|
|
Rather predictably, my remarks tonight will focus on banking issues. More specifically, I would like to talk about international banking regulation. Later this year we expect the European Commission to issue the legislative proposals to implement the final package of the reforms agreed by the Basel Committee on Banking Supervision (BCBS). This last step completes the policy response to the great financial crisis. The Governors and Heads of Supervision (GHOS) in turn have committed to refrain from launching major adjustments to the international standards for the foreseeable future.[1]
But the last lap of this long process is still facing fierce opposition from some in the banking industry who argue that the impact of the reforms might adversely affect banks’ capacity to support the recovery from the coronavirus (COVID-19) pandemic shock. I am firmly convinced that a full and timely implementation of this last set of international standards is in the interest of all stakeholders. It requires only limited adjustments in the short term but will deliver the necessary structural improvements to our regulatory framework as well as sizeable and long-lasting benefits for our economies. Most importantly, I believe that the effectiveness of international standards, which is of great value to supervisory authorities and international banks alike, crucially relies on the commitment of all signatories to faithfully implement them in their domestic jurisdictions.
Before I go into the detail of the final stretch of this journey – the implementation of the final Basel III package – I would like to briefly look back on how the journey began and how banks’ internal models, which the package addresses, came to the fore of international discussions.
The BCBS was established in the mid-1970s after the collapse of the German bank Herstatt. Its creation reflected the awareness that, after three decades of financial stability (although some would perhaps say three decades of financial repression), the collapse of the Bretton Woods system of fixed exchange rates and the ensuing free-floating exchange rate system had given new momentum to the international banking business, and banking regulation and supervision needed to catch up.
As well documented by Charles Goodhart in his excellent book on the early years of the Basel Committee[2], the newly constituted Committee was initially more concerned about supervisory cooperation than standard setting. The Basel Committee’s first document of note was in fact the Basel Concordat. In the wake of the collapse of Herstatt, the Concordat sought to set out the respective responsibilities of what we refer to in supervisory jargon as the “home” and the “host” authorities of international banking groups, as well as their duties of cooperation. The Concordat was revised in the aftermath of other international banking crises, like that of Banco Ambrosiano, and remained for many years the main point of reference for the activities of the Committee, and its main focus was international supervisory cooperation.
The standard-setting drive came later, as it became clear that supervisory cooperation, which sought to avoid disruptive crises of international banking groups, and standard setting, which aimed to foster an international level playing field, were closely connected. Competing regulatory standards from different jurisdictions endangered the very possibility of global financial stability and, in the end, would threaten both the level playing field across regional markets (including the United States and Europe) and the feasibility of international supervisory cooperation.
That is why, to effectively fulfil their national mandates, the national competent authorities and the central banks of the most relevant banking jurisdictions in the world not only had to cooperate ex post – in the run-up or, worse, in the aftermath of the crisis of a specific bank – but they also had to cooperate ex ante, so to speak, to align the regulatory requirements of internationally active banks at the technical level. The firm commitment by all participants to transpose what had been agreed in Basel into legally binding legal acts in each national jurisdiction – or supranational in Europe’s case – coupled with the peer review of the existing legislative frameworks, completed the institutional set-up of this global standard-setting body and laid the foundations of its credibility and reputation vis-à-vis banks and other market participants at the global level.
Now, it is important to underscore that Basel regulatory cycles are very long policymaking processes that take many years to complete. The highly technical nature of the subject matter requires detailed discussions at the negotiating table, consultations with the industry, impact assessment exercises and appropriate transitional periods to be defined before the new rules can actually become fully applicable. This last point is crucial to avoid disrupting the macroeconomic cycle with sudden changes in the regulatory requirements which modify incentives, alter the functioning of banks’ business models and, ultimately, affect the allocation of financial resources to the real economy.
Let’s take the first Basel Capital Accord as our starting point. Negotiations on a global regulatory capital standard for banks had already started in the early 1980s in the aftermath of the Latin American debt crisis, but the Committee only reached an agreement in 1988, with an implementation date being set for the end of 1992. This was achieved in the European Union (or European Community, as it was at the time) in 1989 through a suite of Directives, including the Solvency Ratio Directive and the Second Banking Coordination Directive, which were then transposed into the legal frameworks of all Member States.[3] Compared with the later accords, Basel I was notable for its simplicity but also its very limited scope for risk sensitivity. It established a simple minimum capital requirement of 8% of own funds of variable quality (common equity could, in the end, be as low as 2% of risk-weighted assets) and a few coarse risk weights for specified asset classes divided into 0%, 20%, 50% and 100% risk buckets. One of the key trade-offs in prudential regulation – between the simplicity of the rules and their risk sensitivity – would soon take centre stage.
Two developments made supervisory authorities realise that these broad-brush requirements were fast becoming obsolete. First, exactly in the same years of the definition of the Accord banks’ internal risk management techniques started to become more sophisticated[4], so that the risk sensitivity of the regulatory framework was falling out of touch with best market practices for the internal allocation of economic capital. Second, research conducted by the Federal Reserve System showed that banks had managed to develop a number of practices, such as expanding off-balance sheet activities, like securitisation, or adjusting effective exposures to risk within the broad buckets defined by the first Accord, to increase their level of risk-taking without being captured by regulatory requirements.
The need to bring regulation closer to the real risks involved in certain activities and prevent regulatory arbitrage opened the door to the regulatory recognition of banks’ internal models. If it was possible to measure risk in a more granular fashion for risk management purposes, then the same method should be used to quantify regulatory capital, as it would make the regulatory framework more efficient and more incentive-compatible.
In this context, in 1996 a market risk amendment to the Basel I Accord was agreed. For the first time, banks could use internal risk models to calculate regulatory requirements for market risk (but not credit risk). With the second Basel Capital Accord (known as Basel II), which was agreed in 2004 to become applicable in 2007, the internal model approach was extended to the calculation of capital requirements for credit risk, counterparty credit risk and operational risk.[5] The use of internal models was embedded in prudential requirements to calculate risk-weighted exposure amounts, the denominator of the capital requirement ratio.
It is sometimes erroneously argued that this move towards the use of internal models in the calculation of capital requirements was at the origin of the great financial crisis. Some critics saw the reform as epitomising a broader policy shift towards deregulation of financial markets. For these critics, relying on banks’ own internally developed methodologies to calculate their regulatory requirements was like letting the fox guard the chicken coop.
In actual fact, some of the jurisdictions at the epicentre of the great financial crisis, such as the United States, had not yet allowed banks to use internal models. The Basel Committee’s “use test”, which required banks to use the same models for internal risk management purposes as for the calculation of regulatory requirements, was intended to foster reliance on best market practices. And the ease with which sophisticated international banks managed to circumvent the simpler regulatory requirements of Basel I had, in the end, left the international supervisory community with no credible alternative.
Yet, it would be equally misleading to overlook the weaknesses in banks’ internal models and their use for regulatory purposes that were highlighted during the crisis. The regulatory framework for market risk had to be amended in 2009 because of shortcomings in the ability of value at risk (VaR) models to capture default risk and migration risk, with stressed VaR requirements introduced as a result.
What’s more, the increase in the risk sensitivity of prudential requirements caused by the Basel II standards generally led to an unintended reduction of risk density, i.e. the average risk weight for the largest and internationally active banks, and to unwarranted variability in risk-weighted assets, meaning similar portfolios were carrying vastly different capital charges without apparent differences in the underlying risk.
As the risk density decreased, banks were able to carry out more business based on the same absolute amount of own funds, thus expanding their balance sheets. They increased their leverage while still complying with the risk-weighted capital requirements and, in some cases, even increased their regulatory ratios.
This is well documented in a 2014 report by the European Systemic Risk Board’s Advisory Scientific Committee. The report shows that in the run-up to the great financial crisis, the median risk-weighted capital ratio – so Tier 1 capital to risk weighted assets – remained almost stable at around 8%, whereas the leverage ratio fell by almost half. By 2008 the median leverage ratio of the 20 largest European banks had dropped to just over 3% (assets were 32 times capital) from a previous average of 6% (17 times), while median risk-weighted regulatory capital ratios remained essentially unchanged.
An even more striking finding from the report is the negative correlation between the risk-weighted Tier 1 capital ratio and the equity-to-assets ratio, another measure of leverage. In short, banks that were more capitalised on a risk-weighted basis also had a lower leverage ratio, something that in itself called for a careful investigation into the role of the risk-weighted assets framework, and internal models, in banking regulation.[6]
Other analyses[7] show that shortly before the great financial crisis, the average risk-weighted solvency ratio of the large international banks that entered resolution or needed government support was well above the minimum requirements and not statistically different from the average solvency ratio of the banks that did not experience a crisis. The Tier 1 capital ratios appeared uninformative about banks’ true default probabilities and, therefore, the actual risks on their balance sheets.
We all know the intrinsic limitations of risk measurement models and the implications of their use for regulatory purposes. Of course, it is as much a supervisory problem as it is a regulatory one. As a supervisor I have always been deeply interested in the risk sensitivity of capital requirements. Intellectually, it seems clear that riskier assets should attract higher capital charges than less risky ones. But the real issue, which is almost a philosophical question, is whether financial risk can be accurately measured. As you probably know, there are splendid pages by Keynes, among others, on the irreducible uncertainty of economic outcomes and the impossibility of measuring real risk through probabilistic calculations.[8] To make this fundamental problem more tractable, supervisory authorities need to be in a position to robustly challenge banks’ modelling choices and their conservatism....
More at EBA