System-wide implications of counterparty credit risk

Prepared by Claudio Barbieri, Maciej Grodzicki, Grzegorz Hałaj and Riccardo Pizzeghello^[1]

Published as part of the Macroprudential Bulletin 15, January 2025.

The aim of this article is to assess the scale and systemic nature of counterparty credit risk (CCR) stemming from banks’ derivatives activities and securities financing transactions. Using supervisory data, along with data collected from the EU-wide stress test carried out by the European Banking Authority in 2023, the article analyses the distribution of CCR across banks. It focuses on the concentration of risk within specific bank business models and products, and on links between the banking and NBFI sectors. It also examines not only the role of collateral in risk mitigation but also its potential negative impact on systemic risk. Exposures to CCR are concentrated in a group of global systemically important banks (G-SIBs) and investment banks, which play a vital intermediation role in European financial markets. Banks’s counterparties mainly operate in the non-bank financial intermediation (NBFI) sector. To quantify systemic risk in a network of CCR exposures, we use stress test techniques to see how widely hypothetical defaults among more vulnerable NBFI counterparties may spread across the banking system. In such an event, banks under European banking supervision may face considerable losses.

1 Introduction

Intermediation in financial markets gives rise to counterparty credit risk in the euro area banking sector. Banks’ operations in financial markets expose them to the risk that, at the time of settlement, the other party to a trade might be unable to pay the due amounts resulting from the trade. This risk is affected by market movements and volatility – which lead to fluctuations in the distribution of the payouts to each counterparty – and also by the creditworthiness of the counterparties. Counterparty risk arises from over-the-counter and exchange-traded derivatives, as well as from securities financing transactions. Counterparty credit risk can be mitigated by netting and margining. Netting means that counterparties settle their trades on a net basis, so that amounts due from individual trades offset each other. Margining means that the amounts due between two counterparties are secured by pledging cash or securities, so that, in the event of a failure of a counterparty, the creditor would have recourse to a specific asset.

The aim of this article is to assess the scale and systemic nature of counterparty credit risk in the euro area banking sector. Building on supervisory banking data, Section 2 discusses the magnitude, origin, and distribution of counterparty credit risk exposures. Section 3 introduces a methodology for simulating the impact caused by the failure of a major counterparty in the banking system, so that the resulting contagion effects can be analysed. This sheds light on the transmission of stress across the banking sector through bilateral counterparty credit risk exposures.

2 Counterparty credit risk in the euro area banking system

The aggregate exposure of the euro area banking system to counterparty credit risk is limited, but exposures are concentrated among G-SIBs and investment banks. In total, counterparty credit risk accounts for about €340 billion, or 3.9%, of significant institutions’ risk-weighted assets (Chart 1, panel a). However, exposures to counterparty credit risk are concentrated in a small group of G-SIBs and specialised banks, including investment banks, asset managers and custodians (Chart 1, panel b). Such banks play a significant role in market-making and enable less complex banks, as well as NBFI entities and the non-financial sector, to manage risks more easily.^[2] Therefore, these banks may be difficult to replace, which could make them systemically important.

Chart 1

Counterparty credit risk exposure in the euro area banking system

a) Total exposure and risk-weighted exposure to counterparty credit risk	b) Contribution of counterparty credit risk exposures to CET1 ratio, by bank business model
(Q1 2022 - Q2 2024, EUR billions)	(Q2 2024, percentage points of CET1 ratio)

Non-centrally cleared interest rate derivatives and securities financing transactions are the main source of counterparty credit risk exposure. Exposure arises mainly from bilaterally cleared transactions (Chart 1, panel a) while trades cleared via central counterparties generate only a small part of counterparty risk exposures. Exposure measures cannot be broken down by risk type.^[3] However, there are data on the market value of products generating counterparty credit risk. Interest rate derivatives, which are the largest segment of derivatives traded by euro area banks, are also the largest contributor to banks’ exposure. Their contribution has increased since the monetary policy normalisation started, as shifts in policy rates have led to an increase in the positive and negative market values of derivatives. Securities financing transactions are the second-largest source, while foreign exchange, equity and other derivatives have a lower weight (Chart 2, panel a). The contribution of commodity derivatives increased in early 2022, driven by the heightened volatility of energy prices.^[4] However, it decreased again quickly as market conditions normalised, and there is no evidence of a lasting shift in trading activity away from central counterparties and towards bilateral clearing.

Chart 2

Interest rate risk derivatives and securities financing transactions are the largest source of counterparty credit risk

a) Market value of transactions exposed to counterparty credit risk	b) Transactions exposed to counterparty credit risk, by bank business model
(Q2 2021 - Q2 2024, EUR billions)	(Q2 2024, EUR billions)

Banks’ business models are reflected in their exposure to counterparty credit risk associated with specific derivatives products. Euro area G-SIBs play a key intermediation role in repo markets. Accordingly, they account for a large majority of counterparty risk exposure arising from securities financing transactions. By contrast, investment banks and universal banks are exposed to counterparty risk through foreign exchange and interest rate risk trades (Chart 2, panel b).

Banks actively mitigate counterparty credit risk, but risk mitigation gives rise to systemic liquidity risks for banks and their counterparties. A vast majority of counterparty credit risk exposure is subject to margin requirements, regardless of the underlying risk type (Chart 3, panel a). About 80% of margins are met with cash (Chart 3, panels b and c). This means that, in the case of large market movements, banks’ counterparties should either have sufficient cash buffers or be able to use other collateral to borrow through repo lines. Even if market movements do not threaten a counterparty’s solvency, financial stability may be at risk if counterparties cannot borrow to meet margin calls and are forced to close out trades. In this way, market stress translates into heightened liquidity needs across the financial system. Recent episodes in the euro area and the United Kingdom show that systemic liquidity risk can be triggered when NBFI counterparties have taken large positions in volatile and illiquid markets, although policy interventions have stopped the emerging negative feedback loops.^[5]

Chart 3

The majority of counterparty credit risk exposures is collateralised with cash

a) Average share of margined exposures based on current positive market value	b) Average collateral composition (derivatives)	c) Average collateral composition (securities financing transactions)
(Q2 2024, shares)	(Q2 2024, shares)	(Q2 2024, shares)

NBFI entities are the main counterparty sector for euro area banks, after “other banks”. In the repo market, euro area G-SIBs intermediate between NBFI counterparties and smaller banks (Chart 4, panel a), while smaller banks interact mainly with other banks. In the derivatives markets, NBFI entities are the most important counterparty sector besides banks themselves (Chart 4, panel b). Repo and derivatives trading are both heavily concentrated within a small group of counterparties. For example, 10% of all repo counterparties account for 80% of the outstanding value of such transactions (Chart 4, panel c). This market structure suggests that a shock originating from large NBFI counterparties could have severe repercussions for the euro area banking system. The next section assesses how such a shock could propagate through the banking system and affect financial stability.

Chart 4

Banks are interconnected with other banks and NBFI entities via counterparty credit risk

a) Outstanding amount of securities financing transactions, by counterparty sector	b) Net notional of derivative transactions of euro area banks, by counterparty sector	c) Concentration of bank lending through securities financing transactions, by counterparty sector
(Q2 2024, EUR billions)	(Q1 2024, EUR trillions)	(Q2 2024, EUR billions)

3 Counterparty credit risk as a contagion channel

In addition to its direct impact on the individual bank, a counterparty default can affect the entire banking system. Exposures related to counterparty credit risk (CCR) may act as a shock transmission and amplification channel in the banking system from both a solvency and a liquidity risk perspective (Markose, 2012). Given the available data, we focus on the solvency channel related to the credit risk of CCR exposures, without considering the impact of market liquidity conditions and NBFI leverage on contagion. Notably, excessive leverage and ineffective liquidity risk management have already been identified as material factors related to the NBFI sector that may amplify stress in the financial system (Mosk et al., 2023). Nevertheless, the impact of a counterparty default in an interconnected system depends not only on direct exposures but also on how interconnected the banking system is through a network of banks’ derivatives contracts. Neglecting to take into account the interconnectedness of banks through CCR exposures may result in the CCR impact on banks’ resilience being underestimated, since banks may not internalise the related systemic risk. In addition, the magnitude of the contagion impact depends on how the structure of the network would change in stressful market conditions if CCR exposures were to grow (Haldane, 2012. NBFI entities’ use of derivatives has been growing, increasing the potential contagion via CCR exposures (see McCaul, 2024, and Paddrik, Rajan and Young, 2020^,).

We use a unique and granular dataset from the EU-wide stress test carried out by the European Banking Authority in 2023 to assess CCR-induced contagion risk. The data cover the ten largest counterparties of each participating bank, excluding those that are centrally cleared. Since the analysis focuses on the credit risk aspects of CCR contagion, the exclusion of central counterparty clearing, which transforms CCR into liquidity risk, is not a limitation in this study. The data include characteristics of CCR exposures under stress market conditions, such as default probabilities and stressed loss given defaults. The data allow us to shed light on the shape of the counterparty network and on how CCR-related losses can be conducive to shock transmission between directly and indirectly connected counterparties.

Several large euro area banks are central to the transmission of shocks in a counterparty network. Chart 5 shows some useful metrics of interconnectedness. First, there are five groups of institutions in this network (panel a). Large banks which are more central in the network structure (measured by “betweenness centrality”) may contribute to knock-on effects as they are connected with each other and with multiple counterparties. NBFI entities appear less central from that perspective, particularly if only CCR exposures are considered. Except for one financial auxiliary (large green dot in Chart 5, panel a), most NBFI entities would generally not transmit shocks from one default of a counterparty to another. Finally, smaller banks are connected through just a few links, so shocks stemming from the NBFI sector may propagate quite broadly in just a few stages. Meanwhile, there are eight “community hubs” (panel b), indicating how many banks may be directly affected by the default of a counterparty (both banks and NBFI entities).

Chart 5

Network structure of the counterparty credit risk exposures

a) Betweenness centrality	b) Community hubs
(which banks can transmit CCR default shocks across the system)	(which banks could be immediately impacted by CCR-related default)

Contagion simulation methodology

The network of CCR exposures allows the indirect impact of counterparty defaults to be quantified. To assess the CCR-induced contagion risk, a simulation tool is applied to determine which links between banks and NBFI entities may contribute towards the propagation of distress across the system of CCR exposures. The default of a counterparty such as an NBFI entity would imply a loss to all banks exposed to that counterparty. Under stressful market conditions, it would be measured as a net stressed exposure, which would mean multiplying the collateral value under stress specified in the 2023 EBA stress test scenario by loss given default. Notably, liquidity implications of the shock are not taken into account, for example it is assumed that no collateral is posted in the run-up to the default. This would be called a first-round impact.

In the second round, higher credit risk requires a revaluation of CCR exposures and leads to further losses to banks. Assuming that these banks would not actively respond to this event, for instance by de-risking the assets, the losses incurred in the first round would instantly lead to a deterioration in the banks’ solvency metrics. Consequently, the banks’ credit quality would drop. This can be captured by the Merton model, which links default probability with the leverage ratio and risk in the asset portfolios of these banks. Even if there were no defaults in the second round, lower credit quality would force counterparties of these banks to reassess their exposures. Banks exposed to other banks with declining credit quality would book higher expected CCR losses, leading to a decline in their capital ratios and, in turn, a change in their credit quality in the CCR network. In accounting terms, changes in counterparties’ probabilities of default are reflected in the credit valuation adjustment (CVA). This combines information on the projected exposures (for instance, expected positive exposures) and changes in counterparties’ credit quality, assuming there is no hedging of the CVA risk.^[6] According to the Merton model, since CVA changes are booked to banks’ profit and loss accounts, they will be reflected in the credit quality of the exposed bank.

The results of the simulations should be seen as a lower bound of the CCR contagion estimates due to data limitations and price-impact externalities. The data collection in the 2023 EBA stress test, given its purpose, covers only the ten largest counterparties and some smaller but material ones, so several NBFI counterparties are potentially missing from the coverage. Even though the data account for the impact of stress on collateral value, they do not capture the possible impact of higher CCR on asset prices, i.e., on externalities which may arise if banks close out trades with defaulting counterparties and liquidate pledged collateral. They also do not account for higher liquidity needs and margin calls triggered by the deteriorating credit quality of the counterparties.

Assessment of contagion in the euro area banking sector

The simulations assume that contagion would be triggered by system-wide defaults of vulnerable NBFI counterparties. We assess contagion risk stemming from the NBFI sector by adapting the approach to CCR taken in the 2023 EBA stress test (see EBA, 2023). Consistent with the stress test methodology, NBFI counterparties with the highest default probabilities reported before application of the stress scenario are considered vulnerable. Unlike in the EU-wide stress test, where default is assumed by each bank in isolation, in our simulation the two most vulnerable NBFI counterparties of each bank are assumed to default for all other banks (even if these NBFI counterparties are ranked third or lower for vulnerability). Therefore, the simulated scenario should be considered a severe system-wide shock scenario, since it assumes that multiple counterparties will default at the same time, based on their vulnerability rankings across all banks in the stress test sample. Overall, this results in defaults by 77 NBFI counterparties. Although the likelihood of such a default event is low, there are precedents: market disruptions related to the COVID-19 outbreak in March 2020 revealed broad-based vulnerabilities in the NBFI sector, as did liquidity pressure from sharply increased volatility in energy markets and from liability-driven investment funds in 2022. Using simulation tools to assess the impact of hypothetical future scenarios can help in preparing adequate policy responses.

Chart 6

Simulated contagion in a scenario of system-wide NBFI counterparty defaults

a) Impact of a system-wide NBFI default scenario across banks with similar business models	b) Tail risk in contagion amplification of CCR losses in a system-wide NBFI default scenario
(basis points of the CET1 ratio)	(x-axis: CET1 ratio depletion, basis points; y-axis: number of banks)

Shocks to common CCR exposures to the NBFI sector may lead to material first-round losses which are unevenly distributed across banks’ business models. Investment banks and asset managers stand out, with around 150 basis points of additional impact on the capital ratio for both business models (see Chart 6, panel a). Banks with these business models tend to have higher stressed CCR exposures relative to their risk exposure amounts than other banks in the sample. G-SIBs and universal banks are less affected, despite holding more sophisticated trading portfolios and larger aggregate exposures to CCR. More retail-focused or specialised banks are much less affected by the shocks. Notably, the impact is far more severe than that in the 2023 EBA stress test scenario, where the average CCR losses amounted to slightly more than 20 basis points of the capital ratio. This significant difference in the impact following the system-wide NBFI default scenario indicates that for some banks, stressed exposures to NBFI counterparties would be very material.

Contagion channels can materially amplify the shocks but only to a limited extent, given the relatively sparse network of CCR exposures. The CVA adjustment following the credit quality deterioration described in the setup of the simulation framework can add about 50 basis points on average to the capital impact. The most material impact is seen for banks with the highest direct CCR losses. In addition, banks with the largest and most sophisticated trading portfolios that rank halfway in the list of most affected business models, namely Asset Managers and Wholesale Lenders, would face the highest relative increase in the capital impact, coming from the contagion losses. However, this rather limited impact confirms the conclusions of the contagion risk assessment based on network centrality measures (see Chart 5).

Chart 7

Links between banks aggregated at business model level and the NBFI sector showing the flow of counterparty credit risk

Thickness of edges proportional to stressed CCR exposure, net of collateral

However, the contagion mechanism may more than double the CCR loss originated in the NBFI sector for some banks. For 16 banks, the second-round CCR losses measured by changes in CVA are higher than the direct CCR losses (see Chart 6, panel b), and five of the banks are at risk of being hit hard (as the CVA add-on quadruples the first-round losses). These banks may require closer scrutiny to identify the likelihood of the most vulnerable NBFI counterparties defaulting and to understand the risk profile of the CCR exposure. In addition, since the contagion loss amplification is related to the structure of the direct connections between banks and NBFI counterparties and within the banking system, only macroprudential tools may be effective in mitigating the risk of exposure to vulnerable NBFI counterparties. These tools could address the concentration of exposures or capital buffers of banks that are more central in the market.

Finally, analysis of the transmission channel shows that investment banks are transmitting the initial NBFI default shocks to other banks in the system. This is related to the fact that investment banks have relatively large exposures to NBFI entities and are also material counterparties of other banks. However, as illustrated in Chart 7, the shocks do not feed back from other banks to the investment banks. This means that the shock transmission path is typically rather short, explaining the limited contagion loss amplification effects.

4 Conclusions

Banks’ derivatives market activities and securities financing transactions give rise to CCR. Although the aggregate exposure of the banking system to this risk is limited to under 4% of total risk-weighted assets, exposures are concentrated within a group of G-SIBs and investment banks, which play a vital intermediation role in European financial markets. Their financial interconnectedness may enable shocks arising from banks’ CCR exposures to NBFI entities to be transmitted to the broader banking system. NBFI entities are the main group of euro area bank counterparties.

CCR is interwoven with liquidity risk, as banks require their counterparties to post collateral, mainly in the form of cash, to cover potential future payouts on derivatives trades and securities financing transactions. When market prices move against banks or their counterparties, these movements have to be reflected in the valuation of CCR and could have repercussions for bank capitalisation. These two mechanisms could amplify market stress and transform it into a systemic concern for the euro area banking system. The contagion analysis presented in this article shows that, in the event of a default by several of the most vulnerable counterparties, euro area banks may be faced with considerable losses, even before accounting for the impact of such defaults on asset prices.

Granular exposure data, meaning data at the counterparty level, is essential for monitoring interconnectedness and contagion arising from counterparty risk exposures. These data should include (i) exposure characteristics that allow the regulators and supervisors to assess the sensitivity of the exposures to adverse market conditions and (ii) the leverage characteristics of the NBFI entities that influence liquidity risk-related contagion channels. In addition, supervisors and regulators should strive to cover exposures between key counterparties of directly supervised institutions. This is a challenge that can be addressed by means of close cooperation between supervisors in different jurisdictions.

References

European Banking Authority (2023), “2023 EU-wide stress test – methodological note”, 31 January.

European Central Bank (2020), “Interconnectedness of derivatives markets and money market funds through insurance corporations and pension funds”, Financial Stability Review, November.

European Central Bank (2022a), “Euro area interest rate swaps market and risk-sharing across sectors”, Financial Stability Review, November.

European Central Bank (2022b), “Financial stability risks from energy derivatives markets”, Financial Stability Review, November.

European Central Bank (2023), “Key linkages between banks and the non-bank financial sector”, Financial Stability Review, May.

Haldane, A. (2012), “On counterparty risk”, Lead Comment in Journal of Risk Management in Financial Institutions, Vol. 5, No 3, pp. 224-226.

Markose, S. (2012), “Systemic Risk from Global Financial Derivatives: A Network Analysis of Contagion and Its Mitigation with Super-Spreader Tax”, IMF Working Paper, No 282.

McCaul, E, (2024), “Beyond the spotlight: using peripheral vision for better supervision”, speech at the S&P European Financial Institutions Conference, Frankfurt.

Mosk B., O’Donnell, C., Telesca, E. and Weistroffer C. (2023), “Non-banks’ liquidity preparedness and leverage: insights and policy implications from recent stress events”, Financial Stability Review – Special Feature, ECB, May

Paddrik M., Rajan S. and Young, P. (2020), “Contagion in Derivatives Markets”, Management Science, Vol. 66, No 8, pp. 3295-3798

Pinter, G. (2023), “An anatomy of the 2022 gilt market crisis”, Bank of England Staff Working Paper No. 1019.