Know these first
- L3.2.11: Monte Carlo VaR
- L3.8.04: Premium risk and the underwriting cycle
Math Atoms
By the end you will be able to
- 1Simulate dependent risks by Cholesky and by copula sampling, attaching any margins, and check the joint tails the draws produce.
- 2Measure the uncertainty of a historical VaR or ES with the bootstrap, and use block bootstraps when losses cluster.
- 3Aggregate RiskAtoms Insurance SE's non-life risks by simulation under Gaussian and t copulas, and read capital, diversification and expected shortfall beside the booked figures.
- 4Explain which choices (copula, ν, a hand-set correlation, the number of draws) move the result, and by how much.
Why it matters
RiskAtoms Insurance SE writes motor, property and liability business and carries catastrophe risk. Each risk's own 1-in-200 loss adds up to €4,222.0m. Simulated together with a Gaussian copula, the 1-in-200 total is €2,794.0m; with a t copula, same margins and same correlations, €3,059.0m. Choosing the copula moves the insurer's capital by €265.0m. This lesson builds those simulations, and the bootstrap that says how much to trust a figure estimated from history.
Module complete