Abstract
We propose a dynamic model for systemic risk using a bipartite network of banks and assets in which the weight of links and node attributes vary over time. Using market data and bank asset holdings, we are able to estimate a single parameter as an indicator of the stability of the financial system. We apply the model to the European sovereign debt crisis and observe that the results closely match real-world events (e.g., the high risk of Greek sovereign bonds and the distress of Greek banks). Our model could become complementary to existing stress tests, incorporating the contribution of interconnectivity of the banks to systemic risk in time-dependent networks. Additionally, we propose an institutional systemic importance ranking, BankRank, for the financial institutions analyzed in this study to assess the contribution of individual banks to the overall systemic risk.
Introduction
Traditional economic models often rely on "representative agents," effectively ignoring the complex web of connections between individual institutions. While newer "network" models have looked at interbank lending, the 2011 European sovereign debt crisis highlighted a different channel of contagion: shared portfolio overlap.
When multiple banks hold the same assets (like Greek or Italian government bonds), a price drop in those assets forces banks to sell to maintain their regulatory capital. This "fire sale" further depresses prices, creating a feedback loop that can crash the entire system. This paper introduces a dynamic bipartite network model to simulate these collapses and identify a critical stability threshold.
Methodology and Data
The financial system is represented as a weighted bipartite graph where one set of nodes represents banks and the other represents assets, specifically sovereign bonds. Unlike static models, the link weights—representing the amount of an asset a bank holds—evolve based on market conditions.
The study utilizes holdings data from the European Banking Authority (EBA) "Transparency Exercises" for 119–136 major European financial institutions (2006–2020). It focuses on debt from the GIIPS countries (Greece, Italy, Ireland, Portugal, and Spain), combined with daily stock prices and bond yields to track systemic health.
The Stability Parameter and Phase Transitions
A core contribution of this model is the stability parameter $\gamma$, defined as the product of market liquidity ($\alpha$) and investor "panic" ($\beta$). Using second-order differential equations to model the momentum of financial shocks, the researchers identified a phase transition between a stable regime ($\gamma < 1$) and an unstable regime ($\gamma > 1$).
In the unstable regime, even minor shocks can lead to exponential crashes. The model's accuracy was confirmed retroactively; $\gamma$ spiked above 1 for Greece, Portugal, and Ireland precisely when they required international bailouts, while stable economies like Germany and France remained consistently below the threshold.
Assessing Systemic Importance: BankRank
The authors introduce BankRank, a new centrality measure to rank an institution's systemic importance. Unlike traditional metrics like Eigenvector centrality, which often fail in unstable markets, BankRank calculates the total potential loss to the entire network if a specific bank reaches its "survival threshold." This allows for the identification of specific institutions that exert disproportionate downward pressure on asset prices during a crisis, even if they are not the largest banks by total assets.
Conclusion and Future Directions
The model demonstrates that network topology—specifically who holds what—is a primary driver of systemic risk. "Home bias" in bank holdings creates dangerous risk clusters that are more volatile than randomized portfolios. While currently focused on sovereign bonds, future work could expand this framework to include diverse asset classes like real estate and derivatives, providing a more holistic tool for financial regulators to assess systemic vulnerability.
Summary updated