The Accounting Network: How Financial Institutions React to Systemic Crisis.

Area of Science:

• Financial Economics
• Network Science
• Computational Finance

Background:

• Network Theory is increasingly used to understand systemic financial crises.
• Previous studies highlight how network topology and dynamics can trigger crises.

Purpose of the Study:

• To introduce and analyze the "Accounting Network" derived from companies' financial statements.
• To investigate the role of accounting data in network structures preceding the 2008 financial crisis.

Main Methods:

• Constructed an Accounting Network from worldwide bank financial statements (2001-2013).
• Applied vector similarity techniques and a "Quality Ratio" for network construction accuracy.
• Utilized Louvain community detection and Principal Component Analysis (PCA).

Main Results:

• Identified distinct regional bank communities (e.g., Japanese, US clusters) and a mixed supranational community.
• PCA revealed key economic factors driving community heterogeneity.
• The Accounting Network signature clearly indicated the 2008 financial crisis.

Conclusions:

• Accounting Networks offer a novel perspective on financial crisis prediction.
• The study demonstrates the utility of financial statement analysis within a network framework.
• Future work can refine Accounting Networks for enhanced financial statement analysis and crisis detection.