The researchers model a bank that must balance two competing objectives. Holding reserves provides protection against unexpected deposit withdrawals, but idle money generally earns less than funds invested through lending or other profitable activities. A bank that keeps a large reserve is better positioned to meet immediate claims from depositors, but it may sacrifice returns. A bank that lends more aggressively can increase expected profits, yet it becomes vulnerable if withdrawals arrive faster or in greater volume than anticipated. In the model, each bank first chooses its reserve level while considering the possibility of a liquidity shock. It then evaluates whether joining a network of banks that can lend to one another would produce better outcomes than remaining financially independent. This two-stage structure allows the study to capture an important feature of banking markets: institutions choose both their individual safety margins and the relationships that may substitute for those margins.

The model focuses on liquidity withdrawals, represented as a sudden demand for cash that can force a bank to liquidate assets or seek assistance. A bank with reserves greater than the withdrawal can survive without outside help. A bank whose reserves are insufficient may rely on credit from connected institutions, provided those partners have enough liquidity available and are willing to lend. If support cannot be obtained, the bank may fail. This framework does not treat interbank links as automatically beneficial. Each connection can provide insurance against an isolated shock, but it can also transmit stress when several institutions need cash at the same time. The outcome depends on the interaction between reserve choices, the distribution of deposits across banks, the structure of the network and the incentives created by access to shared liquidity.

The most striking mechanism identified by the researchers is a form of strategic free-riding. Once a bank knows that it can obtain assistance from other institutions, it may reduce its own reserve holdings. That decision can be individually rational: reserves are costly because they represent funds that could otherwise generate income. Yet when many banks make the same calculation, the network’s collective safety cushion becomes thinner. Each institution benefits from the protection provided by others while attempting to avoid bearing the full cost of maintaining liquidity. The result is a tension between private incentives and system-wide resilience. Risk-sharing can lower the expected damage from a shock for an individual bank, but free-riding can reduce the reserves available to absorb shocks across the network, potentially increasing the probability of failures and lowering expected profits.

According to the analysis, the relationship between network size and market performance is not simply positive or negative. When only a relatively small number of banks are connected, adding participants can improve outcomes because the benefits of diversification and liquidity sharing are substantial. A withdrawal affecting one institution may be met by several partners whose reserves remain intact, reducing the need for costly asset sales or emergency intervention. As the network expands, however, the incentive to economize on individual reserves becomes stronger. Every additional connection can make a bank feel less responsible for maintaining its own liquidity, while the number of institutions that may simultaneously seek support also increases. The model therefore produces a rise-and-fall pattern in expected profits: performance improves at first, reaches a peak and then declines as free-riding begins to dominate risk-sharing.

This result points to an optimal range of connectivity rather than a universal case for larger and denser networks. The study finds that relatively small networks can be Pareto optimal, meaning that no participating bank can be made better off without making at least one other participant worse off. In practical terms, a compact network may deliver enough insurance to manage isolated withdrawals while limiting the erosion of individual reserves. A much larger network can appear safer because more institutions are linked, but its apparent strength may be misleading if every bank has reduced its own liquidity buffer. Network size, in this view, is not a measure of resilience by itself. The quality of resilience depends on whether institutions retain sufficient capacity to withstand stress before drawing on their partners.

The researchers also examine networks containing banks of different sizes. Traditional views of interbank relationships might suggest that institutions with similar deposit bases and balance sheets would have the strongest reasons to connect. The model instead indicates that smaller and larger banks may both benefit from forming links even when their deposit sizes differ considerably. A large bank may provide liquidity because it has greater reserves or more diversified funding, while a smaller bank may value access to that support because it is more exposed to a localized withdrawal shock. Such relationships can help explain core-periphery structures in banking, in which a relatively small number of highly connected institutions occupy the center of the network and interact with many smaller banks around them. The same structure, however, can concentrate risk if peripheral institutions become dependent on central providers or if central banks are hit by demands from many partners at once.

The findings also raise questions about the role of government guarantees. When banks believe that authorities will protect depositors, creditors or major institutions during a crisis, they may perceive the consequences of failure as less severe. That implicit safety net can encourage greater risk-taking, lower reserve holdings and more extensive interbank connections. Such behavior may remain profitable during calm periods, when liquidity can be obtained easily, but it can amplify instability during a systemic shock. The researchers argue that capital requirements and related safeguards should be designed with these strategic incentives in mind. Regulation that focuses only on the balance sheet of each bank may miss the way one institution’s reserve decision affects the behavior and vulnerability of its partners. Requirements that preserve adequate buffers could reduce free-riding and prevent large networks from becoming excessively interconnected.

The study is theoretical and relies on computational modeling rather than observations from a particular banking crisis, so it does not claim that every interbank network has the same ideal size or that connections inevitably increase danger. Its contribution is to show how market efficiency can emerge from a trade-off between insurance and incentives. Interbank lending remains a potentially valuable mechanism for absorbing unexpected liquidity withdrawals, especially when shocks are limited and partners are sufficiently strong. But the model warns that connectivity can produce diminishing returns when banks treat shared liquidity as a substitute for prudent reserve management. For policymakers, the message is clear: a safer financial network is not necessarily the one with the most links. It may be the one in which banks share risk while still maintaining enough independent liquidity to survive when everyone needs help at the same time. The research offers a mathematical explanation for why cooperation can stabilize markets at one scale and destabilize them at another, turning the architecture of interbank finance into a critical question for the future of banking regulation.

Subject of Research: Not applicable

Article Title: Interbank network and market efficiency

Web References: https://doi.org/10.1016/j.risk.2026.100059

References: Yu, T., He, X.-Z. (Tony), & Zhang, N. “Interbank network and market efficiency.” Risk Sciences. DOI: 10.1016/j.risk.2026.100059

Image Credits: Yu, T., He, X.-Z. (Tony), & Zhang, N.

Keywords: interbank lending, banking networks, liquidity risk, financial stability, free-riding, bank reserves, market efficiency, core-periphery networks, capital requirements, computational modeling

Over the last decade, the growing prominence of shadow banking in China has sparked intense scrutiny due to its role in channeling liquidity outside the traditional banking framework. Collateral-based monetary instruments, widely employed by Chinese policymakers to regulate liquidity and credit growth, have until now been insufficiently examined in terms of their second-order effects on shadow banking and systemic risk. This study decisively bridges that gap, offering a detailed empirical analysis that contextualizes collateral policy within the broader financial ecosystem.

Crucially, the research adopts an innovative methodological framework integrating SHAP (SHapley Additive exPlanations) values within a Bayesian-enhanced Extreme Gradient Boosting (XGBoost) model. This approach transcends conventional econometric constraints by unraveling the intricate, often opaque pathways through which monetary transmission influences off-balance-sheet activities and risk accumulations. The combined interpretability and predictive power of SHAP-Bayesian-XGBoost mark a significant step forward in econometric analysis of complex, adaptive financial systems.

The findings reveal that collateral monetary policy distorts liquidity allocation in ways that disproportionately stimulate shadow banking entities. These distortions stem from the preferential liquidity access granted to entities holding eligible collateral, which skews funding patterns toward non-primary banks and off-balance-sheet vehicles. Such liquidity misallocation feeds shadow banking growth, embedding fragilities that may not be immediately visible within traditional regulatory metrics.

Moreover, the expansion of shadow banking activities under collateral policies is closely linked to elevated bank risk exposure. These risks manifest through increased leverage, maturity mismatches, and interconnectedness that amplify systemic vulnerabilities. Importantly, the study highlights that non-primary banks exhibit heightened sensitivity to these liquidity distortions, suggesting heterogeneity in risk transmission across different banking institutions based on their market positioning and balance sheet structures.

The research also underscores the salutary effects of targeted regulatory interventions, particularly the 2018 New Asset Management (NAM) Regulation. This policy milestone appears to have effectively curbed the stimulative impact of collateral monetary policy on shadow banking expansion and associated risk buildup. By tightening oversight and enforcing risk controls, the NAM Regulation represents a critical inflection point, demonstrating that well-calibrated macroprudential measures can effectively counterbalance adverse policy spillovers.

This revelation has profound implications for emerging financial markets globally, where collateral frameworks form cornerstone elements of monetary toolkit but also carry latent risks of systemic distortions. The study’s findings shed light on the potential unintended consequences that collateralized liquidity provision may engender, challenging central banks to rethink collateral eligibility criteria, monitoring mechanisms, and the broader regulatory ecosystem to mitigate shadow banking excesses and systemic risk.

Furthermore, this research makes a valuable methodological contribution by illustrating how machine learning models, particularly those enhanced for interpretability, can unlock previously inscrutable relationships in financial policymaking. The fusion of SHAP values with Bayesian inference and XGBoost captures nonlinearities and interaction effects that evade traditional regression techniques, offering policymakers a powerful diagnostic tool for real-time risk assessment in complex monetary environments.

The implications extend beyond regulatory design to practical considerations within banking operations. Financial institutions, especially smaller and non-primary banks, must recalibrate their risk management frameworks to better monitor their exposure to shadow banking activities influenced by collateral-based liquidity strategies. Enhanced transparency, stress testing, and scenario analysis will be vital components in safeguarding institutional resilience in the face of evolving monetary policy landscapes.

In summary, this pioneering investigation peels back layers of complexity surrounding collateral monetary policy in China, elucidating its role as a catalyst for shadow banking growth and bank risk intensification. By merging cutting-edge machine learning with nuanced financial theory, the research not only enriches academic discourse but also furnishes actionable insights for regulators and market participants confronting the dual challenges of fostering liquidity and maintaining financial stability in fast-evolving markets.

The study serves as a critical reminder that monetary policy design must account for the broader financial fabric it penetrates, recognizing institutional heterogeneity and market adaptation. The success of the New Asset Management Regulation in mitigating adverse consequences signals that carefully calibrated interventions can restore balance, but vigilance remains paramount as financial ecosystems continue to innovate and evolve under policy influences.

As policymakers globally grapple with the complexities of shadow banking and systemic risk, this research highlights the indispensable value of interdisciplinary approaches that combine economic theory, regulatory insight, and advanced computational methods. It offers a blueprint for analyzing monetary policy transmission in environments where traditional assumptions about market behavior and liquidity channels no longer hold.

By quantifying the linkages between collateral policies, shadow banking activities, and emerging risks with unparalleled precision, the study paves the way for more informed macroprudential strategies. It advocates for an integrated regulatory stance that harmonizes monetary policy goals with financial stability imperatives, ensuring sustainable growth trajectories for emerging markets undergoing rapid financial deepening and innovation.

Ultimately, the findings underscore that while collateral monetary policies remain powerful instruments for liquidity management, their design and implementation necessitate careful calibration and ongoing empirical monitoring. Through this lens, the research enriches understanding of the shadow banking phenomenon and underscores the critical role of regulation in maintaining a resilient and transparent financial system.

Subject of Research: Monetary policy transmission, shadow banking dynamics, and bank risk in China.

Article Title: Collateral monetary policy, shadow banking and bank risk evidence from China.

News Publication Date: 18-Apr-2025.

Web References: http://dx.doi.org/10.1108/CFRI-07-2024-0423

Keywords: Monetary policy, Shadow banking, Bank risk, Collateral policy, Machine learning, SHAP, XGBoost, Financial stability, New Asset Management Regulation, Liquidity distortion, Macroprudential regulation, China financial system.