Risk taxonomy

During a winter storm, a ransomware attack locks Example Insurance SE's claims system for a week. Thousands of storm claims wait. The regulator later fines the insurer for treating customers unfairly, and to clear the backlog quickly it sells bonds at a poor price.

The insurer reports in euros and holds €2,000m of US dollar corporate bonds with a modified duration of 6. A scenario has US Treasury yields up 0.5 percentage points, the bonds' credit spreads up 0.4 points, and the dollar down 5% against the euro.

A manufacturer has drawn £30m of a £50m credit line. The bank expects 60% of any unused limit to be drawn before a default. The borrower's one-year PD is 2.5% and the LGD 45%.

Swap A, with a highly rated manufacturer, is worth £40m to the bank and is not collateralised (PD 0.2%). Swap B, with a hedge fund, is worth £60m, and the fund has posted £55m of collateral (PD 3%). LGD is 60% on both.

Over a month, 25% of Example Bank plc's £330bn of customer deposits leave. It pays first from its £50bn of cash, then sells securities. Because other banks are selling too, it can only sell at an 8% discount to market value.

Criminals send an email that appears to come from the chief executive, asking a clerk in Example Bank plc's treasury to pay £2m urgently to a new supplier. The clerk pays. No employee was in on it.

A storm costs Example Insurance SE €3,200m before reinsurance. Its catastrophe cover pays losses above €500m, up to €1,500m. In the same quarter, its €10,000m of non-life reserves prove 3% too low. Own funds were €9,300m and the SCR is €5,000m.

A steel maker borrows €1,200m and earns EBITDA of €300m. It emits 1.5 million tonnes of CO₂ a year. A carbon price of €120 a tonne is coming, and it expects to pass only 25% of the cost to customers. The bank's PD model, built on 2010–2019 data, uses last year's accounts only, so it rates the borrower exactly as before.