Sessional Research Event: Good Practice in Setting Inputs for Operational Risk Models

The recent introduction of Solvency II has increased the importance of operational risk assessment and modelling for insurers. Banks already have to allow for operational risk in Pillar I capital under Basel II, but PPI misselling, fines for fixing LIBOR and other large losses have highlighted the need for banks to properly model and understand their operational risk exposures. Meanwhile operational risk poses an existential threat to asset managers.

A key challenge in modelling operational risk is data which is often sparse and/or irrelevant. This paper assesses the types of data available to model operational risk and the limitations of each source. Having regard to regulatory requirements, it puts forward good practice for the collection and use of loss data, scenario analysis and other data in operational risk modelling, as well as for setting correlation assumptions and other inputs to operational risk models.