Analysis of the structural vulnerability of the Hungarian interbank market network with simulation techniques

Abstract

The article focuses on systemic risk and contagion in financial networks. Systemic risk, as a key concept of financial stability, came to the forefront of economic studies in the last two decades of the twentieth century, and only then did the general concept of systemic risk becomes clearer. Although there are many different definitions of systemic risk, they all have in common that they refer to the fragility and instability of financial systems. According to the relevant literature (De Bandt - Hartmann [2000], Lublóy [2004]) the systemic risk is the risk that a series of events triggered by a particular event adversely affects one or more financial institutions or markets over time.

The aim of the article is the application of a model published by the European Central Bank (Montagna-Kok [2016]) on Hungarian data which can identify the systemically important financial institutions and potentially vulnerable interbank market network structures to external shocks. The novelty of the model lies in the fact that interbank market participants were examined through a multi-layered network which can take into account that in reality financial institutions can connect to each other through several markets at the same time. The first layer of the multi-layered network embodies interbank counterparty risk on long term interbank market while the second layer try to capture the funding risk on short term interbank market. The third layer is meant to reproduce the network of overlapping portfolios namely when two banks invest in the same mark-to-market financial securities. Then their balance sheets can be correlated which means that when one bank is forced to sell some securities and the resulting price decline from such fire sale will affect the balance sheets of the banks which hold the same asset. In addition, the model relies on agent-based simulation techniques namely in case of an exogenous shock to the system, through a predefined set of rules and algorithms financial institutions can make decision over a number of periods to renew their short-term interbank loans and sell their securities to fulfill their obligations and regulatory requirements. With appropriate modifications to the model we identified systemically important Hungarian financial institutions which failure could cause most institutions to fail and we also defined those conditions that make networks potentially vulnerable to external shocks.

The data of Hungarian banks and banking groups used in the model comes from multiple sources. Balance sheet data was collected from the Aranykönyv published by Central Bank of Hungary as of 12.31.2019 while the source of the total assets and liabilities from the interbank market was the supplementary annexes of income statements published by Hungarian banks and banking groups individually as of 12.31.2019. Risk Weighted Assets (RWA) and Capital Adequacy Ratios (CAR) of the Hungarian banks and banking groups were obtained from the Pillar 3 disclosures published by the market participants individually.

According to Montagna-Kok [2016], the systemic importance of a financial institution was measured by how many other financial institutions can fail due to its bankruptcy. Based on 100 thousand simulation Hungarian banks/banking groups can be ranked according to the maximum number of failures of other banks caused by the bankruptcy of a given bank. According to the results obtained from the 19 examined financial institutions only a few can cause further failures in the system and these events can be considered rare which are typically caused by a special circumstance or a network structure. Based on that the bankruptcy of a fraction of the banks examined can trigger further bankruptcies, hence it is advisable to analyze those network structures which are the most sensitive to the failure of the largest systemically important banks. Montagna-Kok [2016] measured the fragility of a multi-layered network as the average of bankruptcies caused by the failures of network members. In our calculation 200 thousand multi-layered networks were simulated and the most fragile ones were identified using the referred indicator. According to the results in fragile networks both long term and short term interbank lending are more concentrated and long run key players are likely to provide liquidity to smaller institutions to a greater extent. In order to analyze the role of the number of securities in the model the degree of overlap and the maximum bankruptcy indicators were calculated in

numerous networks using different number of securities. According to the results the higher the maximum number of bankruptcies, the fewer securities we use in the model.