Demystifying copulas: An interactive lesson

Read the README file!

Summary of activity: An EXCEL-based exploration, creating your first copula in a step-by-step exercise. Then, using this to understand how correlation affects joint risk in multi-hazard / compound risk scenarios through simulation.

 Why is the training important: In short, because diversification is the basis of all (re)insurance, and correlation (a.k.a. dependency) destroys diversification. So, it’s fundamental.  If we get dependency wrong, we mis-estimate risk, mis-price risk etc ….  And, most people feel that copulas are scary, yet they really matter for tail-end risk (i.e. rare yet severe impacts).

Scenario: A firm has assumed complete independence between risks (e.g. lines of business). You’re sceptical. Before talking to them, you would like to:

• remind yourself of correlation coefficients (i.e. Pearson, Spearman)
• get your head around these things called ‘copulas’
• estimate how much of an effect pairwise dependency could have upon a firm’s risk (e.g. at 1-in-200 level)

 Level & Audience: Intermediate level. No equations! For those with of some experience the field of catastrophe risk management, or PhD students studying compound or multi-hazard risk. It can also serve as a reminder to those experienced in modelling dependency.

Duration: ~2-3h of effort.

Take home messages:  Also see Briefest_Guide_1.3.pdf – A 2-pager summarising the key things to know!