Explainable Graph Neural Networks for Interbank Contagion Surveillance: A Regulatory-Aligned Framework for the U.S. Banking Sector

arXiv cs.LG / 4/17/2026

📰 NewsSignals & Early TrendsIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper introduces ST-GAT, a spatial-temporal graph attention network framework designed for explainable early-warning detection of bank distress and regulatory-aligned macro-prudential surveillance of interbank contagion in the U.S.
ST-GAT reconstructs a dynamic directed weighted exposure graph for 8,103 FDIC-insured institutions using bilateral exposures derived from publicly available FDIC Call Reports via maximum entropy estimation across 58 quarterly snapshots (2010Q1–2024Q2).
The framework reports the best AUPRC among GNN architectures (0.939 ± 0.010), narrowly behind XGBoost (0.944), suggesting competitive performance for contagion monitoring.
Ablation and interpretability analyses show that the BiLSTM temporal component improves AUPRC by +0.020 and that temporal attention weights follow a monotonic decreasing pattern, while permutation importance highlights ROA and the NPL ratio as key predictors.
The study emphasizes reproducibility and policy relevance by releasing code and results, and by relying only on publicly available FDIC Call Reports and FRED series.

Abstract

The Spatial-Temporal Graph Attention Network (ST-GAT) framework was created to serve as an explainable GNN-based solution for detecting bank distress early warning signs and for conducting macro-prudential surveillance of the interbank system in the United States. The ST-GAT framework models 8,103 FDIC insured institutions across 58 quarterly snapshots (2010Q1-2024Q2). Bilateral exposures were reconstructed from publicly available FDIC Call Reports using maximum entropy estimation to produce a dynamic directed weighted graph. The framework achieves the highest AUPRC among all GNN architectures (0.939 +/- 0.010), trailing only XGBoost (0.944). Ablation analysis confirms the BiLSTM temporal component contributes +0.020 AUPRC; temporal attention weights exhibit a monotonically decreasing pattern consistent with long-run structural vulnerability weighting. Permutation importance identifies ROA (0.309) and NPL Ratio (0.252) as dominant predictors, consistent with post-mortem analyses of the 2023 regional banking crisis. All data are publicly available FDIC Call Reports and FRED series; all code and results are released.