Restoring confidence in banks

Keynote address by Mr Stefan Ingves, Chairman of the Basel Committee on Banking Supervision and Governor of Sveriges Riksbank, to the 15th Annual Convention of the Global Association of Risk Professionals, New York, 4 March 2014.

As prepared for delivery

Introduction

Good morning. It is my great pleasure to be with you here today, and I would like to thank the Global Association of Risk Professionals for their hospitality.

The invitation to speak at this important event is especially welcome as we in the Basel Committee view conferences like this one as an essential part of the conversation between regulators and the industry. Moreover, since some of what I have to say today will delve into technical issues, an audience of risk experts must be close to ideal. The location, too, is appropriate. It was in this city, seven years ago, that the seemingly solid ground under the financial system started to shift under our feet. Even now, it's fair to say that we are still working to rebuild confidence in the stability of the financial system. Today, I would like to briefly revisit some of the reasons why confidence was lost. Then I will outline what the Basel Committee is doing to help restore it.

How confidence was lost

I do not need to remind this audience that banking is a business built on confidence. As I have recently said on another occasion,¹ modern banks are highly leveraged institutions, and fractional reserve banking can only be successful when lenders to a bank have full confidence that it has the financial strength to meet its obligations as and when they fall due.

The financial crisis, which began seven years ago, was in essence a collapse in confidence. It stress-tested regulation just as much as it did banks. Both were found wanting. Bank risk managers severely underestimated the risk that such a widespread loss of confidence in the banking system could occur. The regulatory framework was inadequate to protect against such a crisis of confidence when it did. The result was a crisis that continues to impose substantial costs on society.

This crisis of confidence - which is what underlies a liquidity crunch - reflected uncertainty. Everyone knew that banks had taken hits from the subprime fallout. But in which banks were those losses hiding, and how large were they? Was the capital that banks reported really present and ready to absorb these losses as intended?

The shortcomings in risk management and regulatory settings exposed by the crisis make for a long litany. Today, I will focus on just three issues - ones that directly relate to confidence in the underlying soundness of banks.

First, in some jurisdictions the existing regulatory framework allowed equity capital - the main shock-absorber for losses - to be run down as low as 2% of risk-weighted assets. And it was not just the amount of capital that was deficient: so was the quality. Notably, banks could stock the remaining 6% of required capital with cheaper, debt-like instruments that, as soon became clear, could only be used to cover losses after a bank had failed.

Second, the prudential framework filtered out - that is, ignored - some mark-to-market losses on holdings of securities. But investors weren't willing to overlook such losses - on the contrary, they were extremely concerned about them, and they started to reassess bank capital ratios in that light. The focus on simpler measures - such as a leverage ratio based on tangible common equity - became stronger as a result.

Third, under existing accounting standards, bank provisioning for credit losses on their loan books was backward-looking. The incurred loss model that underpinned IFRS and US GAAP prevented banks from making forward-looking assessments of likely losses. So provisions had not been adequately built up in good times, and there was considerable uncertainty about what additional provisioning might be needed.

All of this meant that, when confidence in banks was most needed, the key regulatory metric of financial health - the regulatory capital ratio - was increasingly discounted because it potentially overstated a bank's true loss-absorbing capacity.

How Basel III is helping to restore confidence

The Basel III package is designed to address these shortcomings. As you all know, banks will be required to hold substantially more capital, with the minimum common equity requirement - CET1 - rising from 2% to 4.5% of risk-weighted assets. The capital has to be of higher quality too, with common equity at the core, and standards to ensure that other types of capital instruments are truly loss-absorbing. And the prudential filters, which allowed banks to avoid facing up to losses on their holdings of marketable securities, have been removed.

Then there are measures to protect that capital backing. The capital conservation buffer provides an additional 2.5% of CET1, but allows banks to dip into their capital if their capital ratio deteriorates. At the same time, it provides disincentives for banks to do so, by imposing corrective actions on dividend and bonus payouts to help restore a bank's capital strength. The incentive is clearly there for banks to operate with capital levels above this buffer level, and to only use it when truly needed.

We have also sought to make the capital framework more countercyclical. Banks will hold more capital in good times to prepare for inevitable downturns: this is the so-called countercyclical capital buffer, which potentially adds up to an additional 2.5% of CET1. And then the very largest banks - those designated as globally systemically important - will need to hold extra capital to take account of the extra damage their failure would inflict on society.

Beyond more and better quality capital, Basel III introduces other measures to bolster bank soundness. Key among them is a simple backstop in the form of a (non-risk-based) leverage ratio. In addition, there are the world's first international standards for bank liquidity and funding in the shape of the Liquidity Coverage Ratio (LCR) and the Net Stable Funding Ratio (NSFR).

In parallel, accounting standard setters have been working to improve the framework for provisioning against expected credit losses.

The good news is that both the IASB and the FASB have developed new standards that will move the accounting framework from one based on incurred losses to one based on expected losses. The Basel Committee has strongly supported this switch, and we are pleased that both Boards have seen fit to make this move. It is somewhat disappointing that the Boards have not been able to converge on a single standard, which would improve things even further. But the core objective of the supervisory community has been, as I said, to promote more forward provisioning practices in banks, as this will also aid confidence that bank balance sheets more accurately represent the underlying financial position at any given time.

Especially in the present company, I don't need to go into detail on these reforms. Their import will be clear to you. In short, the banking system will in future have materially greater loss-absorbing capacity. This should provide greater confidence in the ability of banks to survive periods of stress, whenever they next occur.

The problem of risk-weighted asset variability

To borrow an automotive metaphor, the financial system now has in place a greatly improved set of airbags to protect it from adverse shocks and minimise damage. However, one needs to be careful with that airbag comparison. I have to concede that, when carmakers first installed these devices (as some of you will be old enough to remember), they stirred up quite a bit of controversy. People asked if they would work as intended, or whether they'd go off too soon, and so on.

It's not so different with Basel III. People are asking if we can rely on the risk-weighted asset measurements that stand at the Basel framework's core - or, more precisely, in its denominator. At a stretch, you might compare these measurements with the sensors that set off the airbags in your car. If the sensors don't work as intended, the airbags won't be of much use.

The Basel Committee takes these questions very seriously. As matters stand, we've so far published three studies on risk-weighted asset variability - two for the trading book and another for the banking book. Taken together, these show that actual risks drive the lion's share of differences in risk weights and capital requirements. This, of course, is just as it ought to be.

But variations also arise from supervisory and bank practice-based idiosyncrasies, and these can result in material discrepancies. While it is difficult to be precise on how much scatter is "too much", the range of bank practice-based variations is uncomfortably wide.

Let me give you a sense of that scatter. If we just take the banking book results - which only focused on risk-weighted assets for sovereigns, banks and wholesale credit - two banks with exactly the same assets could report capital ratios that differ by 4 percentage points. That is, if the most conservative bank reported a regulatory capital ratio of 8%, the least conservative one would report 12%. Of course, this simply compares the extremes in the sample - most banks were much closer to the average. But the potential for differences this wide, particularly as they are derived from only a part of a bank's business, weakens confidence in the measurement of bank capital.

The first step to solving a problem is acknowledging you have one. From the Committee's perspective, we see there is a key problem that needs to be fixed. To bolster confidence in the reliability of their capital ratios, banks should also have a keen interest in ensuring that their risk measurement methods are viewed as robust and credible. Any doubts in this area also begin to call into question stress-testing results and risk management systems more generally, since these are all built on the same foundations.

Thus far, the response from the banking industry on this issue has been mixed. Some banks - and organisations like our hosts today, GARP, and the IIF - have been keen to engage with the Committee and are endeavouring to offer constructive ideas and suggestions on how to reduce variability. This is very welcome. Some others in the industry have been less keen to acknowledge there is a problem. The message I would like to leave you with today is that there is one, and we plan to do something about it. And, ideally, we would like your help. Success is critical to restoring confidence in bank capital ratios and this is of great importance to both banks and regulators alike.

So what is being done? Clearly, when the causes are multifarious, there can be no single "silver bullet" solution for risk-weighted asset variability. It follows that the regulatory response also has to pursue several different leads. In fact, these efforts can be grouped into three main strands, that of (i) policy, (ii) supervision and implementation, and (iii) disclosure.

I'll start with the policy strand.

At this stage, the Basel Committee's goal is to reinforce the risk-based framework, not to replace it. That is, we want to develop a set of simplifications and safeguards that will help limit variability but still provide for appropriate risk sensitivity in risk-weighted asset measures. More importantly, they must also provide greater comfort that the results are comparable from bank to bank, and over time.

One way to limit variability is to constrain banks' modelling choices - for example, with respect to the choice of stressed periods or the scaling of short-horizon estimates to longer horizons. In addition, one could envisage improvements to the data requirements on which models are built, limiting certain modelling assumptions or techniques, and strengthening validation requirements. Proposals along all these lines have been built into the Committee's fundamental review of the trading book, supplemented by more guidance on expected supervisory practices.

For a more direct impact, we are also looking at the role of floors and benchmarks within the Basel III framework. Floors are not a new phenomenon in the capital framework: we have them already at the aggregate level - in the form of the transitional floor relative to Basel I that remains in place - and for some risk parameters - for example, the minimum loss-given-default for residential mortgage loans. But we need to critically examine whether the current floors are effective. It is clear, for example, that the aggregate floor on modelled risk weights based on Basel I calculations needs to be replaced, and the Committee is actively looking into the design of a suitable successor.

We also need to examine whether there is a case for the greater use of floors at a more granular level, particularly for products and markets where data are limited or where there are other characteristics that make them hard to model reliably. In these instances, the case for relying completely on bank models to determine regulatory requirements is far from compelling.

One of the criticisms we hear of any move to place constraints on internal modelling is the potential for adverse impacts on banks' risk management incentives. It is claimed we will drive adverse behaviour, or limit the investment in risk management. Such criticisms reflect poorly on banks, in my view, as they imply banks will only invest in more advanced risk management practices if incentivised (or subsidised) to do so by regulators offering them lower capital requirements. And from a regulatory perspective, maybe the right structure of incentives, at least for large international banks dealing in complex financial products and markets, is to impose a capital penalty where those banks do not have advanced risk management capabilities. In other words, these capabilities should be regarded as something regulators should expect as a minimum, and incentives can be created by imposing additional costs where they do not exist. So while I agree that we should be careful not to provide disincentives to improve risk management, equally we must avoid creating a regulatory system that merely incentivises modelling for the purposes of optimising risk-weighted assets.

Finally, as mentioned earlier, we intend to ensure that the leverage ratio fulfils its intended role as a backstop to the risk-based regime. It should be clear to everyone that the case for a stronger leverage ratio will only grow further if risk-weight variability is not adequately dealt with. Some people continue to question the merits of a leverage ratio. But given the very high levels of leverage that could be generated under the risk-based regime, this is a sensible measure to introduce as a backstop.

That brings me to supervisory implementation measures.

Most immediately, supervisory action is already taking place. Many supervisors are responsible for only a handful of banks that use internal ratings to gauge risks, and it can be difficult to generate reliable benchmarks from a small sample. Our three studies on risk-weighted assets have provided supervisors with a much clearer picture of how their banks stack up against a much larger, international peer group, and supervisory action is already being taken on some outlier banks.

In like fashion, the Committee's country-by-country or jurisdictional RCAP assessments are helping to reduce the type of variability that arises from undesirable differences in national regulations. Indeed, this is one of the Committee's major aims on the Basel III implementation front - RCAP stands for "regulatory consistency assessment programme", after all. The differences it picks up may seem to be small issues of technical detail, but when aggregated they can often have a material impact on capital outcomes. Thus, their removal adds to the consistency of RWA calculations.

In the context of implementation work under the RCAP, we are considering ways of monitoring risk-weighted asset outcomes over time. A proper process for assessing such outcomes could have several benefits. First, it could provide the Committee with evidence for how far Basel III implementation, supervision, and changes in banks' own internal risk management practices are having the desired impact. In addition, it will better equip all of us to respond to changes in modelling behaviour and the evolution of average risk-weighted asset density levels. It will also speed the development of improved benchmarks and validation tools. Some have even suggested setting up a permanent capacity for periodic monitoring of risk-weighted asset variations for the core asset classes based on hypothetical portfolio exercises.

To improve implementation, the Committee is also looking at the issue of national discretions. This entails an evaluation of whether more can be done to reduce variability from these sources by eliminating discretion where possible, and strengthening supervisory guidance on model validation, review and approval.

Finally, let me say a few words about disclosure.

The financial crisis made it clear that the existing disclosure regime did not allow market participants to get a clear and timely picture of a bank's material risks. Nor did it provide enough information to let them assess a bank's ability to withstand market turmoil. Thus, it was already apparent that a more fundamental revision of the disclosure regime and an overall strengthening of the Basel framework's Pillar 3 were needed.

When it comes to bank modelling practices, the Committee plans to propose improvements to the information that banks provide on their risk profiles and risk-weighted asset differences. The Committee is conducting a thorough review of Pillar 3 disclosure requirements with a particular focus on comparability across banks as a prerequisite for efficient market discipline. At present, that comparability is lacking. As an example, counterparty credit risk is not always distinct from credit risk in banks' disclosures. One natural response would be to standardise the presentation of core information on prudential metrics, and proposals to that end are in the pipeline. I hope that we will be able to release these to you for comment over the summer.

A balance does need to be struck between the use of prescribed templates to promote reporting comparability, and the need to allow banks' management enough scope for meaningful commentary on how they see their institution's specific risk profile. But for key data on a bank's capital and risk-weighted assets, a more standardised disclosure framework will form a vital part of the global response to the existing shortfall in market discipline.

Conclusion

There's no doubt that Basel III and the other post-crisis reforms will produce a more resilient financial system. We intend that banks should be able to ride out the next set of financial shocks much more capably than before, both for their own sake and for that of the credit they intermediate. Restoring confidence in banks is critical to the end-objective, which is to buffer the real economy against financial stress.

Basel III has materially addressed shortcomings in the numerator of the risk-based capital ratio, and it has added additional safeguards in the form of leverage and liquidity requirements. But, to use another Swedish proverb, don't celebrate until you have crossed the creek (Ropa inte hej förrän du är över bäcken). Question marks remain about the reliability and comparability of risk-weighted asset calculations and, until these are resolved, confidence in capital ratios cannot be fully restored.

When deciding what to do about this issue, let me assure you that supervisors are keen to engage the industry in a constructive dialogue on the best way forward. Indeed, I hope my remarks today will be taken as part of that dialogue. My focus today has been on the actions already taken by the Basel Committee, or in its pipeline, that will help restore confidence in banks. But this is just one element in the equation. Banks themselves must take a leading role, as the ultimate restoration of confidence depends on the actions they take. The Committee is eager to hear from banks and other interested parties on their preferred course of action. Industry-led initiatives to foster improved understanding of these issues are therefore more than welcome - they are a vital part of the process.