My love-hate relationship with bow-tie analysis
Bow-Tie Analysis is often presented as a panacea for the analysis of risk. On the one hand, they are intuitive, provide a useful structure for the discussion of risk, control and mitigation and they provide a natural bridge between qualitative and quantitative assessment methodologies. At the same time, however, they are illogical, prohibitively restrictive in what they can model, and potentially value-eroding.
In this talk, I will present bow-tie analysis and the natural way in which it supports the quantitative analysis of event-based loss risks. I will discuss their categorical shortcomings and the modelling challenges they represent; and I will suggest an interpretation that resolves many of these challenges, while maintaining much of their apparent intuitive appeal.