Heterogeneity in buffers set for systemically important banks in the European banking union

Prepared by Matteo Greco, Maciej Grodzicki and Ursula Vogel^[1]

Published as part of the Macroprudential Bulletin 30, August 2025.

Unwarranted heterogeneity in O-SII buffer levels in the European banking union may have adverse consequences for financial stability and the level playing field in the banking market. Analysis of national buffer-setting yields evidence of heterogeneity which does not result from differences in the size, concentration and funding structure of the domestic banking systems. From a banking union perspective, buffer-setting by national authorities results in heterogeneity at both the upper and the lower end of the distribution of a bank’s systemic relevance. The recent enhancement of the ECB’s O-SII floor methodology is designed to mitigate unwarranted O-SII buffer heterogeneity at the lower end of the buffer range. As the ECB can impose higher macroprudential requirements but not reduce macroprudential requirements, it is not possible for the ECB methodology to address instances of unwarranted heterogeneity at the upper end.

1 Framework for setting capital buffers for systemically important banks

Systemically important banks are required to hold more capital than other banks as their failure can have serious consequences for the economy and the financial system as a whole. The global financial crisis demonstrated how failures of large, interconnected banks which play a key role in financing the economy can have systemic repercussions. In 2011 the Basel Committee on Banking Supervision published a framework governing how systemically important banks should be identified at the global level and prescribing the additional capital surcharges for such banks (see BCBS, 2011). The Basel Committee also recognised that banks may be systemically important in a given country. However, the framework it designed for domestic systemically important banks (D-SIBs) is principle-based, making it less prescriptive than the equivalent for global systemically important banks (G-SIBs) (see BCBS, 2012). It complements the G-SIB framework, focusing on the systemic importance of a given bank for its domestic economy. As such, it leaves a larger margin of discretion to the national authorities responsible for its implementation. Principle 8 of the D-SIB framework provides some guidance to authorities on how country-specific factors could influence their buffer-setting decisions. The size of the banking sector relative to the economy, its concentration and its reliance on volatile, foreign funding could encourage authorities to set the buffers at a higher level.

The EU implemented the Basel framework by introducing two buffer requirements for systemically important institutions. Since 2013 macroprudential authorities in the EU have been empowered by the Capital Requirements Directive (CRD)^[2] to identify banks in their jurisdiction as either global or other systemically important institutions (G-SIIs/O-SIIs). The framework for identifying G-SIIs closely follows the Basel G-SIB framework. The framework for identifying O-SIIs builds on the Basel D-SIB principles, with some additional clarifications. National authorities are expected to identify O-SIIs based on at least four criteria: size, importance for the economy of the Union or of the relevant Member State, significance of cross-border activities and interconnectedness of the institution or group with the financial system.^[3] Authorities can require O-SIIs to hold additional capital equal to up to 3% of the total risk exposure amount.^[4] The higher of the G-SII and O-SII buffer requirements applies.

The decentralised setting of O-SII buffers leads to unwarranted heterogeneity in buffer levels. Whereas the identification of O-SIIs is governed by guidelines which provide a scoring methodology for national authorities to use (see EBA, 2014), no further guidance has been provided by legislators or the European Banking Authority (EBA) on how to translate O-SII scores into buffers. The choice of a specific methodology for setting O-SII buffers is left fully to the discretion of the national authorities. The EBA reviewed the methodologies used to calibrate O-SII buffers in 2020, documenting a wide range of methods applied by the national authorities for this purpose (see EBA, 2020). These methods often rely to a large degree on policymakers’ judgement. Even in cases where the national authorities apply a model-based approach to buffer calibration, the results require a judgement-based correction. For example, the frequently used equal expected impact approach often produces raw buffer levels which fall well outside of the range deemed adequate by policymakers, and the assumptions underlying the approach may provide grounds to deviate from the results (see Passmore and von Hafften, 2017, and Geiger et al., 2022). Relying so heavily on judgement in the buffer framework raises concerns that differences between O-SII buffer levels may not always be driven solely by economic factors.

Unwarranted heterogeneity in O-SII buffer levels across the banking union may have negative consequences for financial stability, including beyond national banking systems. If the O-SII buffers for some banks insufficiently capture the serious impact of an O-SII failing, the economy and the financial systems may not be well prepared for the materialisation of financial stability risks. Given the substantial degree of financial and economic integration in the banking union, shocks are unlikely to remain limited to the domestic banking system but spread across the entire banking union. In addition, systemically important banks may have no incentive to keep their systemic footprint in check.

Buffer heterogeneity may also distort the level playing field in the banking union. As the EU forms a single market, O-SIIs compete not only domestically but also with banks from the rest of the EU. Misaligned O-SII buffers may give competitive advantages to specific banks, and differences in buffer levels may not be warranted by countries’ fundamentals. Inconsistent buffer-setting may also influence the allocation of assets by banks. Banks located in a country that sets its buffers at relatively low levels could be able to provide banking services throughout the EU at a lower cost than their peers based in other countries, and vice versa. Similarly, the direction of cross-border banking consolidation may be affected by O-SII buffer-setting (see ECB, 2021), meaning that banks facing a relatively high buffer would be disadvantaged if they intended to acquire a bank located in a country which sets its buffers at a relatively low level.^[5]

The remaining sections of this article review the heterogeneity that has been observed in O-SII buffers. Taking a national perspective, Section 2 discusses the dispersion of O-SII buffers in the 21 countries participating in the European banking supervision and analyses whether the heterogeneity observed can be attributed to country-specific factors recommended by the Basel Committee. Section 3 considers the perspective of the banking union as a whole, and looks at how the heterogeneity observed in buffers aligns with the systemic importance of O-SIIs for these 21 countries as an aggregate. Section 4 concludes.

2 Buffer heterogeneity from the national perspective

Descriptive analysis shows that the relationship between O-SII buffer rates and the size and concentration of the national banking systems is weak. There is no clear correlation between bank assets-to-GDP and weighted average O-SII buffer rates in the 21 countries participating in European banking supervision (Chart 1, panel a). A group of countries seem to set O-SII buffer rates at particularly high levels, despite the small aggregate size of their banking systems. At the same time, some countries seem to impose buffer rates close to the lower end of the range of their European peers. While the more concentrated banking systems seem to face higher O-SII buffers on average, there is considerable dispersion among countries, and two banking systems with a similar level of concentration may be subject to very different buffer levels (Chart 1, panel b).

Chart 1

O-SII buffer rates show little relationship with size and concentration of domestic banking systems

a) O-SII buffer rates and size of the banking sector	b) O-SII buffer rates and banking system concentration
(2023; x-axis: bank assets-to-GDP, percentages; y-axis: weighted average O-SII buffer rate, percentages)	(2023; x-axis: share of top three banks in total assets, percentages; y-axis: weighted average O-SII buffer rate, percentages)

A regression model is set up to investigate the relationship between the country-specific factors suggested by the Basel Committee and O-SII buffer rates more formally. In this set-up, the level of O-SII buffer is explained using the logarithm of O-SII scores (in order to account for the mostly concave shape of national O-SII calibrations) together with three country-specific variables. Total bank assets, expressed as a share of GDP, measure the relative size of a country’s banking sector. The Herfindahl-Hirschman Index (HHI) measures the concentration of the banking system and also the share of interbank funding in total liabilities serves as a proxy for (international) wholesale funding. Regressions are run separately for each year from 2018 to 2024 in order to explore changes in the relationships between country characteristics, scores and buffers over time.

Higher O-SII buffers are associated with higher O-SII scores, with the strength of this relationship increasing over time. In line with the objectives of the O-SII buffer, national authorities set higher buffers for those banks which are assessed to be more systemically important. The estimated coefficient representing the impact of scores on buffer levels has increased over time by about 25-30%. This could be due to countries refining their national calibration frameworks, partly in response to the implementation of the ECB floor methodology. In recent years, an increase of around 10% in the O-SII score is associated with an O-SII buffer that is 25 basis points higher.

O-SII buffers tend to be lower in countries with large banking systems, although the economic significance of this association is low. The relationship between the size of the banking sector and O-SII buffers is negative and, in many years, statistically significant. This conflicts with the Basel Committee principles. However, estimation results indicate that this relationship has little economic significance. Given the variation across different regression model specifications, an O-SII buffer that is 25 basis points higher could be expected for an increase of between 150 and 1,000 percentage points in the total bank assets-to-GDP ratio, while this ratio comes in a range between 100% and 300% for most banking union (BU) countries. The other country-specific factors are, in most cases, not statistically significant (see Chart 2).

Some of the heterogeneity observed could stem from preferences of national authorities which host large foreign bank subsidiaries for higher O-SII buffers. Controlling for differences in O-SII buffer levels between countries that have significant numbers of foreign bank subsidiaries and those with banks that are predominantly domestically owned does not affect the relationship between scores, assets-to-GDP ratios and buffers. It does, however, reveal a statistically and economically significant difference in buffer levels between the two groups of countries. O-SIIs in countries with significant numbers of foreign banks (Bulgaria, Estonia, Croatia, Latvia, Lithuania and Slovakia) are charged on average an O-SII buffer that is about 50 basis points higher, all else equal, than their peers in the home countries.

Chart 2

While O-SII buffers increase with higher O-SII scores, the economic significance of buffers’ relationship with country-specific factors is weak

Estimated relationship between O-SII scores, country-specific factors and O-SII buffers

(2018-26, estimated model coefficients and confidence intervals)

3 Buffer heterogeneity from the BU perspective

From the BU perspective, heterogeneity in O-SII buffers is particularly concerning if it affects buffers for the most systemically important banks in the banking union. With the harmonisation of banking regulation, supervision and resolution, some of the safety net arrangements covering the failure of systemically important banks have been transferred from the national level to the European level. The cost of any such failures would be shared by the banking systems of all EU Member States participating in the banking union and no longer by the affected domestic banking system alone. This suggests it would be prudent to review O-SII buffers not only from the national perspective but also from the perspective of the banking union as a whole (see Grodzicki et al., 2025). At the same time, buffer heterogeneity leads to a distorted playing field, which has a particularly strongly impact on large banks active in multiple jurisdictions that compete with each other directly.

From a BU perspective, buffer-setting by national authorities results in heterogeneity at both the upper and the lower ends of the distribution of a bank’s systemic relevance. O-SII scores, when calculated from the BU perspective, reveal that several European G-SIIs and large O-SIIs stand out, as their buffer rates are similar to those assigned to much smaller banks (see Grodzicki et al., 2025). A similar conclusion can be drawn from a comparison of simpler indicators, such as relative size and share of lending in the banking union (Chart 3). Several O-SIIs whose systemic importance in the banking union as a whole appears to be very low are required to maintain comparatively high buffers, while some G-SIIs and some highly systemically important O-SIIs face substantially lower requirements.

Chart 3

O-SII buffer rates for some of the most systemically important banks in the banking union are close to those assigned to the median O-SII in the banking union

a) O-SII buffer rates and O-SIIs’ share of total banking assets in the banking union	b) O-SII buffer rates and O-SIIs’ share of total loans in the banking union
(2023, percentages)	(2023, percentages)

Statistical analysis enables O-SII buffers to be benchmarked from the BU perspective. An ordered probit approach is used to estimate the relative stringency of each O-SII’s buffer rate applicable in 2024 on the scale of possible buffer rates. In this model, the O-SII buffer is treated as a cardinal dependent variable, taking one of ten possible values between 0.25 and 2.50 percentage points.^[6] It is explained using the four sub-scores of the O-SII score from the BU perspective together with country-level variables that may affect O-SII buffer-setting by the macroprudential authorities.^[7] A model-based probability distribution conditional on its O-SII score from the BU perspective is derived for each O-SII. Then the actual buffer rate is compared with this distribution, yielding an estimated percentile of the distribution to which the buffer rate corresponds. Given that this approach is based on the results of countries’ buffer-setting processes, its results cannot be used to claim that a particular bank’s buffer rate is too high or too low in absolute terms. That said, they indicate that some banks’ buffer rates may be too high or too low relative to the rest of the sample of O-SIIs.

The model reveals a markedly negative relationship between the systemic importance of an O-SII in the banking union and the relative stringency of its buffer rate. Focusing on the 24 most systemically important O-SIIs whose scores stand above the 80 basis point threshold of the second bucket BU floor in the ECB’s enhanced floor methodology, it becomes clear that most O-SIIs in the lower half of this group are required to have relatively high buffers, compared with the group of the most systemically important O-SIIs in the banking union. Buffer rates for some of these most systemically important O-SIIs, in particular for a few G-SIIs, are placed in the lowest quartile of their implied distributions (Chart 4).

Chart 4

The more significant an O-SII is in the banking union, the less stringent its O-SII buffer requirement is – but the ECB’s enhanced floor methodology mitigates this

Relative stringency of O-SII buffer requirements following implementation of the ECB’s enhanced floor methodology

(2023; x-axis: O-SII score from the BU perspective, basis points; y-axis: percentile of the model-implied buffer distribution)

The recent enhancement to the ECB’s floor methodology will reduce the heterogeneity in O-SII buffers at the lower end of the buffer range. A counterfactual analysis assumes that, all else equal, O-SII buffers would increase in line with the minimum rates implied by the enhanced floor methodology. This assumption is highly simplified and disregards the potential evolution of the systemic importance of O-SIIs through to the implementation deadline on 1 January 2028. Nonetheless, it demonstrates that the buffer rates for the most systemic O-SIIs would be brought closer to the median of the model-based distribution, and none of these O-SIIs would be placed any longer in the lowest quartile of the distribution.

As the ECB’s macroprudential powers are asymmetric, it is not possible for its methodology to address instances of unwarranted heterogeneity at the upper end. The ECB’s macroprudential tasks and tools allow it to apply higher requirements for capital buffers than applied by the national competent authorities or national designated authorities of participating Member States, but the ECB cannot lower the capital buffer requirements. The ECB methodology therefore includes a BU perspective that is implemented as a floor. This cannot address the relatively stringent buffer-setting at the upper end of the distribution, notably for several O-SIIs whose scores come in the range between 80 and 300 basis points. These O-SIIs would continue to face more stringent requirements than several of the most systemic O-SIIs in the banking union.

4 Conclusions

This article presents evidence of unwarranted heterogeneity in O-SII buffer levels for banks in countries participating in European banking supervision. Such unwarranted heterogeneity is a matter of concern in the banking union for two main reasons. First, if it implies buffers are being set too low for some of the most systemic banks, the economy and the financial system of the entire banking union may not be well prepared for the materialisation of financial stability risks. Second, heterogeneity at the upper and lower ends of the spectrum may distort competition and incentives for financial integration in the single European market.

While national authorities have substantial latitude to set O-SII buffers as they see fit, the heterogeneity observed in O-SII buffers seems to go partly against the direction set by the principles laid down by the Basel Committee. At the national level, more systemically important banks are required to hold higher buffers than their less important peers. In a cross-country context, however, neither the size nor the concentration of the national banking systems appears to be related to the O-SII buffer levels in the way envisaged by the Basel Committee. Such unwarranted heterogeneity has persisted for several years, as national O-SII buffer frameworks have been broadly stable. Taking the banking union perspective confirms the presence of heterogeneity in buffer requirements among both large and smaller O-SIIs.

The recent enhancement to the ECB’s floor methodology for assessing O-SII buffers will mitigate unwarranted heterogeneity at the lower end of the buffer range but, due to the asymmetric powers of the ECB, cannot mitigate the heterogeneity at the top end. Further alignments aimed at reducing buffer heterogeneity at both ends of the distribution, possibly also considering country characteristics, would require a more harmonised methodology for setting O-SII buffers to be developed and implemented in the banking union.

References

BCBS (2011), “Global systemically important banks: assessment methodology and the additional loss absorbency requirement”, Bank for International Settlements, November.

BCBS (2012), “A framework for dealing with domestic systemically important banks”, Bank for International Settlements, October.

EBA (2014), “Guidelines on the criteria to determine the conditions of application of Article 131(3) of Directive 2013/36/EU (CRD) in relation to the assessment of other systemically important institutions (O-SIIs)”, EBA/GL/2014/10, December.

EBA (2020), “EBA Report on the appropriate methodology to calibrate O‐SII buffer rates”, EBA/Rep/2020/38, 22 December.

ECB (2021), “O-SII buffer calibration and cross-border bank mergers”, Financial Stability Review, November.

ECB (2022), “Annex 1: Analytical addendum underpinning the policy proposals”, March.

Geiger, S., Heires, M., Krüger, U., Ludwig, J. and Vogel, U. (2022), “Calibrating capital buffers for other systemically important institutions (O-SIIs) in Germany – Utilising the equal expected impact approach”, Technical Papers, No 02/2022, Deutsche Bundesbank.

Grodzicki, M., van der Kraaij, A., Vogel, U. and Zsámboki, B. (2025), “Enhancing the ECB’s O-SII framework”, Macroprudential Bulletin, Issue 30, August, ECB.

Passmore, W. and von Hafften, A. (2019), “Are Basel’s Capital Surcharges for Global Systemically Important Banks Too Small?”, International Journal of Central Banking, Vol. 15, No 1, pp. 107-156.