Banking supervision and the SSM: five questions on which research can help

Speech by Ignazio Angeloni, Member of the Supervisory Board of the European Central Bank,
at the Centre for Economic Policy Research's Financial Regulation Initiative Conference organised by Imperial College Business School/CEPR,
Panel 1 “Banking Supervision and Regulation”,
London, 30 September 2015

***

Summary

The SSM has been operating for 11 months and all national authorities contribute, but none is predominant. Three features of the SSM receive less attention than others: with the SSM the ECB has been given both micro- and macroprudential responsibilities; the coexistence of national and European components in the legal basis of the SSM is a source of asymmetry, which conflicts at times with a level-playing field; and the founding regulation of the SSM – the SSM Regulation – gives it extensive powers over the banks under its direct responsibility, covering all supervisory tasks.

Mr Angeloni then highlights five areas in which research could benefit supervisors. The first area concerns financial stability – how to define it and how to measure it. Good measures of financial stability remain elusive. In contrast to monetary policy, where an extensive literature on the definition and measurement of price stability exists, no quantified objective is available for supervisors. Financial stability is multidimensional and hard to enshrine in a single notion or measure.

The second area relates to how much capital should banks hold. Answers are required for how much capital is needed in “normal” conditions, how it depends on preferences and other factors, at what point capital should be considered excessive, and why. These dilemmas are routine for banking supervisors because of their responsibility in setting Pillar 2 requirements.

The third area concerns the diversification and optimal mix of capital. As hybrid instruments are inherently complex, their use may add uncertainty to the financial system. The question is whether the additional complexity is outweighed by benefits in terms of added flexibility, better incentive structure, or lower costs.

Another question relates to what role there is for other prudential requirements. A framework is needed on how liquidity and capital requirements interact. In principle, the distinction between insolvency and temporary cash shortages is clear; but in practice, it is blurred because temporary illiquidity may lead to premature asset liquidation in stressed market conditions, and hence to losses.

A final question has to do with where – specifically, in which sub-entity of a banking group – should capital and liquidity be located. For the SSM, the problem is important because of the presence of large cross-border groups. The SSM has launched a project on harmonising national options and discretions in the CRD IV and CRR, which will increase the flexibility to manage capital and liquidity for banking groups, while still respecting basic prudential requirements.

Mr Angeloni concludes that some of these issues are very complex, and that understanding is still incomplete and evolving; which is why more research is needed.

***

A return ^[1]

I am grateful to Franklin Allen for inviting me to this conference. I very much welcome the launch of the CEPR Financial Regulation Initiative, which I regard as promising and timely.

Being here today feels a little like a return. Part of my career was devoted to research, first at the Banca d’Italia, then at the ECB in its early years. I have always believed in the value of research at central banks. If combined with policy work, research helps discipline the internal debate and makes external communication more articulate and convincing. There is no evidence that more research-oriented central banks are less prone to error, but they tend to give better explanations of their policy. And the benefit is reciprocal, I think, because researchers are stimulated by questions that need concrete answers, often urgently.

After many years working on monetary policy and related areas, I have recently moved towards banking regulation and supervision. Relative to other central bankers, supervisors are less inclined to discuss research or to admit that they can benefit from it. While some reasons may be valid ^[2], I believe that more exposure to research could help supervisors at the present time. Banking supervision (I will use this term to encompass regulation) is undergoing a re-examination. Old wisdoms are being destroyed and new paradigms introduced, most of which are tentative and untested. In this process, the need for good analysis is immense. Second, supervisory practices are going through a “transparency revolution”, as monetary policy did after the 1970s. Policymakers are increasingly asked to pursue clearly defined objectives and make their actions understood and subject to scrutiny. Transparency calls for clear thinking, which is a main by-product of research.

For these reasons, I thought I could contribute today by suggesting a few questions faced by supervisors to which research can contribute. The choice is influenced by my experience, but also – I declare this at the outset – by my personal preference for issues that (1) can be explained in simple, intuitive terms, (2) have direct concrete application, and (3) are analytically and empirically tractable. I have chosen five questions, which I will present in a moment.

Before that, however, I will recall some basic elements regarding the Single Supervisory Mechanism (SSM) which may help in the subsequent discussion.

A primer on the SSM

The banking union project was launched by the Euro summit of June 2012 with the goal of severing the link between weak banks and weak government finances that was threatening the stability of the euro. The conclusions of that summit did not specify in detail what elements would be part of the union, but hinted that it should involve bringing up to euro area level ^[3] both the supervisory powers (the Single Supervisory Mechanism) and the bank safety net. The two elements complement and balance each other. The timing, however, differed. The SSM was set up by the ECB at record speed. The single bank resolution authority is being prepared and will start operating next January. A resolution fund has been agreed, but will take years to phase in. A single deposit insurance scheme does not exist, and its modality and timing are still being discussed.

The SSM has been operating for 11 months now. It is governed by a Supervisory Board of 25 members – 19 from national supervisors and six from the ECB. In size and composition it mirrors the ECB Governing Council, whose format has proved successful over the years in formulating monetary policy. The new system brings in a clear European orientation: decisions are made by a single authority, by simple non-weighted majority rule. All national authorities contribute, but none is predominant. The European dimension is also built in the internal organisation. Day-to-day supervision of individual banks is conducted by groups of ECB and national supervisors, called Joint Supervisory Teams, acting under the coordination of the ECB. A team is in place for each supervised bank. ^[4] The teams are supported by the ECB structures and submit draft decisions to the Board.

While these features are largely known, others have received less attention. I will mention three. First, with the SSM the ECB has been given both micro- and macroprudential responsibilities, in recognition of the close linkage between the two. This makes the institution potentially stronger, but also complicates the process of defining objectives and assigning instruments to them. In addition, national and European prerogatives are intricately interlinked. In view of these complexities we are not, in my view, yet benefitting fully from the combination of the two policies.

Second, the legal basis of the SSM is EU law, mainly the Capital Requirements Regulation and Directive (CRR/CRD IV) and also the national laws that transpose CRD IV. The coexistence of national and European components and some flexibility available to member countries under EU law itself are a source of asymmetry, which conflicts at times with a level-playing field. I mention this technical legal aspect because it impacts on the ECB’s approach in supervising cross-border groups, an issue I will mention again later.

A final point regards the supervisory powers. Its founding regulation ^[5] gives the SSM extensive powers over the banks under its direct responsibility, covering all supervisory tasks: granting and withdrawing licenses, authorising mergers and acquisitions, determining how much capital banks should hold and in what form above the legal minima, setting other prudential requirements, all the way to asking for recovery plans and enacting early intervention for ailing banks. A key step in the supervisory cycle is the annual Supervisory Review and Evaluation Process (SREP), under which banks, taken at group level, are scored according to their risk and the Pillar 2 requirements are set.

Question 1: how to define and measure financial stability?

Let me now turn to my first question. Not surprisingly given my premises, it has to do with the goal of supervision. What is supervision exactly about, and what is it for?

Economists see policies as a means to correct market failures, generated for example by externalities. Accordingly, some of them argue that microprudential supervision should be concerned with correcting “safety net distortions” (that is, the tendency for bank managers to take on too much risk because they are protected by taxpayer funds), whereas macroprudential supervision should correct for “network externalities” (namely, risk transmission from systemically relevant institutions to the rest of the system, and back). ^[6] The Basel Committee on Banking Supervision (BCBS), in one of its core principles, stipulates that “the primary objective of banking supervision is to promote the safety and soundness of banks and the banking system”, and that any secondary objectives are subordinate to this one. ^[7] Now, what does “safe and sound mean”? How do the micro and the macro dimensions relate? And how can “safety and soundness” or “stability” be measured? Some metric is necessary, even if only to ensure that the responsibilities and objectives are “clearly defined and publicly disclosed” in line with another Basel core principle.

Unfortunately, good measures of financial stability remain elusive. In contrast to monetary policy, for which an extensive literature on the definition and measurement of price stability exists, no quantified objective is available for supervisors. Financial stability is multidimensional and hard to enshrine in a single notion or measure. ^[8] Broadly speaking, the above definitions imply two conditions. First, each bank should have a default probability within an “acceptable” range. The measurement of default probabilities (at least subjective ones, perceived by the market), is well developed and published measures exist. Second, default correlations across institutions must also be within “acceptable” bounds. If not, systemic events would be frequent and possibly catastrophic. Large-scale crises typically involve the failure (or a high risk thereof) of several institutions jointly. Hence, measures capturing the probability of simultaneous failures could be useful benchmarks.

To do this, two avenues have been followed. The first builds statistics capturing co-movements of asset prices. The best known are CoVar of Adrian and Brunnermeier ^[9] and the Systemic Expected Shortfall of Acharya and co-authors. ^[10] Market data are used to quantify default probabilities (under the assumption that markets operate efficiently, a strong one in a crisis). The second approach measures systemic risk from a quantitative angle. This approach draws on network science and is inspired by work of Allen and Gale ^[11]. Network analysis allows us to weigh financial institutions (hence the supervisory pressure to be placed on them) according to their systemic linkages. This approach is attractive because it does not depend on market efficiency, but is statistically demanding. ^[12]

The existence of different approaches, relying on separate data sets and assumptions, should be good news, since their combination may give rise to more robust indicators. The problem is that the links between them are unclear and the combination of the two sources of information has not been explored. This seems to me an empirical line worth pursuing. In parallel, the concept of stability, and what it implies in terms of market failure and welfare losses, should be clarified further.

The distinction between the micro and macro dimensions also needs further elaboration. After the crisis, the sharing of bank risk has shifted, in all countries, away from taxpayers and towards bank shareholders and creditors, with the imposition of stricter capital requirements and bail-in provisions. This enhances the relevance of the macroprudential focus. Micro and macro risks are covered by different capital requirements – Pillar 2 add-ons and systemic buffers, respectively. But the quantification of these layers and their interaction for the purpose of covering losses have not yet been sufficiently analysed.

Question 2: how much bank capital?

The second question is easy to ask, but very hard to answer: how much capital should banks hold? Capital occupies centre stage among prudential policies. It measures the maximum unexpected loss a bank can suffer without going out of business. All else being equal, the greater the capital, the more remote bankruptcy is. Hence, should supervisors conclude, like economists, that “more is better”?

Unfortunately, the answer is not that simple. Capital affects managerial decisions. If the incentives of managers and shareholders are aligned ^[13], it fosters good risk management. However, some have argued that stringent requirements may shift bank preferences towards more risk. ^[14] Another problem is that partial equilibrium reasoning does not consider second-round effects. Stricter requirements may discourage lending or force deleveraging, hence weakening the condition of borrowers and the broader economy. The ultimate effects are hard to determine and may even be adverse.

The crisis prompted a global drive towards higher capital standards under the aegis of the Financial Stability Board and the BCBS. The ratio of high-quality capital to risk-weighted assets for US banks, equal to 13.1% in 2009, had risen to 14.4% in 2014. For European banks, the median ratio increased from 13.1% to 17.5% over the same period. Some are suggesting that this increase may be going too far. We need to answer questions such as (1) how much capital is needed in “normal” conditions, (2) how it depends on preferences and other factors, (3) at what point capital should be considered excessive, and why. These dilemmas are routine for banking supervisors because of their responsibility in setting Pillar 2 requirements. ^[15]

Scholarly debates here are intense but have not been conclusive. Based on the Modigliani-Miller theorem with some frictions, Admati and Hellwig ^[16] calculated that a 1 percentage point increase in the risk-weighted capital requirement increases average cost of capital by only 1 basis point. This is because both equity and debt become safer (and thus cheaper) with greater equity holdings. Kashyap et al. ^[17] estimate the increase in the cost of capital to be between 2.5 and 4.5 basis points, still very low. This literature therefore suggests that setting high requirements should be appropriate. ^[18] Others have argued that the theorem does not apply to banks because of the special services they provide. The existence of demand deposits can incentivise monitoring and discipline the bank. In such a framework, higher requirements reduce the supply of valuable liquidity services. ^[19] Some case-studies point to a negative effect on lending after the increase in capital requirements. ^[20]

Another critical aspect is the link between capital and the rest of the bank’s balance sheet. Basel III introduced a leverage ratio (LR), based on total unweighted exposures, to complement the traditional risk-based framework. This ratio is not susceptible to risk weight optimisation (known as “gaming”) and model risk. However, it is unclear how the whole architecture performs over the cycle, considering also that risk weights themselves may be cyclical and tat there are also countercyclical buffers. In addition, critics of the leverage ratio suggest that it incentivises riskier lending. This is not decisive however. Research at the ECB showed that this effect is small and outweighed by the benefit of greater loss absorbing capacity. ^[21]

Question 3: which capital instruments?

My next question regards the diversification and optimal mix of capital. Capital instruments exist in various forms, ranging from ordinary shares to various forms of hybrids, subordinated debt and contingent instruments. International standards admit such instruments, within limits, to satisfy prudential requirements. ^[22]

One should recognise at the outset that as these instruments are inherently complex, their use may add uncertainty to the financial system – hardly a desirable outcome. After the crisis the prudential focus has shifted to a simpler instrument with more certain loss-absorbing capacity: Common Equity Tier 1 (CET1), which includes, in addition to shares, retained earnings and some form of reserves. EU law (CRD IV and CRR) requires most Pillar 1 capital to be composed of CET1 (4.5% out of 8% in total). ^[23] In addition, the macroprudential buffers (for capital conservation, countercyclical and systemic risk) all need to be satisfied with CET1. At the end of last year, the total capital ratio for banks supervised directly by the ECB was 15.4 %, with a CET1 ratio of 12.5 %. ^[24]

Is this additional complexity outweighed by benefits in terms of added flexibility, better incentive structure, or lower costs? For one instrument at least this has been argued strongly, namely, contingent convertibles (CoCos) ^[25], debt instruments that are automatically bailed-in in certain trigger situations. Relative to other bail-in debt, CoCos have the advantage of discouraging shareholders from taking risk. ^[26] Under Basel III and EU Law, CoCos qualify as either Additional Tier 1 (AT1) or Tier 2 (T2) capital, depending on whether or not they meet all the criteria set out in regulation. ^[27] In the comprehensive assessment conducted by the ECB in 2014, CoCos were accepted for covering losses in the stress scenario, within limit and under restrictive conditions, such as strong legal certainty and an appropriate trigger point. ^[28] At present, however, CoCos represent less than 6% of total capital of SSM banks.

All in all, I think the jury is out on whether diversification of capital will play a greater role in the foreseeable future. At present, the burden of the proof is probably more on the supporting side. This is not a definitive view, however, but an encouragement to researchers to sharpen their case further.

Related to this question is the issue of “gone concern capital”, loss-absorbing instruments that are activated not in the course of the normal business, but in resolution. ^[29] The FSB will soon issue standards for building a total loss-absorbing capacity (TLAC), for the globally systemic banks. In Europe, an analogous framework, will be applied to all banks (minimum requirement for own funds and eligible liabilities or MREL), in combination with the bail-in provisions of the Bank Recovery and Resolution Directive that will be in force next year. The coexistence of both going and gone concern capital raises important issues. How should the two requirements interact? Evidently, they cover different risks, the first offering protection primarily to more junior bank debtors, the second to more senior ones, to other banks (via the deposit insurance scheme) and ultimately to the taxpayer. That said, if gone-concern capital effectively absorbs losses in a failure, what does that imply for the optimal capital mix? Can we accept that less of the capital of banks is composed of the highest quality capital – CET1 – in the presence of a sufficiently large amount of Tier 2? Do we for example still need a buffer add-on for global systemically important banks in the form of CET1 if a credible resolution regime exists which converts in resolution (a large amount of) Tier 2 into equity? Reflection can contribute, going forward, to the finalisation and the phasing-in of the new instruments.

Question 4: what role for other prudential requirements?

Basel III has introduced new prudential requirements to complement the capital framework, the Liquidity Coverage Ratio (LCR) and the Net Stable Funding Ratio (NSFR).

The LCR requires banks to hold a minimum buffer of liquid assets to survive periods of stress in which cash outflows materially exceed inflows. The measure is based on a flow concept. This survival period has been set at 30 days to provide sufficient time to banks and the authorities to take corrective action. The NSFR, in contrast, aims to limit reliance on short-term funding which can rapidly destabilise banks. While the LCR focuses on the short-term, the NSFR is calibrated to reduce maturity mismatches for up to one year. The two ratios are thus complementary to each other.

We need framework on how liquidity and capital interact. In principle, the distinction between insolvency and temporary cash shortages is clear. In practice, it is blurred because temporary illiquidity may lead to premature asset liquidation in stressed market conditions, hence to losses. Liquid assets and equity are substitutes in that they can both reduce bank risk. However, recent work suggests a role for liquidity which is distinct from capital. ^[30] Going forward, we need modelling frameworks in which capital and liquidity requirements optimally correct market failures, calibrated to inform policy-makers about the optimal combination of the two.

From a macroprudential perspective, three areas are worth considering. First, the new liquidity requirements may negatively affect funding costs, profitability and lending, with possible adverse impact on output. ^[31] This would imply a trade-off between an improved resilience against liquidity shocks, on the one hand, and provision of credit and liquidity transformation services by banks on the other hand. The existence and magnitude of this trade-off is not yet fully understood.

Second, it has been suggested that the liquidity requirements may affect the central banks’ operational framework and the interbank market. ^[32] LCR also means that banks will have to hold more liquid securities; this may impact banks’ collateral pledging behaviour in central bank operations and the market for these securities. Further research is needed on how liquidity requirements may affect the ability of central banks to implement monetary policy.

Third, work on the macroprudential use of liquidity standards is still in its infancy. Progress in making macroprudential liquidity tools operational has been limited. An integrated framework for capital and liquidity, needed for micro-supervision, can deliver tools to improve financial stability as well.

Question 5: where should the prudential safeguards be located?

Let me conclude with the question that has received the least attention of all. The question has to do with where – specifically, in which sub-entity of a banking group – should capital and liquidity be located. Location evidently matters only if intra-group mobility of capital and liquidity is less than perfect, which usually turns out to be the case.

The issue is particularly important for cross-border groups, but it is relevant also for groups established in a single jurisdiction. Even without cross-border obstacles and under a single legal structure, there may be factors preventing the parent company from intervening in support of a subsidiary (or the reverse, though this is less frequent). For example, managers of the subsidiary may be committed to partly different shareholders. The commitment may differ also depending on the resolution strategy chosen. In the United States, for example, a single point of entry resolution strategy provides an incentive for intra-group support in going concern, unlike in the multiple point of entry case, in which subsidiaries are resolved individually without support from the group.

For the SSM, the problem is important because of the presence of large cross-border groups. 45 out of the 123 banks that the ECB supervises directly, representing two thirds of bank assets in the SSM area, are established in more than one SSM country. Different Member States have different legal frameworks, leaving room for asymmetries. For the SSM, a trade-off arises. On the one hand, intra-group flows promote banking integration and an effective single market, and may also enhance the resilience of the group. On the other, interconnectedness and dependence might facilitate contagion, which in extreme cases may involve the countries where the group operates. There is an analogy here with the literature on international capital flows, where a similar trade-off arises between integration and stability. ^[33]

European law stipulates that prudential requirements apply to individual establishment, but allows for exceptions (“options and discretions”) subject to conditions. In some cases, strict prescriptions apply; for example, derogation from capital requirements at subsidiary level are not admitted across borders, but only within countries (Article 7 CRR). Substitutability across borders is allowed, by way of derogation, for liquidity and net-stable-funding requirements, again under certain conditions (Article 8 CRR). A similar approach holds for large intra-group exposures (Article 400(2)(c) CRR). Options and discretions are exercised by banking supervisors and, in less frequent cases, by Member States via national law. Up to recently decisions were made largely by national supervisors; the SSM is now in a position to exercise options and discretions in an area-wide consistent way. To this end, it has recently undertaken a project to harmonise and rationalise the exercise of those provisions across the banking union. The goal is to increase the margins of flexibility for cross-border groups, selectively and gradually so as to keep contagion risks under control. The results of the project will be published soon and will be the object of a public consultation, open to all. Public consultations are not usual fora for academic debates, but we nonetheless encourage input from all sources.

Conclusions

My remarks have turned out to be a bit longer and more detailed than I intended. At the start I made a plea for simplicity and intuition, but now I am not sure I kept the promise. For this, I could blame a lack of sufficient time to prepare, like Blaise Pascal, who once wrote “Had I had more time, I would have written a shorter letter”. But that would not be fully true. The reality is that some of these issues are very complex, even more since the understanding of them is still incomplete and evolving. This is also why more research is needed. I hope that collaboration between banking supervisors and researchers will increase in the future.

Thank you for your attention.