SUNANDO ROY – On Banking, Finance and Society

Managing interest rate risk in the banking book, or IRRBB, is a vital task for banks, especially when dealing with assets and liabilities in multiple currencies. A key challenge is determining how interest rate changes in one currency, like the US dollar, relate to those in another, such as the euro. These relationships, known as correlations, play a major role in calculating overall risk and ensuring banks hold enough capital. Recent banking crises, such as the 2023 Silicon Valley Bank collapse, have shown what happens when these risks are underestimated, making accurate correlation assumptions more critical than ever. Regulatory bodies, including the Basel Committee on Banking Supervision and the European Banking Authority, stress that banks must use clear, well-documented, and thoroughly tested methods to model these correlations. This blog explores how banks approach correlation modeling for IRRBB across currencies, highlights recent research, dives deeper into regulatory expectations, and provides a practical example to illustrate the concept.

When banks operate in multiple currencies, they need to understand how interest rate movements in different economies affect each other. If rates in two currencies tend to move together, the bank’s risk could increase significantly. However, if they move independently or in opposite directions, this might reduce overall risk by balancing things out. Getting these assumptions wrong can lead to misjudging the bank’s financial health, which could mean holding too little capital to cover losses. The 2023 banking turmoil, where rapid rate hikes exposed weaknesses in some banks’ risk management, underscored the need for robust correlation modeling. To address this, banks typically use three main approaches: looking at historical data, testing extreme scenarios, and using advanced statistical tools. Each method has its strengths, but regulators demand that banks combine them thoughtfully and back them up with solid evidence.

One common approach is to study historical data, often spanning five to ten years, to see how interest rates in different currencies have moved together in the past. For example, US dollar and euro rates often move similarly because of close economic ties between the US and Europe, while US dollar and Japanese yen rates might show less alignment due to different economic policies. However, recent research from 2024 warns that past patterns don’t always hold, especially during turbulent times like the 2020 pandemic, when global markets behaved unpredictably. To improve accuracy, some banks now use machine learning to analyze historical data alongside economic indicators, helping them better anticipate shifts in these relationships.

Another method involves preparing for worst-case scenarios by assuming that interest rates in all currencies move together perfectly. This stress testing helps banks ensure they can handle extreme market conditions, like those seen during the 2022–2023 global rate hikes when central banks acted in sync. A 2023 analysis of European banks found that assuming strong correlations in these tests revealed hidden risks, particularly for banks with complex deposit structures. Regulatory bodies, including the Basel Committee, have pushed for more rigorous stress tests since the recent banking crises, urging banks to consider how synchronized rate movements could impact their finances.

For a more sophisticated approach, banks turn to advanced tools like statistical models or even artificial intelligence. These methods can capture complex, changing relationships between interest rates, especially during volatile markets. For instance, they might reveal that US dollar and British pound rates tend to align more during global economic downturns. A 2023 study showed that newer technologies, like deep learning, can predict these shifts more accurately than older methods, helping banks reduce errors in their risk calculations. However, these tools are complex and require careful testing to ensure they’re reliable.

Regulatory guidance on correlations in IRRBB has become more detailed in recent years, reflecting the growing complexity of global banking. The Basel Committee’s 2016 IRRBB standards set the foundation, requiring banks to use transparent correlation assumptions and validate them with historical data and stress tests. In 2023, the Basel Committee issued updated guidance, emphasizing the need for banks to account for recent market volatility and recalibrate their models accordingly. This includes testing scenarios where correlations spike during crises, ensuring banks are prepared for sudden shifts. The European Banking Authority has also tightened its rules, particularly with its 2022 guidelines on IRRBB and credit spread risk, which is the risk that changes in bond credit ratings affect a bank’s portfolio. These guidelines require banks to model how interest rates and credit spreads interact, adding another layer of complexity to correlation assumptions. The EBA expects banks to report these risks separately and use realistic correlation estimates, even when data is limited. Additionally, the Central Bank of the UAE introduced model management standards in 2022, urging banks to establish clear governance processes for correlation modeling, including board-level oversight and regular audits. These regulations collectively demand that banks not only use robust methods but also explain their choices clearly to regulators, with detailed documentation and evidence of testing.

Beyond interest rates, banks must now consider credit spread risk in the banking book, known as CSRBB, which has gained attention since the EBA’s 2022 guidelines. This involves understanding how changes in credit spreads—essentially the extra yield investors demand for riskier bonds—relate to interest rate movements. These relationships are harder to predict and can vary widely, making accurate modeling a challenge. A 2023 study of European banks highlighted that many struggled to integrate CSRBB into their IRRBB frameworks, often due to inconsistent correlation assumptions. Regulators are pushing banks to address this by using more granular data and advanced tools, though smaller banks may find these requirements resource-intensive.

Despite these advancements, challenges remain. Historical data might not be available for emerging markets, making it hard to estimate correlations accurately. Advanced models, while powerful, can be sensitive to errors if not properly tested. Regulatory demands for detailed reporting and validation also add to the workload, with the EBA and Basel Committee expecting full compliance by mid-2024. To overcome these hurdles, banks should use a mix of historical analysis, stress testing, and advanced modeling, while regularly checking their models against real-world data. Tools like standardized checklists from consulting firms can help banks align with best practices. Strong governance, including oversight from senior management, is also essential to meet regulatory expectations and ensure models remain reliable.

In conclusion, correlations are at the heart of managing IRRBB across multiple currencies, shaping how banks assess and prepare for interest rate risks. By combining historical insights, stress tests, and cutting-edge tools, banks can build robust models that reflect real-world complexities. Regulatory guidance from the Basel Committee, EBA, and others provides a clear roadmap, emphasizing transparency, rigorous testing, and adaptability to recent market lessons. As global markets continue to evolve, ongoing research and technological advancements, like machine learning, will further strengthen IRRBB management. The example below illustrates how different correlation assumptions can affect a bank’s risk calculations, offering a practical perspective on these concepts.

References

1. Basel Committee on Banking Supervision. (2016). Standards: Interest Rate Risk in the Banking Book. Bank for International Settlements.
2. Basel Committee on Banking Supervision. (2023). Consultative Document: Recalibration of Shocks for IRRBB. Bank for International Settlements.
3. European Banking Authority. (2022). Guidelines on the Management of Interest Rate Risk and Credit Spread Risk. EBA/GL/2022/14.
4. KPMG. (2023). IRRBB and CSRBB: Benchmarking Analysis of 47 European Banks. KPMG Publications.
5. Bank for International Settlements. (2023). Annual Economic Report 2023. BIS Publications.
6. International Monetary Fund. (2024). Global Financial Stability Report. IMF Publications.

Annexure: Example of How Correlation Assumptions Affect Risk Calculations

Imagine a bank with significant operations in US dollars and euros, holding a portfolio of fixed-rate loans in both currencies. The bank wants to measure how a sharp increase in interest rates might affect the value of its assets, a metric known as economic value of equity, or EVE. Regulators require the bank to test what happens if rates rise significantly, using a standard scenario from the Basel Committee. The bank tests two cases: one where it assumes interest rates in both currencies move together perfectly, and another where it uses a more realistic estimate based on past data, where rates are somewhat aligned but not identical.

In the first case, the bank assumes that a rate increase hits both the US dollar and euro portfolios at the same time and with full force. This leads to a noticeable drop in the value of the bank’s assets, as both sets of loans lose value simultaneously. The bank calculates that this scenario reduces its EVE by a significant amount, equivalent to about 5.5% of its core capital, which is a key measure of financial strength.

In the second case, the bank uses historical data showing that US dollar and euro rates don’t always move in perfect sync—sometimes one rises more slowly or not at all. This assumption suggests that the impact on the bank’s assets is less severe, as losses in one currency might be partly offset by stability in the other. The result is a smaller EVE decline, around 5% of core capital, indicating lower risk.

Both scenarios show the bank’s risk is within regulatory limits, which cap acceptable losses at 15% of core capital. However, the difference between the two cases highlights how correlation assumptions can affect risk estimates and influence decisions about capital reserves or hedging strategies. This example underscores why regulators insist on testing multiple scenarios and validating assumptions with real-world data.