I recently spent an enjoyable few days leading seminars on the Fundamental Review of the Trading Book (FRTB) with a financial supervisory authority. It was a great opportunity to share thoughts on the subject and consider its impact from both a banking and a supervisory perspective.
What struck me is that the FRTB has two initial goals: first, to ensure that capital better reflects banks’ trading risk, and second, to create consistency so capital measure can be compared between banks. In this article, I’ll look at how the FRTB has measured up to the first objective.
The FRTB Standardised Approach Explained
One of the FRTB’s great achievements is its new Standardised Approach (SA) to making capital reflect risk. It has several clear advantages over previous regulatory regimes.
The SA calculates market risk capital for credit in much the same way as any other risk class. It’s a long-overdue recognition that credit, like interest rates or equity, is now a traded risk class and should be treated as such. The SA also now measures risk capital for both delta and vega. This marks a sensible improvement in regulation, clearly tying together capital and standard risk management practices for derivative products.
It’s also clear that the new curvature risk capital captures non-linear risk from derivative products. Similarly, the default risk charge captures idiosyncratic risk from the default of an underlying. These two capital charges should be considered an achievement in making capital reflect trading risk.
However, there are some effects that these charges miss. For example, cross risk from changes in underlying and volatility is missing from the curvature charge and the SA in general. Time will tell whether this oversight will be material in the efforts to align risk and capital.
The ‘Simplified’ Standardised Approach Could Reverse Regulatory Progress
Given the significant achievement of the SA in ensuring that capital reflects risk, it’s a little disappointing that the FRTB introduced an option for banks to use a ‘Simplified’ Standardised Approach. This simplified method has none of the advantages of the full SA and doesn’t offer anywhere near the same levels of regulatory rigour.
For some people, the last-minute addition of the simplified SA to the FRTB is a retrograde step. I recognise that the full SA places an additional burden of operational complexity and cost on some of the smaller banks, and some concession for those with very small trading books might have been necessary. However, offering a pared-back approach to the FRTB comes with a considerable trade-off; it means capital isn’t comparable between banks as not all firms will implement the full SA. Further, for banks that adopt the simplified SA, capital does not fully reflect risk.
The inconsistency enabled by the simplified SA undermines the two initial goals of the FRTB. I know financial supervisory authorities have mixed views on the introduction of the simplified SA - I can only hope that they will restrict its use to banks with insignificant trading activities.
The Internal Models Approach
There have been significant improvements in making the Internal Models Approach (IMA) calculation of market risk capital better reflect trading risk. It’s universally acknowledged that the move from VaR to Expected Shortfall (ES) more accurately reflects the tail risk in trading P&L. Incorporating different liquidity horizons into the ES calculation is another major step forward and recognises a key interaction between market liquidity and market risk.
From a regulators point of view, the continued use of a stressed period - although in a more sensible way than Basel 2.5 - and the control of correlation between risk classes may be necessary complications. However, for all the additional complexity to the ES calculation, the question remains; does FRTB ES make capital better reflect trading risk?
I believe a fundamental problem persists. FRTB uses historical data - either directly, or to calibrate a Monte Carlo method - to try and predict the behaviour of future returns. Even retail financial products frequently warn that ‘past returns are no guarantee of future behaviour,’ yet this is the entire basis of the IMA.
However, in my view, the new P&L attribution testing and the heavy level of backtesting should more than compensate for this reliance on historical data. After the 2019 changes to the testing regime, these tests now provide a robust framework for assessing the real-world effectiveness of an internal model. And they’re not only robust, they have teeth; for moderate levels of testing failures, they can increase capital requirements. For more extreme levels of failure, they can remove a trading desk from internal model capital treatment.
I hope that supervisory authorities will gain confidence in this testing regime and that this, in turn, will lead to a more relaxed approach to model specification and a more sympathetic treatment of Non-Modellable Risk Factors.
There’s no doubt that the FRTB’s achievement of making capital better reflect trading risk has come at a considerable cost. Notably, the dramatic increase in computing power needed to calculate ES and the increased sophistication required to implement the SA. However, despite these drawbacks, the FRTB is a positive step towards improving trading book capital calculations. Of course, some areas of the new regime won’t work as well as expected, or will require alteration when field-tested in coming years. Hopefully, the Basel Committee will be quick to respond to market feedback as and when needed.
If you want to find out more about FRTB and how regulators and banks have been responding to this challenge, I provide in-house courses given by London Financial Studies. You can find more details about our courses here.