Interconnectedness and systemic risk in financial networks: Fresh evidence from India

Nodes in a financial network, such as banks, cannot assess the true risks associated with lending to other nodes in the network, unless they have full information on the riskiness of all other nodes. These risks can be estimated by using network metrics (as DebtRank) of the interbank liability network. With a simple agent based model we show that systemic risk in financial networks can be drastically reduced by increasing transparency, i.e. making the DebtRank of individual banks visible to others, and by imposing a rule, that reduces interbank borrowing from systemically risky nodes. This scheme does not reduce the efficiency of the financial network, but fosters a more homogeneous risk-distribution within the system in a self-organized critical way. The reduction of systemic risk is due to a massive reduction of cascading failures in the transparent system. A regulation-policy implementation of the proposed scheme is discussed.

Financial markets are exposed to systemic risk (SR), the risk that a major fraction of the system ceases to function, and collapses. It has recently become possible to quantify SR in terms of underlying financial networks where nodes represent financial institutions, and links capture the size and maturity of assets (loans), liabilities and other obligations, such as derivatives. We demonstrate that it is possible to quantify the share of SR that individual liabilities within a financial network contribute to the overall SR. We use empirical data of nationwide interbank liabilities to show that the marginal contribution to overall SR of liabilities for a given size varies by a factor of a thousand. We propose a tax on individual transactions that is proportional to their marginal contribution to overall SR. If a transaction does not increase SR, it is tax-free. With an agent-based model (ABM) (CRISIS macro-financial model), we demonstrate that the proposed ‘Systemic Risk Tax’ (SRT) leads to a self-organized restructuring of financial networks that are practically free of SR. The SRT can be seen as an insurance for the public against costs arising from cascading failure. ABM predictions are shown to be in remarkable agreement with the empirical data and can be used to understand the relation of credit risk and SR.

The drivers of systemic risk in financial networks: a data-driven machine learning analysis

The 2008 financial crisis has been attributed by policymakers to “excessive complexity” of the financial network, especially due to financial derivatives. In a financial network, financial institutions (“banks” for short) are connected by financial contracts. As banks depend on payments from contracts with other banks to cover their own obligations, such a situation creates systemic risk, i.e., the risk of a financial crisis. Some of the contracts are financial derivatives, where an obligation to pay depends on another variable. In this thesis, I study in what sense derivatives make a financial network fundamentally “more complex” compared to one without derivatives. I capture the notion of “complexity” formally using tools from finance and theoretical computer science. I reveal new kinds of systemic risk that arise in financial networks specifically because of derivatives and I discuss the impact of recent regulatory policy. I first focus on a type of derivative called a credit default swap (CDS), in which the writer insures the holder of the contract against the default (i.e., bankruptcy) of a third party, the reference entity. I show that, when the reference entity is another bank, then such CDSs introduce a new kind of systemic risk arising from what I call default ambiguity. Default ambiguity is a situation where it is impossible to decide which banks are in default following a shock (i.e., a loss in banks’ assets). At a technical level, I show that the clearing problem may have no solution or multiple incompatible solutions. In contrast, without CDSs, a unique canonical solution always exists. I then demonstrate that increased “complexity” due to CDSs also manifests as computational complexity. More in detail, I show that the clearing problem leads to NP-complete decision and PPAD-complete approximation problems if CDSs are allowed. This implies a fundamental barrier to the computational analysis of these networks, specifically to macroprudential stress testing. Without CDSs, the problems are either trivial or in P. I study the impact of different regulatory policies. My main result is that the aforementioned phenomena can be attributed to naked CDS positions. In a final step, I focus on one specific regulatory policy: mandatory portfolio com- pression, which is a post-trade mechanism by which cycles in the financial network are eliminated. While this always reduces individual exposures, I show that, surprisingly, it can worsen the impact of certain shocks. Banks’ incentives to compress may further be misaligned with social welfare. I provide sufficient conditions on the network structure under which these issues are eliminated. Overall, my results in this thesis contribute to a better understanding of systemic risk and the effects of regulatory policy.

We provide an overview of the relationship between financial networks and systemic risk. We present a taxonomy of different types of systemic risk, differentiating between direct externalities between financial organizations (e.g., defaults, correlated portfolios and firesales), and perceptions and feedback effects (e.g., bank runs, credit freezes). We also discuss optimal regulation and bailouts, measurements of systemic risk and financial centrality, choices by banks' regarding their portfolios and partnerships, and the changing nature of financial networks.

We provide an overview of the relationship between financial networks and systemic risk. We present a taxonomy of different types of systemic risk, differentiating between direct externalities between financial organizations (e.g., defaults, correlated portfolios, fire sales), and perceptions and feedback effects (e.g., bank runs, credit freezes). We also discuss optimal regulation and bailouts, measurements of systemic risk and financial centrality, choices by banks regarding their portfolios and partnerships, and the changing nature of financial networks.

This paper has two main objectives: first, to provide a formal definition of endogenous systemic risk that is firmly grounded in equilibrium dynamics of financial networks; and second, to construct a discounted stochastic game (DSG) model of the emergence of equilibrium network dynamics that fully takes into account the feedback between network structure, strategic behavior, and risk. Based on our definition of systemic risk we also propose a formal definition of tipping points. Using these tools we then provide a strategic approach to making global assessments of systemic risk in financial networks. Our approach is based on three key facts: (1) the equilibrium dynamics which emerge from the game of network formation generate finitely many disjoint basins of attraction as well as finitely many ergodic measures (implying that, starting from any financial network, in finite time with probability one, the dynamic sequence of financial networks arrives at one of these basins and once there stays there), (2) each basin of attraction is homogenous with respect to its default characteristics (meaning that if a basin contains states having a particular set of defaulted players, then all states contained in this basin have the same set of defaulted players), and (3) the unique profile of basins generated by the equilibrium dynamics carries with it a unique set of tipping points (special states) - and these tipping points provide an early warning of network failure.

Evolutionary systemic risk: Fisher information flow metric in financial network dynamics

The global financial system has become highly connected and complex. It has been proved in practice that existing models, measures and reports of financial risk fail to capture some important systemic dimensions. Recently, advisory boards have been established at a high level, and regulations are being directly targeted at systemic risk. In the same direction, a growing number of researchers employ network analysis to model systemic risk in financial networks. Current approaches are concentrated on interbank payment network flows at national and international levels. This work builds on existing approaches to propose systemic risk assessment at the micro level. The proposed model (partially) captures the systemic aspect of credit risk of bank customers and argues that this part of risk is neglected in both credit scoring models and interbank systemic risk calculations. In particular, the analysis of intra-bank financial risk interconnections is introduced by examining the real case of a ‘receivables-as-collateral’ network. In the data set examined, an initial failure of five customers, representing 17 per cent of the network’s total value, results in the subsequent failure of 15 customers, representing 41 per cent of the total value. The author’s model could be complementary to existing credit scoring models that account for mainly idiosyncratic customers’ financial profiles. Private or public organisations could further elaborate the specification to include a wider range of parameters, such as transactions on contracts, assets and cash flows. Identification of the systemic risk could be beneficial for a business entity in assessing unexplored sources of risks in its portfolios of assets and customers. Understanding and modelling these ‘particles’ of risk could enable more realistic monitoring and the provision of early warning messages for market supervising bodies.

In highly connected financial networks, the failure of a single institution can cascade into additional bank failures. This systemic risk can be mitigated by adjusting the loans, holding shares, and other liabilities connecting institutions in a way that prevents cascading of failures. We are approaching the systemic risk problem by attempting to optimize the connections between the institutions. In order to provide a more realistic simulation environment, we have incorporated nonlinear/discontinuous losses in the value of the banks. To address scalability challenges, we have developed a two-stage algorithm where the networks are partitioned into modules of highly interconnected banks and then the modules are individually optimized. We developed a new algorithms for classical and quantum partitioning for directed and weighed graphs (first stage) and a new methodology for solving Mixed Integer Linear Programming problems with constraints for the systemic risk context (second stage). We compare classical and quantum algorithms for the partitioning problem. Experimental results demonstrate that our two-stage optimization with quantum partitioning is more resilient to financial shocks, delays the cascade failure phase transition, and reduces the total number of failures at convergence under systemic risks with reduced time complexity.

This chapter discusses how the widespread adoption of blockchain technology (distributed ledgers) might contribute to solving a larger class of economic problems related to systemic risk, specifically the degree of systemic risk in financial networks (ongoing credit relationships between parties). The chapter introduces economic network theory, drawing from Konig and Battiston (2009). Then, Part I develops payment network analysis (analyzing immediate cash transfers) in the classical payment network setting (Fedwire (Soramaki 2007)) synthesized with the cryptocurrency environment (Bitcoin (Maesa 2017), Monero (Miller 2017), and Ripple (Moreno-Sanchez et al. 2018)). The key finding is that the replication of network statistical behavior in cryptographic networks indicates the robust (not merely anecdotal) adoption of blockchain systems. Part II addresses balance sheet network analysis (ongoing obligations over time), first from the classical sense of central bank balance sheet network analysis developed by Castren (2009, 2013), Gai and Kapadia (2010), and Chan-Lau (2010), and then proposes how blockchain economic networks might help solve systemic risk problems. The chapter concludes with the potential economic and social benefits of blockchain economic networks, particularly as a new technological affordance is created, algorithmic trust, to support financial systems.

The COVID-19 pandemic precipitated a surge in the non-performing assets held by financial institutions, elevating systemic risk in financial networks. Therefore, developing strategies to alleviate this risk, with a focus on non-performing assets, has become a research area of interest. Supported by policies related to the Chinese insurance market, this study proposes the establishment of a non-performing assets disposal fund backed by insurance capital. This fund will invest in the non-performing assets of financial institutions with the aim of mitigating systemic risk. Using a linear threshold model, we identify an asymptotically optimal scheme for disposing of non-performing assets. Additionally, we construct a payment model integrated with non-performing assets, from which we derive an optimal payment and clearing strategy. Our research also proposes a robust set of criteria to assist regulators in determining whether to use the non-performing assets disposal fund. To demonstrate the efficacy of the fund in reducing systemic risk, we conduct simulations and analyze data from the Chinese interbank financial network. Through this rigorous analysis, we confirm the role of the fund in enhancing the stability of the financial system.

We consider default by firms that are part of a single mechanism. The obligations of all firms within the system are determined simultaneously in a fashion consistent with the priority of debt claims and the limited liability of equity. We first show, via a fixed-point argument, that there always exists a clearing payment that clears the obligations of the members of the system; under mild regularity conditions, this vector is unique. Next, we develop an algorithm that both clears the financial network in a computationally efficient fashion and provides information on the systemic risk faced by individual system firms. Finally, we produce qualitative comparative statics for financial networks. These comparative statics imply that, in contrast to single-firm results, unsystematic, nondissipative shocks to the system will lower the total value of the network and may lower the value of the equity of some of the individual network firms.

A systemically important firm could be too-connected-to-fail and/or too-important-to-fail, two properties which centrality measures and community detection methods can capture respectively. This paper examines the performance of these measures in a variance decomposition global financial network. Too-connected-to-fail risk and vulnerability rankings are quite robust to the choice of centrality measure. The PageRank centrality measure, however, does not seem as suitable for assessing vulnerabilities. Two community identification methods, edge betweenness and the map equation (Infomap) were used to identify systemic communities, which in turn capture the too-important-tofail dimension of systemic risk. The first method appears more robust to di erent weighting schemes but tends to isolate too many firms. The second method exhibits the opposite characteristics. Overall, the analysis suggests that centrality measures and community identification methods complement each other for assessing systemic risk in financial networks.

A systemically important firm could be too-connected-to-fail and/or too-important-to-fail, two properties which centrality measures and community detection methods can capture respectively. This paper examines the performance of these measures in a variance decomposition global financial network. Too-connected-to-fail risk and vulnerability rankings are quite robust to the choice of centrality measure. The PageRank centrality measure, however, does not seem as suitable for assessing vulnerabilities. Two community identification methods, edge betweenness and the map equation (Infomap) were used to identify systemic communities, which in turn capture the too-important-to-fail dimension of systemic risk. The first method appears more robust to different weighting schemes but tends to isolate too many firms. The second method exhibits the opposite characteristics. Overall, the analysis suggests that centrality measures and community identification methods complement each other for assessing systemic risk in financial networks.