Fundamental Review of the Trading Book: Beyond the IMA

The 2008-09 financial crisis brutally exposed lacunae in the assessment of trading book capital requirements. It triggered a fundamental need to reconsider guidelines governing the overall market risk framework.

The Basel Committee for Banking Supervision (BCBS), through a series of consultative papers released between 2012 and 2014, set out a Fundamental Review of the Trading Book (FRTB) regime for trading businesses and their support functions.

The proposal seeks to address broad areas of the banking trading book division, standardised approach methodology and internal models approach methodology. While a wide range of issues are addressed regarding the internal models approach (IMA), the key idea is to avoid the ‘black-box’ approach of banks when adopting the IMA while also strengthening the market risk measurement methodology.

The proposal can be considered as a welcome change from a prudential perspective since the market risk regulatory requirements have remained largely unchanged for a long time, other than certain additions in Basel 2.5 to revise Basel II norms. While the key proposals touched upon a wide range of topics in market risk measurement methodologies, a key area of interest could be that of proposed changes to the IMA.

There are certain qualitative and quantitative criteria prescribed for banks looking to adopt IMA, to be adhered to on an ongoing basis. The supervisory authority should ensure that the bank satisfies these requirements to be deemed eligible for adopting IMA.

Qualitative Criteria:

It is noteworthy that most of the qualitative criteria mentioned in the proposal fall under the auspices of process and governance mechanisms. With requirements ranging from the setting up of an independent risk control unit capable of strategising, designing and implementing the bank’s overall risk management system to laying down policies for implementation of stress testing and backtesting programmes, enhanced involvement of senior management and tighter model approval and audit processes, the proposals would mean enhanced prudence from a regulatory perspective.

While that is an upside, it carries with it cost and implementation challenges for banks if they are to realise all of the recommendations, both in letter and spirit. Banks would need to identify the required policies and devise a mechanism to implement and achieve firm-wide adherence within the given time limits. It is also imperative that the proposals are implemented consistently across the organisation.

It is also critical to note that certain requirements – such as production of daily risk reports, implementing back testing and penins explained (PnL) attributions based on daily as well as hypothetical changes – demand a significant investment in technology architecture and change management systems. Additionally, there are several further requirements – such as risk measure calculation on full set of positions, building robust stress testing capabilities and model documentation systems – that banks needs to revaluate as well from this perspective. These investments, when viewed from a long-term perspective, clearly give the bank better insight and tighter control over its exposure and risk appetite.

Figure 1: IMA – Qualitative criteria

Figure 1: IMA – Qualitative criteria

Click image for enlarged view

Quantitative Criteria:

FRTB replaced the long-standing risk measure of Value at Risk (VaR) with Expected Shortfall (ES). ES looks to plug the shortfall of VaR (mainly capture of tail risk and the principle of sub-additivity), which renders VaR as an incoherent risk measure. ES needs to be calculated on a daily basis with a 97.5th percentile, one-tailed confidence interval.

The committee, in its second consultative paper, prescribed modelling of risk factor shocks directly over the actual length of the given liquidity horizon. This would mean, for longer liquidity horizons, that banks could not scale up from shorter horizons and would need to gather the data and compute PnL for all the days in the given liquidity horizon. However, based on industry concerns around the area of data availability, potential lack of comparability across firms and implementation challenges in terms of sample sizing becoming very small for illiquid risk factors, the committee conceded a revision in its third consultative paper.

According to this, the revised model prescribed ES model for a base horizon of 10 days for all the risk factors and a pool of incremental ES (scaled from “base” horizon) for subsets of risk factors with longer liquidity horizons. This nested approach helps preserve the short horizon correlations across risk factors. Clearly, banks cannot use the scaling methodology for liquidity horizons of less than 10 days. For those banks which currently use scaling from shorter horizon (like 1 day), a 10-day full revaluation could pose a significant challenge.

ES should be calibrated to a period of severe 12-month stress available over the observation horizon ranging back to at least 2005. While it is unlikely that all the risk factors used in ES model are available for a full history, the committee proposed an indirect approach. According to this, the calibration is carried out using a reduced set of risk factors (under supervisory approval) with an observable history of 10 years and would explain a minimum of 75% of the variation in the full ES model. This value would then be scaled up for the historical period based on a ratio of full risk factors to reduced risk factors of current period (recent 12 months) ES.

A critical point to note here is the significant impact that the multiple “runs” for computing ES (as required by the nested approach) would put on banks’ infrastructure. This effectively amounts to a multiplicative effect of calculating ES on all the five asset classes, five liquidity horizons with both reduced and full set of risk factors and for current as well as stressed periods for every desk. It is also proposed that the level of cross-asset class diversification will be constrained by the supervisory scheme.

Figure 2: IMA – Quantitative criteria

Figure 2: IMA – Quantitative criteria

Click image for enlarged view


While the finer points are still being discussed with the industry and the final recommendations not yet imminent, it is becoming increasingly clear that the road ahead for banks is a challenging one. There are fundamental policy and governance issues that the banks need to iron out – and prune – while also balancing other advanced challenges in terms of scaling up and readying the infrastructure, data models and risk reporting mechanisms to support and achieve compliance within the given timelines.

This is a challenge that banks need to solve from a multidimensional perspective!


Related reading