Extending Maximum-Entropy Interbank Reconstruction to a Multi-Country Framework with Cross-Border Exposures

Sbaraglia, Simone

First

;Guo, Pin

Second

;Zedda, Stefano

Last

2026-01-01

---

Abstract

Estimating bilateral interbank exposures is essential for assessing systemic risk and contagion in global banking systems. We propose a reconstruction framework that extends the maximum-entropy approach to a multi-country setting by integrating domestic balance sheet total interbank exposure values with country-level aggregate cross-border exposures. The method distributes each country’s external positions across its resident banks and then reconstructs the domestic interbank matrices under maximum-entropy constraints, yielding a globally consistent exposure matrix. Cross-border aggregates—such as those published by the BIS—enter the framework only as optional external constraints rather than as objects of empirical validation. The resulting dual-layer interbank network improves the internal coherence of multi-country exposure estimates and provides a consistent input for contagion simulations and systemic risk analysis.