A Tale of Two FRTB’s
In January 2016 the Basel Committee published the final rule of the Fundamental Review of Trading Book (FRTB), which represented the revised standards for minimum capital requirements for market risk.
In July 2015, BIS proposed the FRTB-CVA framework. Together with the Default CCR Capital, introduced as a part of Basel II in June 2006, CVA Risk Capital reflected Counterparty Credit Risk capital charges and has become an important part of analysing trade profitability.
Market Risk FRTB
Final major piece in Basel 3 puzzle
FRTB is intended to address the undercapitalisation of trading book exposures witnessed during the financial crisis. While the basic goals and ideas of FRTB are simple, it differs materially from the existing Market Risk regulations. It is therefore likely that the new rules will substantially change both the operating and business models of a large number of industry players.
The final release of FRTB introduces various changes to Basel 2.5 Market Risk capital rules, both qualitative and quantitative. For example, the definition of banking and trading books are more prescriptive, with tighter restrictions on trading/banking book reclassifications designed to reduce regulatory arbitrage. The rules governing the separation of trading and banking books are now more robust.
Internal risk transfers are also restricted as internal trades are only recognised if they are hedged with an external party (with the exception of interest rates risk). FRTB also includes more stringent and granular trading desk level Internal Model Approach (IMA) approvals.
Standardised Approach (SA)
Every bank, regardless of its IMA accreditation status, must also use SA to calculate capital. SA is not only applicable to banks with smaller and less sophisticated trading operations, it is now a viable fall-back method for IMA, allowing a more granular IMA accreditation than previous regulation.
Internal Model Approach
With approval from the banks’ supervisory authority, institutions can use the IMA to meet market risk capital requirements. However, the IMA approach cannot be applied to all products, such as securitisation and correlation-trading portfolios, where market risk capital must be calculated using only SA charges.
IMA consists of three different components:
Expected Shortfall Component
Whilst this is a highly visible change we do not believe it to be the most important. What we consider more significant is that the resulting capital charge is now calculated across a potentially very high number of scenarios, as opposed to the current regulation which only requires VaR and SVaR scenarios.
Default Charge and Non-Modellable Risk
This is an incremental charge intended to capture losses that stem from an obligor defaulting – similar to a Jump to Default charge. In this instance a VaR model with 99.9% confidence level is required. Capital charge for non-modellable risk factors (NMRF) is another IMA component, which by some banks is considered a key challenge area within the FRTB. NMRF are factors that affect pricing, but cannot be included in the extended shortfall calculation.
One of the new features in FRTB is the more granular approach to IMA approval. Historically, approval has been at the institution level, whereas now individual trading desks are subject to approval. As part of the switch-over process, institutions are required to submit details of the desks they intend to request IMA approval for. Any desks not submitted cannot migrate to IMA for a period of twelve months.
While BCBS aimed to have final regulations capital neutral, the latest Quantitative Impact Study indicated a median increase in capital of 18% and the weighted average of 75%. Expected Shortfall of IMA generally decreased compared to VaR measure, whilst both Default Risk Charge and Non-Modellable Risk Factors significantly contributed to the increase. The standardised charge increased across the board, and was two to three times more than the internal Model charge.
Despite the lengthy consultation process for the revised market risk framework there are still areas of ambiguity. For example, with Residual Risk Add On one would think that there could scarcely be something less controversial than multiplying a notional by a constant.
A concern for market participants is that regional regulators may adopt different approaches to resolving ambiguities, which would consequently create a maze of complex rules for institutions operating in multiple regulations.
The FRTB-CVA framework was proposed as a replacement for the current CVA Risk Capital calculations. To better understand its context, it is worthwhile reviewing the history of both Default and CVA capital charges.
Basel II and Default CCR (Counterparty Credit Risk) Capital Charge
Basel II requires banks to set aside capital to cover losses arising from counterparty defaults. These cost provisions, defined as RWA’s (Risk Weighted Assets), rely on a notion of loan equivalent Exposure at Default (EAD).
The most advanced methodology for calculating EAD is the IMM approach as it authorises banks to use their own internal models.
Basel III and CVA (Credit Valuation Adjustment) Risk Capital Charge
During the financial crisis of 2007-2008 many losses incurred by banks were caused by CVA moves. During the crisis, volatility of underlying market factors drastically increased as did credit spreads of counterparties. Both of these effects led to a significant increase in CVA, which is accounted for as a loss to the bank.
As a response, Basel III (BIS, December 2010, finalised June 2011) introduced a new capital charge aimed at mitigating CVA volatility. This charge can be calculated either using Standardised or Advanced methodology.
CVA Capital Charge – Standardised Formula
This is a simplified calculation for the CVA capital charge based on EAD’s and effective maturities M’s. In the absence of credit CVA hedges (which were the only hedges permitted) this formula is:
CVA Capital Charge – Advanced Formula
This formula, also known as CVA VaR, applies to banks with IMM approval for RWA and Specific Interest Rate Risk VaR model approval for bonds. For these banks the CVA capital charge is calculated as a triple sum of 10-day 99% CVA VaR’s for current and stressed periods
In these cases VaR’s are usually calculated historically but also can be done using the Monte Carlo model. Stressed period should be chosen independently for exposure and for credit spread. For exposure it should be based on three years of historical stress data across the whole portfolio. For credit spread it should be based on one year of historical stress data, as part of the three years of historical stress for exposure.
Total CCR Capital Charge
To calculate total CCR capital charges, Basel III requires banks to sum up Basel II RWA capital charges and Basel III CVA risk capital charges. There are currently four possible variants for total CCR capital charge, depending on the level of bank approval. With the introduction of SA-CCR on January 1 2017 the number of variants will reduce to three.
The new CVA framework will further adjust these variants, however as the new framework is work in progress and subject to significant change – such as the recent removal of IMA – it is too early to finalise total capital charge under new CVA framework.
New CVA Frameworks
The Consultative Document ‘Review of the Credit Valuation Adjustment Risk Framework’, published by BIS in July 2015, proposed replacing current Standardised and Advanced Approaches for calculating CVA capital charges with new methodologies. These new methodologies are more aligned with the Basel FRTB framework and accounting practices for evaluating CVA. It’s aim is threefold:
i. Capturing all CVA risks along with enhanced recognition of CVA hedges
ii. Alignment with industry practices for accounting purposes
iii. Alignment with proposed revisions to the market risk framework
New Basic Approach BA-CVA
A simplified formula resembling current Standardised CVA methodology. In the absence of credit hedges (and this approach does not allow market hedges) the formula is:
New SA-CVA approach
To use SA-CVA the following requirements must be fulfilled:
- The calculation of CVA sensitivities for given risk factors must comply with general principles for the calculation of CVA
- A methodology for approximating the credit spreads of illiquid counterparties is applied
- A dedicated CVA risk management function (and control unit) exists
Qualifying banks need to follow general principles to calculate regulatory CVA in line with the FRTB-CVA framework.
Comparison of SA-CVA with SA-TB
Since IMA-CVA is eliminated from FRTB-CVA framework, the only comparison between the two FRTB frameworks is with the SA-CVA and SA-TB methodologies. BIS acknowledges that though SA-CVA is an adaptation of the sensitivity-based approach for market risk to the CVA book, there are several important variances between SA-CVA and SA-TB, e.g. since counterparty default risk is already included in the CCR capital charge, the SA-CVA does not account for default risk. Also, recognising the fact that calculating CVA sensitivities is computationally very expensive, regulators reduced granularity of supervisory market factors in most cases and excluded gamma risk from SA-CVA. To compensate for the elimination or reduction of sensitivities, risk aggregation for SA-CVA is more conservative than it is for SA-TB. Recognising that CVA is almost linear to the counterparty credit spread and that this type of sensitivity is relatively straightforward to calculate, an extra asset class of counterparty credit spreads is created which retains the full granularity of deltas, but no vega.
FRTB is likely to have a substantial influence in the way firms are organised, and their approach to measuring and reporting risk. There will also be an overall business and operational impact. Banks need to decide whether the costs associated with operational and IT change is justified.
Ambiguity remains around methodologies, assumptions, definitions etc. which we expect to be resolved once regulation is enshrined in law – hopefully consistently across the multiple jurisdictions. Even with these ambiguities resolved, implementing FRTB is still likely to be challenging.
PnL tests are critical for any bank that wishes to use IMA, however, only a few sophisticated institutions have in place the necessary framework to successfully run PnL tests.
NMRF can pose capital and operational costs to the bank. At the same time, NMRF may encourage banks or third parties to make the markets more transparent (thus reducing the number of NMRFs). Conversely a lack of transparency imposes high capital costs on counterparties, causing the market to shrink.
Similar issues are relevant to the CVA component of FRTB. Tests carried out by Quantifi demonstrate the increase in CVA risk capital charge under the new basic approach (BA-SVA) compared to the current Standardised approach. The impact of the new CVA risk regulatory framework on the calculation methods and the bank’s current state infrastructure could be a turning point for many medium-sized institutions.
Banks seeking to adopt the SA-CVA method will be interested in fast and accurate CVA sensitivity calculations. Sophisticated technology providers like Quantifi, are able to provide various approaches for increasing efficiency of CVA calculations like streaming, cash-flow optimization, and analyzing dependency graphs.
Banks that have a culture of following a disciplined approach and make good choices in managing their technology will likely be at a distinct advantage and better positioned in the post FRTB world in terms of their operating costs, ability to manage risk, and their ability to understand profitability across their organisation.