Risk modelling: Convergence needed, but some variances are legitimate

Abstract

The level of variance between banks’ internal RWA models has generated concern, although the sources and effects of such variance are not always understood. An extensive review of modelling practices by the IIF RWA Task Force in 2014 has confirmed that, while variances are indeed too large, some components of this reflect underlying differences between banks’ portfolios and risk management policies and practices, among other legitimate factors. In this sense, some variance can have a positive effect, as capturing the inherent differences between banks helps strengthen risk management, making banks more attuned to the real risks and incentivising them to improve their capabilities. Some harmonisation of modelling practices is necessary to help restore RWA credibility, necessitating an extensive process of sifting through which model factors represent an intrinsic difference between banks or jurisdictions and which reflect more common factors. This can help enable the continued improvement of models, and the vital role of risk-sensitive metrics at the core of banking.