Of Dogs, Frisbees and the Complexity of Capital Requirements Andrew Cornford

Both Basel II and Basel III have been criticised for their complexity. But as the process of extension and emendation of the Basel capital framework continues, major figures from the regulatory community have recently been expressing their reservations concerning the framework’s complexity increasingly forcefully and posing the question of whether revisions in the direction of simplification are not now required.

The debate on the complexity of Basel II and Basel III is closely connected to that on the adequacy of the capital requirements. Thus, although in what follows here the principal focus will be on the issue of complexity, there are also references to critics’ observations on capital levels.

The criticisms of two senior regulators have been recently the subject of special attention: that of Thomas Hoenig, currently a director of the US Federal Deposit Insurance Corporation (whose chairman until 2011, Sheila Bair, was one of the earliest and most forceful critics of Basel II) and previously head of the Kansas City Federal Reserve Bank, and that of Andrew Haldane, Executive Director, Financial Stability, Bank of England.

Hoenig’s concerns are equally with the measures of risk used to estimate capital requirements and with the levels of capital prescribed.

Basel II, he notes, was a response to pressures from economic and financial experts and many of the world’s largest banks for “a more sophisticated and flexible risk-based standard”. The attempt to achieve this resulted in an enormous proliferation of the number of risk weights in Basel II. The flaws in the rules of Basel II exposed by the current financial crisis have led to its replacement by Basel III.

The revised framework “is intended to be a significant improvement over earlier rules. It does attempt to increase capital, but it does so using highly complex modelling tools that rely on a set of subjective, simplifying assumptions to align a firm’s capital and risk profiles. This promises precision far beyond what can be achieved for a system as complex and varied as that of US banking”.

Increased complexity has been accompanied by increased leverage and thus a reduction in the “fortress capital” required for a bank’s “fortress balance sheet”. Historically, the ratio of tangible equity (equity without add-ons such as goodwill, minority interests, deferred taxes and other accounting entities) to tangible assets (all assets less intangibles) varied between 13 and 16 per cent.

These figures are in sharp contrast with recent experience: between 1999 and 2007, the United States banking industry’s ratio of tangible equity to tangible assets declined from 5.2 per cent to 3.8 per cent. For the ten largest banks, the ratio reached a level of only 2.8 per cent in 2007 while its risk-based capital ratio, i.e. that estimated in accord with the rules of the Basel capital framework, remained at approximately 11 per cent, a figure achieved by means of a shrinkage of the denominator of the ratio through the application of increasingly favourable risk weights to the assets of which it consisted.

The principal focus of Haldane’s paper is the growth in the length and complexity of the Basel capital framework, and the consequences for the sheer scale of financial regulation and supervision. The 1988 Accord, Basel I, which was the first genuinely international prudential agreement, was a mere 30 pages long. In 1996, the accord was extended to incorporate the Market Risk Amendment which allowed banks to use internal models to calculate the regulatory capital for exposures to market risk.

As Haldane puts it, “With hindsight, a regulatory rubicon had been passed. This was not so much the use of risk models as the blurring of the distinction between commercial and risk judgements… The regulatory backstop had been lifted, replaced by a complex, commercial judgement… A revised Basel Accord, Basel II, was agreed in 2004. It followed closely in the footsteps of the trading book amendment. Internal risk models were… not so much permitted as actively encouraged, with internal models designed to deliver lower capital charges.”

Reflecting the greater detail and complexity, Basel II was 347 pages long. The response to the financial crisis has consisted of efforts to fill the gaps exposed by the crisis, and the new framework, Basel III, agreed in 2010, is contained in documents 616 pages long.

The consequences for national rule books have been dramatic and the domestic documentation in both the United States and the United Kingdom now exceeds 1,000 pages. In 1980 in the United Kingdom, there was one regulator for every 11,000 people employed in the financial sector, but in 2011, there was one regulator for every 300 financial employees.

In the United States, there are three regulators for every bank. In the United Kingdom, when regulatory reporting was first introduced in 1974, there were approximately 150 entries in the typical bank’s regulatory return, but now, approximately 7,500 data cells are required. The quarterly reports required by the Federal Reserve in the United States covered 547 Excel columns in 1986 and 2,271 Excel columns in 2011.

Haldane presents evidence which does not support the superiority of a capital ratio based on risk-weighted assets as a statistical determinant of bank failure in comparison to a simple leverage ratio in which assets are equally weighted. On the contrary, in many of the statistical tests reported in his paper, the leverage ratio (the ratio of capital to aggregate assets and other exposures) outperforms the capital ratio based on risk-weighted assets.

Similarly simpler measures of capital (including only equity) are better correlated with bank insolvency than more complex measures. The performance of complex measures as forecasters of bank failure does improve as the number of observations in the statistical tests (numbers of banks and numbers of periods) increases but with the implication that such measures become useful as policy tools only when the amount of information supporting them is large.

Haldane argues that this forecasting performance exemplifies the way in which optimal decision-making strategies depend importantly on the degree of uncertainty about the environment, which he characterises as model uncertainty. Other things being equal, smaller samples are associated with greater model uncertainty, and simple, heuristic strategies then are likely to outperform more complex ones.

One example chosen by Haldane to illustrate this point is provided by investment strategies. A simple rule such as attributing equal weights to each of the assets in a portfolio is likely to outperform a more complex one incorporating estimates of means and variances.

As another striking example, Haldane cites the catching of frisbees. If approached through the laws of physics, solving this problem would involve weighing a complex array of factors including wind speed and frisbee rotations. Yet, the average dog is capable of mastering the art of catching a frisbee. The secret of the dog’s success is following a simple rule, namely, running at a speed that keeps the angle of gaze to a frisbee roughly constant.

As Haldane points out, regulation of finance has become extraordinarily complex if carried out on the basis of techniques and instruments honed on the basis of the multiple determinants of recent experience with financial crises. As he puts it, “To ask today’s regulators to save us from tomorrow’s crisis using yesterday’s toolbox is to ask a border collie to catch a frisbee by first applying Newton’s Law of Gravity”.

Even before recent criticism of the complexity of Basel III, the Basel Committee on Banking Supervision had established a Task Force on Complexity and Comparability. This group is to report to the Basel Committee in December of this year. A major aim of the Task Force will be to identify features of Basel III which do not contribute to the capital framework’s principal objective of strengthening banks’ defences against risk (and can thus presumably be dropped).

More generally, the reaction of defenders of the overall approach of Basel III amongst senior regulators and policymakers has been to accept the case for simplification so long as it does not compromise the objective of an indispensable level of ranking of different risks.

As Stephen Cecchetti, economic adviser to the BIS (Bank for International Settlements), put it at a recent conference in Lisbon, “The solution is not to throw risk adjustment out of the window… Instead we need the simple response, namely to make risk adjustment simpler. That means a coarse scale of risk weights. When a ranking of riskiness cannot be made with real confidence, assets should be lumped together.”

Nevertheless, in Cecchetti’s view, sole reliance on a leverage ratio (a prescribed level of capital in relation to aggregate exposures) could encourage excessive risk for a given level of leverage. How the greater simplification favoured by such regulators and policymakers should be incorporated in Basel III is not yet clear.

What are the solutions proposed by the critics of complexity? The most drastic, advocated by Thomas Hoenig, is rejection of Basel III and its replacement by a simple tangible equity-to-tangible assets ratio, the historical focus of markets and investors in assessing a bank’s soundness. This uncompromising proposal would not be acceptable to the great majority of regulators in its current form.

Haldane’s approach is more eclectic and comprises instead amendment and extension of the agreement as well as supplementing it by other measures likely to reinforce the strengthening of banks’ defences against risk. At a general level, he would like to see more prominence than currently in Basel III attributed to Pillar 2, that part of the agreement covering the supervisory process.

Here, he would like to see a rebalancing in Basel III towards supervisory judgement and away from prescriptive rules, which in financial supervision as in certain other professions such as medicine, “have generated a wood-from-trees problem”, which leads to “defensive, backside-covering behaviour” on the part of those involved, thus increasing risk in the system. Such an approach would imply not reliance on a greater number of supervisors but rather fewer more experienced supervisors hopefully conducting supervision on the basis of a smaller, less detailed rule book.

But how would Haldane simplify the rules?

  • Firstly, he proposes less reliance on internal models for the estimation of banks’ exposure to risk. One possibility would be the introduction of limits or floors on models output, thus restricting the extent to which they can reduce regulatory capital requirements. Another, clearly favoured by Haldane, would be to remove internal models from the regulatory framework and replace them with standardised approaches to measuring credit and market risk which classify exposures on the basis of broad asset classes.
  • Secondly, he favours placing capital and leverage ratios on an equal footing within the framework. The leverage ratio, already included in Basel III, Haldane would like to see increased – at least for large banks – from three to at least seven per cent.
  • Thirdly, he would target what he regards as the excessive complexity widespread in large, complex financial firms which in his view have become too large to manage. He considers that such complexity is actually “subsidised” by the current Basel framework owing to the incentives which it provides to complexity through its encouragement of the use of internal models. This is an argument for a regulatory (presumably capital) charge for complexity. Such a charge should be directed at, inter alia, banks’ mutual exposures to other banks and financial firms within the financial sector, since such exposures are often associated with opaque transactions whose potential risk is difficult to assess. The rules of Basel III are actually being changed in this direction.
  • Fourthly, he favours structural change in favour of increased separation of commercial and investment banking. There are already indications of likely regulatory reform along these lines due to the recommendations of the Independent (Vickers) Commission on Banking in the United Kingdom and of a group of European Union experts chaired by Erkki Likanen, governor of the Bank of Finland.

Haldane’s proposals have a general character and do not target any particular group of countries. However, the emphasis on simplification and on reduced reliance on internal models for the purpose for estimating risk exposures is arguably of special relevance to Emerging Market and Developing Economies (EMDEs).

Embodying greater simplification in actual rules could be done in various ways.

One might be to follow the precedent already set in the Annex of Basel II by the inclusion of the Simplified Standardised Approach. This was not an alternative approach for determining regulatory capital but rather the collection in a single place of the simplest options in Basel II for calculating risk-weighted assets.

Extension of this approach to Basel III could involve selection of all the simpler options for calculating risk-weighted assets in a single Annex. The selection might be accompanied by some broadening of the classification of exposures, thus leaving more scope for discretion as to the drafting of the national rules through which Basel III would be introduced, and for supervisory judgement in their application.

Alternative rules to those in the existing text of Basel III would be likely to involve not only the broadening of the classification of exposures – a point apparently accepted by some of those defending Basel III against critics of its complexity, as mentioned above – but also an increased dependence on standardised rules in place of excessive reliance on internal models.

An interesting suggestion which might contribute to change for this purpose was recently made by Adair Turner, chairman of the United Kingdom Financial Services Authority. As he put it, “My own conclusion is that when we moved from Basel I to Basel II we made a mistake, and that we should have maintained for each asset class an absolute minimum, using models to identify whether and where higher minimum capital was required. And I suspect we should and will migrate to such a system over time.”

A comprehensive, simplified version of Basel III could eventually include not only the simplification of exposure categories already mentioned but also the setting of minima as suggested by Adair Turner.

In consideration of the global introduction of Basel III, its character as soft international law, i.e. standards agreed as statements of intent which allow for a measure of national interpretation when embodied in national rules, should be remembered. Arguably, the emphasis on the role of Basel III in the enhancement of competitive equality between the banks of different jurisdictions is indeed appropriate in the case of the major developed economies and perhaps also for the emerging-market economies which are now also members of the Basel Committee.

But the appropriateness of a uniform set of rules for all countries is questionable. This was implicitly recognised in the multiplicity of options for the measurement of risk exposures in Basel II. These were designed to accommodate financial systems of different levels of complexity.

The need for such accommodation is also recognised in the rules of Basel III for the control of liquidity risk, which allow for national variation in the availability of liquid financial instruments for this purpose. A measure of diversification in the national rules introducing Basel III, which takes account of variations in local conditions and development objectives in EMDEs, seems not only inevitable but also desirable.

(The quotations in this article are from Thomas Hoenig, “Back to Basics: A Better Alternative to Basel Capital Rules”, address to The American Banker Regulatory Symposium, Washington, DC, September 14, 2012; and from Andrew Haldane and Vasilious Madouros, “The dog and the frisbee”, speech at the Federal Reserve Bank of Kansas City’s 366th economic policy symposium, “The changing policy landscape”, Jackson Hole, Wyoming, 31 August 2012.)

(This article was originally published in SUNS- South-North Development Monitor on 19 November 2012)