While reading the “Fundamental review of the trading book - consultative document” (published on 3 May 2012; comments sought by 7 September 2012) by Basel Committee on Banking Supervision (BCBS), I for some reason remembered this popular quote: “Things not worth doing are not worth doing well.” (Ken Blanchard) Doing well things that should not be done at all seems to be exactly what banking supervisors are doing when introducing one new or updated rule after another for the calculation of the regulatory capital, in this case for the market risk regulatory capital.
First, they convincingly explain some shortcomings of the framework before the crisis started (read: recognise blunders in the earlier regulatory regime in a professional manner). In the above referred document the BCBS is highlighting that:
* Regulatory boundary between the trading book and the banking book has been subjective, and coupled with large differences in capital requirements against similar types of risks across either side of the boundary, the capital framework has proven susceptible to arbitrage;
* When the supervisors witness deterioration in the performance of the internal models, they have no real option to withdraw the model approval (This would require banks to raise new capital which during the times of stress is not very well doable.);
* 10-day VaR metric falls short in ensuring that banks have sufficient capital to survive low probability, or “tail”, events;
* There has been a lack of robust valuation practices.
Comment: Recognising previous blunders is very good, although the mistakes and consequences have been this big that should call into question the competence of supervisory authorities. Indeed, the “but” is that the “confession” is not going far enough. At this point it should be added that due to the extreme complexity of the financial structures, assessing the size of potential losses adequately is well beyond the current human capabilities.
Secondly, they explain what improvements the supervisors have made so far. For example, in response to the weaknesses highlighted by the crisis of 2007-2009 they have introduced Basel 2.5 with the following amendments to the market risk framework:
* The introduction of the IRC (Incremental Risk charge), an additional capital charge intended to capture both default risk and credit rating migration risk;
* The introduction of stressed VaR measure;
* (Some) alignment of the treatment of securitisation exposures across the banking book and the trading book;
* Improved risk factor coverage of internal models;
* Enhanced prudent valuation guidance.
This means that, on average, the market risk capital requirements of large banks would more than double. Further, in Basel III package three more changes have made to the capital treatment of trading activities and market risk: a number of amendments to strengthen the counterparty credit risk framework, unrealised gains and losses will no longer be filtered out of Common Equity Tier 1 capital, and Tier 3 capital is no more available to meet market risks.
Comment: At first glance, it seems that supervisors are learning from their previous mistakes – and so it is intended to be presented, I believe. Note, however, that the above amendments were just quick fixes, an attempt to judge increasing capital requirement for the strongly undercapitalised market risk.
The Basel Committee recognised this and the need for initiating a longer term, fundamental review of the risk-based capital framework for trading activities. Yet, it should have asked if risk-based framework is feasible at all if the risks are barely possible to grasp.
Anyway, thirdly they point out drawbacks of the current market risk regime:
* The framework lacks coherence: overlapping capital charges, development and validation of several distinct sets of models, severe strain on supervisory oversight;
* The boundary issue has not been fully addressed in Basel 2.5 and in Basel III: The July 2009 revisions to the market risk framework made only minor amendments regarding the set of products that should be excluded from the trading book;
* Market liquidity risk is not evenly captured;
* Many of the new approaches are still based on a bank-specific view of risk, not that many banks try to exit risk positions simultaneously;
* The structural shortcomings of the standardised approach unaddressed: lack of risk sensitivity, a very limited recognition of hedging and diversification benefits and an inability to sufficiently capture risks associated with more complex instruments;
* There remains a lack of credible options for withdrawal of model approval: aside from multipliers on VaR and stressed VaR, there are limited options for supervisors to deal with poorly-specified internal models.
* The relationship between the capital charges for CVA risk and the trading book regime has not been clarified.
Comment: We see that this time in some cases current capital charge is implied to be rather too high (bankers’ lobby?). Consider the phrases such as “overlapping capital charge”, “very limited recognition of hedging and diversification benefits” in the standardised approach etc. Secondly, it isn’t that clearly articulated here but still implied and tried to deal with later, that regulators (and analysts as well) are in trouble when it comes to comparing market risk capital of different banks that are applying internal models; there may be major variations simply because of models and subjectivity involved.
After that come the proposed revisions. Logically, these are built upon the identified drawbacks of the current regime:
* Reassessment of the boundary between the banking book and the trading book (there are presented two different alternatives to choose from together with the question: which one?);
* Changes in choice of risk metric and calibration to stressed conditions, including moving from VaR to the Expected Shortfall (ES);
* Factoring in market liquidity, including a regulatory framework for assessing market liquidity risk across the trading book, as well as for incorporating the assessment of market liquidity into trading book capital requirements; (What are commenters’ views on the likely operational constraints?);
* Revised treatment of hedging and diversification: a similar parameterisation is expected to be used under the revised models-based and standardised approaches;
* Establishing a stronger relationship between the standardised and models-based approaches (calibration, mandatory standardised measurement, floor (or surcharge) based on the standardised approach);
* Further revisions to the models-based approach, such as more detailed process for determining eligibility of trading activities for the internal models-based approach, new formula for the aggregate models-based regulatory capital, ongoing monitoring of the already approved models with much greater detail than before;
* Revised standardised approach based on the principles of improved risk sensitivity; credible calibration; simplicity, transparency and consistency; limited model reliance; credible fallback in case a bank’s risk management model is inadequate. There are two approaches proposed: 1) partial risk factor approach, and 2) fuller risk factor approach. (Question: which one, and do you have suggestions for improving these approaches?)
The next steps are collecting feedback from the market participants (“commenters”) and performing a quantitative impact study as always.
Why the revision is still a blunder?
When reading these proposed amendments and the discussion points, I get the impression that the supervisors on one hand want to control everything, but on the other hand are not able to design a suitable regulatory framework by themselves, and thus involve “commenters” (which first of all will be banks, I expect) into the process. This looks more like a loss of control.
Standardised approach and models-based approach get closer to each other; why not simply have one single approach for the regulatory capital? I do not believe that (theoretically and questionably) slightly better capital allocation thanks to the internal models would possibly outweigh greater obscurity that these models create, and the major regulatory costs for approving and monitoring the banks’ internal models. I do not even believe that the regulators or bankers themselves are possibly capable to grasp the universe of all the models and assessments. What does this or that estimate actually show? Is it based on the realistic assumptions?
After all, attempts to regulate extreme and incomprehensible complexity can only lead into even more complexity and to the misleading illusion of being precise. Supervisory community in general seems hesitant about banning gambling with the other people’s money altogether; instead they are trying to discourage it by introducing new and more complicated regulatory frameworks. When thinking about all the working hours that go into it, all the redoing is clearly a waste of human resources. Then on one day it will be discovered, that well, the changes still were not that good... We are back at the conclusion that no matter how you regulate a flawed system / financial architecture, it is still flawed. Perhaps supervisors should think more about the rules that would lead to the better self-regulation?