This is my initial reaction to the fundamental review of the trading book. I’ll set out what is in the FRTB in roman, then my reaction in italics. In many cases I’ll quote the Basel document for the former.
First the good news. The Committee recognises that the Basel 2.5 revisions did not fully address the shortcomings of the market risk framework.
- The framework lacks coherence;
- The boundary issue (between the trading book and the banking book) has not been fully addressed;
- Problems remain unaddressed in the standardized market risk rules.
Basel 2.5 was, in my view, a sticking plaster, designed to ameliorate the immediate problem – and to dramatically increase the capital required for the trading book – but not to provide a coherent solution. The FRTB is an attempt at that solution.
The trading book/banking book boundary
The boundary is an issue in that some things were put in the trading book pre crisis which turned out to be illiquid. Trading book rules designed to capitalise short term market risk assigned too little capital to these positions. Therefore the committee wishes to strengthen the boundary criteria.
The Committee is consulting on two possible solutions to the boundary:
- A “Trading evidence”-based boundary where instruments will be admitted to the trading book provided that the bank has trading intent plus proven ability to execute that intent; or
- A “Valuation”-based boundary where the trading book will be defined by those items which are fair-value accounted.
To a certain extent, if the Committee addresses illiquidity as proposed below, then boundary arbitrage might either go away or, more likely, go the other way; the banking book might be cheaper than the trading book post FRTB. In any case, philosophically, the market risk rules should capitalise variation in fair value, so the second approach is clearly more coherent than the first (which would permit fair value variation to go uncapitalised in the banking book).
The Committee, understandably, wants:
- to ensure that regulatory capital is sufficient in periods of significant market stress; and
- to reduce the cyclicality of market risk capital charges.
Calibrating the capital framework to a period of stress helps to achieve both of these objectives.
I agree, but the devil is in the stress here. What we don’t want is banks all having book that would be very robust if the last crisis happened again, but which are all vulnerable to the next one. That’s the problem with uniform capital rules; they encourage herding.
Moving from value-at-risk to expected shortfall
ES will be used instead of VAR in models-based approaches.
While this aspect of the FRTB has received a lot of attention, it is not a massive change. Banks will be able to use their existing VAR infrastructure to calculate ES, and ES charges – while larger than VAR – should behave much like a stressed VAR.
A comprehensive incorporation of the risk of market illiquidity
This, in contrast, is a big change. It will, if implemented, force firms to take asset liquidity a lot more seriously.
- The framework will incorporate an assessment of market liquidity for regulatory capital purposes. Banks’ exposures would be assigned into five liquidity horizon categories, ranging from 10 days to one year.
- Capital will be based on the assigned liquidity horizon.
- There will also be capital add-ons for jumps in liquidity premia, which would apply only if certain criteria were met. These criteria would seek to identify the set of instruments that could become particularly illiquid, but where the market risk metric, even with extended liquidity horizons, would not sufficiently capture the risk to solvency from large fluctuations in liquidity premia.
- The Committee is consulting on two possible options for incorporating the
“endogenous” aspect of market liquidity. (Endogenous liquidity is the component that relates to bank-specific portfolio characteristics, such as particularly large or concentrated exposures relative to the market.) The main approach under consideration by the Committee to incorporate this risk would be further extension of liquidity horizons; an alternative could be application of prudent valuation adjustments specifically targeted to account for endogenous liquidity.
Let’s suppose that the committee is broadly comfortable with the overall level of capital assigned to the trading book after the Basel 2.5 changes. That’s an assumption, but a reasonable one I think, as B2.5 was a quick fix designed to get the level of capital roughly right. In this case, clearly highly liquid areas (FX, most cash equities, some rates) will benefit from these proposed changes, while less liquid ones (most junk credit, most ABS) will suffer.
Treatment of hedging and diversification
The Committee is proposing to more closely align the treatment of hedging and
diversification between the standard rules and internal models.
Notice that this cuts two ways: reducing the diversification (and perhaps hedging) benefits for banks with models permisssions, while increasing them for standard rules banks.
The Committee proposes
- Constraining diversification benefits in the internal models-based approach to address the Committee’s concerns that such models may significantly overestimate portfolio diversification benefits that do not materialise in times of stress.
- Supervisor imposed (stressed) correlations will be a floor to diversification benefit within the internal models regime; and
- Revisions to the standardised approach that will enhance its risk sensitivity.
The hard part here will be improving the standard rules. There is very little detail from the Committee on this aspect.
Relationship between internal models-based and standardised approaches
The Committee considers the current regulatory capital framework for the trading book to have become too reliant on banks’ internal models that reflect a private view of risk. In addition, the potential for very large differences between standardised and internal models-based capital requirements for a given portfolio is a major level playing field concern and can also leave supervisors without a credible option of removing model permission when model performance is poor. To strengthen the relationship between the models-based and standardised approaches the Committee is consulting on three proposals:
- First, establishing a closer link between the calibration of the two approaches;
- Second, requiring mandatory calculation of the standardised approach by all banks;
- Third, considering the merits of introducing the standardised approach as a floor or surcharge to the models-based approach.
This follows a general theme of backstopping models-based capital requirements by cruder, deliberately less risks sensitive calculations (in case the latter are wrong). Again, though, the devil is in the details; how will the Committee make the calibration ‘closer’ and how will it ensure that the two approaches do not drift too far apart?
Revised models-based approach
The Committee proposes to
- strengthen requirements for defining the scope of portfolios that will be eligible for internal models treatment; and
- strengthen the internal model standards to ensure that the output of such models reflects the full extent of trading book risk that is relevant from a regulatory capital perspective.
A key tool here will be desk-level models permission, based on
- P/L attribution and
In other words, a bank won’t have models permission or not (a binary thing), but rather will have permission for some desks but not others. Even the most sophisticated banks may struggle to meet the standards to use models on some exotics businesses, for instance. While this approach makes sense, it is vulnerable to supervisory arbitrage, i.e. it will be vital that individual national supervisors interpret these criteria equitably.
Revised standardised approach
Two suggestions are given:
- Partial risk factor approach: Instruments that exhibit similar risk characteristics would be grouped in buckets and Committee-specified risk weights would be applied to their market value.
- Fuller risk factor approach: map instruments to a set of prescribed regulatory risk factors to which shocks would be applied to calculate a capital charge for the individual risk factors. The bank would have to use a pricing model (likely its own) to determine the size of the risk positions for each instrument with respect to the applicable risk factors. Hedging would be recognised for more “systematic” risk factors at the risk factor level. The capital charge would be generated by subjecting the overall risk positions to a simplified regulatory aggregation algorithm.
The appropriate treatment of credit risk within the market risk framework
Currently the market risk framework capitalises jump-to-default (and ratings migration) risk via IRC models, while charging for credit spread risk in VAR.
The Committee is consulting on whether, under a future framework, there should
- continue to be a separate model for default and migration risk in the trading book,
- or a single model.
Clearly consistency between the treatment of the same risk in the banking and trading books is to be desired.
Areas outside the scope of these proposals
- Interest rate risk in the banking book. The Committee intends to consider the timing and scope of further work in this area later in 2012.
- Capital for CVA risk.
IRRBB is something the FSA believes should be capitalized, but where it faces strong opposition from other supervisors. The industry is unhappy with the CVA proposals in Basel 3, and in particular with their poor treatment of market risk hedges to the CVA, but the Committee is not proposing to address this in the FRTB.
Overall, I find these proposals fairly sensible, if lacking in detail in a number of key areas. The thinking on liquidity, in particular, is valuable and welcome. The trading book/banking book boundary, while a flaw in the current regime, is in practice not a major vulnerability, and so it is good that the Committee is not suggesting anything too radical here. Expected shortfall is clearly a more stable risk measure than VAR, so again moving to this approach represents a sensible incremental improvement. A lot remains to be done though, especially developing the standard rules proposals.