our results suggest that firewalls between financial services for different sectors would reduce systemic risk without hampering economic growth.

Motivation & Objective

• To understand how economic market interdependence evolves over time and contributes to systemic risk.
• To identify the role of the financial sector in linking otherwise weakly connected economic sectors during market stress.
• To evaluate whether regulatory firewalls between financial services and non-related sectors could reduce systemic risk.
• To assess the impact of deregulation, such as the repeal of Glass-Steagall, on financial network stability and crisis propagation.
• To provide a data-driven basis for systemic risk regulation using network science and correlation dynamics.

Proposed method

• Constructed a dynamic network of 500 major corporations using daily stock return correlations from 2003 to 2008.
• Applied Pearson correlation to log-returns of adjusted closing prices to define links between firms.
• Formed a network with the highest 6.25% of pairwise correlations in each year, with temporal continuity between years.
• Used t-statistics to test statistical significance of within-sector clustering and between-sector linkages.
• Identified network communities and changes in connectivity using self-clustering and merging statistics.
• Incorporated external indices (oil, bonds) as nodes to assess macroeconomic influence on network structure.

Experimental results

Research questions

• RQ1 How do correlations between financial firms and economic sectors change during rising and falling markets?
• RQ2 To what extent does the financial sector act as a bridge between otherwise loosely connected economic sectors?
• RQ3 What is the statistical significance of sector-specific clustering and cross-sector linkages over time?
• RQ4 How do resource constraints and commodity price surges correlate with systemic risk propagation?
• RQ5 What role do historical financial regulations, such as Glass-Steagall, play in mitigating systemic risk through network separation?

Key findings

• The financial sector became increasingly central in linking otherwise weakly connected sectors, especially during economic decline, increasing systemic risk.
• The real estate-related financial cluster merged into the broader financial sector by 2007–2008, with p < 0.18, indicating loss of isolation.
• Technology sector self-clustering declined sharply during the 2008 crisis, with a negative slope in self-clustering t-statistic (p < 10^-66), indicating loss of internal coherence.
• The oil sector showed strong self-clustering (p < 10^-13) and increasingly linked to the broader basic materials cluster (p < 10^-45), indicating growing interdependence.
• The oil price surge and housing crisis coincided closely (p < 10^-5), suggesting investment-driven demand rather than use-driven scarcity as a driver of correlation.
• Positive correlations between financial sector performance and commodity prices indicate that rising asset prices propagate risk across sectors, not just through supply chains.