Simplification and complexity in the banking union: separating the wheat from the chaff

Keynote speech by Pedro Machado, Member of the Supervisory Board of the ECB, at the 12th Conference on the Banking Union: “Simplification of Banking Regulation and Supervision”

Frankfurt am Main, 3 November 2026

Thank you for the invitation.^[1] A few weeks ago I argued, in another forum, that competitiveness and simplification are part of a shared agenda for European banks, as opposed to two agendas running in parallel.^[2] Today I want to go one layer deeper and ask some harder questions. What exactly is the complexity we are trying to remove? Who should bear its cost? And how do we remove it without weakening the foundations that have made European banks resilient?

I ask this because “simplification” has become one of those concepts everyone agrees with until you ask them what it means. For some, it means shorter documents. For others, lower capital requirements. And for others still, it means fewer reports, fewer layers of supervision, or fewer differences between the rules in Europe and the rules elsewhere. Clearly, these are all very different takes on simplification, and some of them are in direct conflict with one another.

So let me present my main argument from the outset, in the spirit of simplicity. Our prudential framework is complex, and while there are understandable reasons for why it has developed the way it has, it may be too burdensome across the board. The solution is to make the framework more proportionate and easier to understand, not to lower capital standards. There is scant evidence that doing so would be helpful, least of all at a time of rising uncertainty. In any case, the biggest hurdle for the competitiveness of European banks lies elsewhere entirely: in a banking market that is still fragmented along national lines at a time when technology is making scale more valuable than ever.

I will take these issues in turn. First, how complex the framework has become and why. Second, what complexity means for different stakeholders. Third, whether capital requirements are the problem. Fourth, where the real gains lie. And finally, a word of caution about the cycle we are in.

How complex has prudential regulation become, and why?

Let me start with the numbers, because they are striking. Basel I, agreed in 1988, ran to roughly 30 pages. Basel II, in 2006, to around 350. Basel III, as finalised in 2017, to around 2,000. And that is only the core standard. Once we add the implementing technical standards, guidelines, Q&As and supervisory guides that turn those standards into operational rules, the body of text that a European bank must navigate runs to well over 15,000 pages.

It is not only the volume that has grown. Banca d’Italia research using text analysis of new euro area financial legislation since 1970 finds that both the length of financial laws and their linguistic complexity, as measured by readability indices, have risen, with a marked step-up after the global financial crisis.^[3] Put simply, laws have become longer and harder to read.

We should be clear about why this is the case. Much of this complexity was a response to a financial system that had, itself, become more complex. Think of Ellis, Giannetti and Jotikasthira, who make exactly this point in a paper presented at this year’s ECB Forum on Central Banking: a complex financial system needs a correspondingly rich regulatory framework, and the stringency of post-crisis regulation has brought about genuine resilience.^[4] The question is not, therefore, whether the rules should be complex, but whether the complexity we have is justifiable and connected to the most critical and relevant risks.

Here, two problems emerge, and they are mirror images of each other.

The first is that complexity is regressive. A large part of the cost of compliance is fixed: the systems, staff and legal review. Spread over a trillion-euro balance sheet, it is a rounding error. But for a small regional lender, it is a real constraint on their ability to compete. Research on the fragmented US regulatory system documents this, and comparative legal work in Europe makes a related point.^[5] Complexity, in other words, favours scale. That is worth remembering when we hear that regulation is holding back the largest banks.

The second problem cuts the other way: complexity creates room for arbitrage, and arbitrage is an exclusive game for those who can afford to play it. The clearest example is internal models for risk weights. Research shows that model-based capital requirements, introduced to make capital more risk-sensitive, have had several unintended consequences. First, they have allowed banks to report lower risk weights where the underlying risk is not actually lower.^[6] Second, banks’ own incentives shape the risk models they build.^[7] And third, risk weights have turned out to be poor at tracking actual risk across large international banks.^[8] The findings from the literature are uncomfortable but important: the parts of the framework that are most complex are often those that give the largest banks the most discretion.

So complexity does not come at a neutral cost that is equal for all banks. Any serious simplification agenda has to address this issue.

Complexity means different things to different people

Turning to my second issue – if you want to know what people mean by simplification, look at what they ask for when given the opportunity. Earlier this year the European Commission ran a targeted consultation on the competitiveness of the EU banking sector, to which the Eurosystem contributed.^[9] The ECB’s Directorate General Research has classified the open-ended responses into five themes, using several large language models as independent coders. These themes are: simplification of buffers and rules, proportionality, minimum capital requirements, external competitiveness and reporting burden.^[10] The patterns that emerge are revealing.

For the largest banks, simplification is mostly about two things. Close to one-third of the issues they raised concerned external competitiveness, notably a level playing field with their US peers. And almost one-fifth concerned minimum capital requirements. The reporting burden and proportionality also featured, while buffer simplification was barely mentioned. In short, for large banks “simpler” tends to mean “less demanding relative to the United States”.

For small banks, the picture is almost the opposite. Nearly half of the issues they raised concerned proportionality, and more than one-third concerned the reporting burden. Capital requirements and external competitiveness were almost absent, which makes sense, as less significant institutions tend to operate with higher capital levels than significant institutions and are not competing with US investment banks.^[11] For small banks, “simpler” means “rules scaled to my size and my risk”. I will come back to this point later.

For supervisors and central banks, the emphasis was different again. One-third of the issues they raised concerned the simplification of the buffer framework and the rulebook itself, with proportionality and the reporting burden close behind. For them, “simpler” means a framework that is easier to understand, communicate and use in a crisis.

One further finding from the same consultation deserves attention. Respondents were asked whether they would support less complexity in the regulatory framework even if it meant somewhat higher capital requirements. Supervisors and central banks, on average, said yes. Banking associations and large banks responded with a clear no. That tells us something important: the industry’s concern is less about complexity as such and more about the level of the requirements.

I say this not to be judgemental, but as a clarification. These are legitimate positions. But if we are going to simplify banking regulation and supervision effectively, we need to be clear about which of these three conversations we are having at any given moment. The small bank asking for proportionate reporting, the large bank requesting lower capital requirements and the supervisor seeking a cleaner buffer stack are touching on three different things.

Are capital requirements the problem?

So let me address that one conversation directly. Two arguments are regularly made in support of lowering capital requirements: that it would support lending, and that it would boost competitiveness. The evidence for both these claims is worth examining carefully.

On lending, the honest answer is that it depends on the economic environment. In a downturn, releasing capital does help. Using euro area credit register data, research finds that a 1 percentage point release of capital requirements during the pandemic boosted lending by around 3%.^[12] That is why releasable buffers matter – and why the macroprudential framework should be simplified accordingly, as I will mention below. But in normal times the effect is very different. The literature shows that the impact of changes in capital requirements on lending is state-dependent and, outside of crises, small.^[13] A similar conclusion is reached for small firms across recessions and recoveries.^[14] Lower requirements in good times buy very little additional credit.

Higher capital, on the other hand, has substantial benefits over the medium term. Banking crises are extraordinarily costly. Across 210 crisis episodes, research estimates an average peak-to-trough decline in real GDP of around 5 to 6%.^[15] Laeven and Valencia examine cumulative output losses relative to trend in high-income economies and find that these losses amount to one-third of a year’s GDP over the crisis period and the three years that follow.^[16] Capital is the loss-absorbing capacity that makes such episodes less likely and less severe. And well-capitalised banks lend more over time, not less. Two influential studies show that stronger capital positions are associated with more lending in the medium term, partly because funding costs fall.^[17] This is the evidence that shows there is no trade-off between resilience and competitiveness.

Another frequently made claim is that capital levels are historically high. But by the only measure that is comparable across a century – the ratio of capital to total assets – they are not. For Germany, France, Italy and the United States, equity ratios of 15-20% were common before the First World War. They fell to low single digits after 1945 and have only partially recovered since the global financial crisis, to between 6% and 9% today.^[18] Recent work on the point at which requirements begin to hurt growth puts it at roughly 11-12% of assets or higher, well above current levels.^[19]

There is yet another claim we often hear: that European banks are at a disadvantage because US requirements are lower. ECB staff have calculated what the minimum Common Equity Tier 1 – or CET1 – capital would be for euro area significant institutions if US rules were applied to them as at the end of 2025.^[20] For the largest banks, the required capital would be around 8% higher under US rules, with a plausible range from slightly above zero to nearly 17%. And for smaller significant institutions it would be somewhat lower, around 5%. So the widely held belief that US rules are uniformly lighter does not seem to hold up under a counterfactual analysis, at least for the large banks that make this comparison most often.

None of this means capital requirements are perfect. Where US and European banks compete head to head – in leveraged lending, securitisation, prime brokerage and market-making – the calibration of specific requirements can matter a great deal for market share. But we should also not lose sight of the fact that quality of capital matters, as the last crisis showed.^[21] Considerable progress has been made since then: in 2008 a large share of what counted as bank capital was subordinated debt, whereas today it consists mainly of CET1, the capital that actually absorbs losses while a bank is still a going concern.^[22] Ownership structures matter for banks’ willingness to raise equity.^[23] And a functioning securitisation market matters for increasing lending room to small and medium-sized enterprises.^[24] These are targeted issues. They are not an argument for lowering the overall level of resilience in the system.

Let me also briefly mention the output floor, which ranks high among the concerns of the largest banks. The floor is the backstop that limits how far internal models can push capital below what the standardised approaches would require, and that is precisely the part of the framework where, as I noted earlier, there is the strongest evidence of arbitrage.^[25] And that is why regulatory backstops, including the output floor, should be maintained to guard against risks being underestimated.

Where the real gains can be found: scale, integration and simpler supervision

If lower capital requirements are not the answer, what then is? Let me offer three answers, in increasing order of ambition.

First, simplify what can be simplified now. This is where the ECB has moved from design to delivery: faster capital decisions, with most own funds reductions now eligible for fast tracking measured in days rather than months; stress test data points cut by about half; a streamlined short-term exercise; shorter and more focused on-site inspections; and a review of more than 100 supervisory guides, many of which are being retired.^[26] None of this lowers the bar. But all of it reduces the fixed cost of being supervised, which, as we have seen, is the cost that hits smaller banks hardest.

Nowhere does this matter more than in small banks, and here euro area supervisors have just put forward a concrete set of proposals.^[27] Rather than creating a separate prudential regime for smaller banks, the approach builds on the existing category of small and non-complex institutions. This category already covers more than 1,400 banks – three-quarters of the less significant institutions in the banking union – and these already report only around 30% of the data required from larger institutions. The proposals would broaden the category in two ways: first by allowing national supervisory authorities to raise the total assets threshold from €5 billion to as much as €10 billion, to reflect the size and structure of their domestic banking sector; and second by treating a liquidation strategy in the resolution plan as the test of non-complexity, which would help banks, particularly in smaller Member States, that today fall outside the category for technical reasons. Up to 85% of less significant institutions could then qualify – around 150 more banks than today. And the category would be used systematically: new legislation should state how it would apply to these banks, ideally in a single, consolidated rulebook specifically for them.

Much can also be done without changing the law. Where their risk profile allows, some small banks could go two to three years without a full Supervisory Review and Evaluation Process, while supervisors would still have discretion to look more closely whenever risks warrant it. Around 1,000 small and non-complex institutions currently run bottom-up stress tests; for most of them, supervisors’ own top-down projections would serve the same purpose. From 2027, a dedicated category in the ECB’s FINREP reporting would reduce the data points these banks have to submit from around 13,500 to roughly 700. Governance requirements can be scaled too, for instance by allowing small banks to combine their nomination and remuneration committees, or their risk management and compliance functions.

Let me be clear about what this is not. It is not about lowering the guardrails for small banks. They face the same geopolitical, cyber, digital and climate-related risks as everyone else, and their depositors should be just as confident that their savings are safe. That is why proportionality should be pursued within the Single Rulebook rather than outside it, and why a lighter regime must go hand in hand with a credible, flexible and efficient crisis management framework for small banks. The aim is not to lower standards but to satisfy them in a more efficient and proportionate manner.^[28]

The same logic applies to the rulebook. The risk-based capital stack in Europe can comprise up to nine separate layers. Merging the macroprudential buffers into two tiers – releasable and non-releasable – would keep the same resilience while making the framework far easier to explain to a board or an investor.^[29] Here, I would add one caveat from our own analysis: relabelling buffers does not change the underlying risks they address, and a simpler stack must not become a false sense of security about systemic risks that the “old” layers were designed to capture. Complexity in the capital stack also has a source that is easy to overlook: requirements set by different authorities can add up in ways that no single authority intended, with cumulative and, at times, even contradictory effects. Research on regulatory overlap finds that it measurably lowers productivity.^[30]

Second, let banks reach scale across borders. The comparison with the United States is instructive, and not in the way it is usually made. The United States removed geographic barriers to banking with the Riegle-Neal Act in 1994.^[31] Europe removed them, on paper, with the Second Banking Directive in 1989.^[32] Yet, today, at least 45% of credit to US firms crosses state lines^[33], while only about 14% of credit to euro area firms comes from a bank whose ultimate parent is in another euro area country.^[34] The research on the US outcome is clear: branching deregulation raised growth and bank efficiency^[35], and geographic expansion reduced bank risk.^[36] Other studies make the broader case that institutional integration has been Europe’s most powerful source of competitive markets.^[37]

Cross-border bank M&A in the euro area – as measured by the assets of target banks – has been close to zero since the financial crisis, while the market value of the largest US banks has outstripped by far that of their European peers.^[38] The barriers seem clear: an unfinished banking union and national legal requirements – with the latter relating to areas as diverse as insolvency law, deposit insurance and withholding tax, to name but a few.^[39] And these barriers are of our own making. Let’s be clear about this: integration inside an incomplete banking union needs to be approached with care for a bigger bank that cannot be resolved is a bigger problem, not a smaller one.^[40]

Third, recognise that technology has changed the economics of banking. This is the point I want to leave you with, because it reframes the whole debate. Banking has become a high fixed-cost industry. In the United States, fixed costs have risen from around 40% of non-interest expenses for large banks in the mid-1990s to around three-quarters today, and from under 20% to over 30% for small banks.^[41] The drivers are digitalisation, cloud technology, cybersecurity and now artificial intelligence. Comprehensive research analysis explains why: a bank has to be at the technological frontier in every function. And the frontier is expensive^[42], particularly considering that, in intangible-intensive industries, a higher share of fixed costs places greater importance on economies of scale.^[43] What’s more, early evidence suggests that AI adopters are more profitable, with the largest banks capturing a substantial share of the gain.^[44]

If that is truly the case, then the choice facing European banking is stark. Despite the importance of diversity in the banking sector, scale may play a determining role. Both cross-border and domestic consolidation may become, in that sense, a matter of attaining the scale needed for coping with the investments required by AI-driven technology. But it also raises the stakes in terms of banks being too big to fail, and that is precisely why the argument that scale should be achieved through weaker regulation, supervision or resolution should not be accepted.

A word of caution: the regulatory pendulum

Before I conclude, let me say something about timing, because the moment matters. We are debating simplification at a point in the cycle when European banks are profitable, asset quality is strong and the last crisis is a fading memory for many in the market. History tells us this is exactly when the pressure to loosen is strongest and the ability to see risk is weakest.

The pattern is well documented. Literature traces regulatory cycles across a century of financial history: as the cycle improves and credit losses decline, memories fade, lobbying intensifies and the political system responds by diluting rules, loosening supervisory intensity and, in some cases, drawing the industry itself into the supervisory process.^[45] Studies of the political economy of US banking deregulation show how it followed the fortunes of the industry.^[46] Research finds the same in nineteenth century usury laws.^[47] Other work documents how political constraints shape the response once the crisis arrives.^[48] And behavioural finance gives us the mechanism: expectations overreact to good news, so risk looks lowest when it is building fastest.^[49]

The backdrop today does not invite complacency. European banks are in a far stronger position than in 2012, when the interbank market had dried up and the central bank had, in effect, to take its place. But uncertainty has been building for more than two years: the interest rate environment, geopolitical risk and its effect on energy prices, imbalances in private asset markets left behind by years of near-zero borrowing costs, and the still-unknown ways in which artificial intelligence may reshape credit, markets and entire business models. Yes, the system is robust today. But as my colleague Patrick Montagner recently put it, “the point at which all these uncertainties converge could be the catalyst for an economic shock”. ^[50] That is why any relaxation of capital requirements should be approached with the utmost caution: it would signal that banks need less capital at precisely the moment when uncertainty is rising.

I raise this issue not to argue against simplification but to distinguish between simplification that removes dead weight and simplification that removes load-bearing walls. The former is overdue; the latter is how the next crisis gets built.

Conclusions

Let me sum up in three points.

First, for small, low-risk banks, the complexity of the framework and the weight of reporting are a real burden on competitiveness. We should reconsider these, and proportionality should be built into the rules rather than bolted on afterwards. This we could do by widening the category of small and non-complex institutions, using it systematically in legislation and supervising these banks in a way that matches their risk, all within the Single Rulebook and backed by a credible crisis management framework.

Second, for large, systemic banks, the evidence for lowering capital standards is weak and the comparison with the United States does not show what it is often claimed to show. We should uphold rigorous capital standards, preserve the role of the output floor as a backstop to internal models and continue to address the costs of resolution, so that scale does not come at a cost to the taxpayer. The resilience we have built is our best asset in an uncertain world; this is not the moment to throw it away.

Third, economies of scale in banking are more important than ever, and digitalisation, cybersecurity and artificial intelligence will only add to their importance. The way for European banks to achieve that scale is through cross-border integration or domestic consolidation inside a completed banking union, not through a race to looser rules.

I referred some weeks ago, in the same forum I mentioned at the beginning, to Padoa-Schioppa’s image of Economic and Monetary Union as a jigsaw puzzle and to his warning not to discard any piece before the overall design is complete.^[51] Simplification, done well, is how we see the picture more clearly. Done badly, it is how pieces go missing.