Property cat capacity: cat bond, ILW or more traditional XL?
Caribbean Wind Re writes £400m of property catastrophe premium, mostly US wind and Caribbean storm. The current programme is £250m xs £50m of cat XL plus a £100m aggregate protection. The board wants to cut earnings volatility after two years of large losses. Three options are on the table: (a) a £150m xs £150m catastrophe bond at a 6.5% coupon, fully collateralised, with an industry-loss trigger; (b) an industry-loss warranty of £80m xs £120m, attaching at a 1-in-10 industry loss, priced at 40% of the bond's risk-adjusted cost; (c) buying £100m more of traditional cat XL at the renewal. The internal model gives a 12% SCR reduction for the bond, 7% for the ILW and 10% for the extra XL, before model uncertainty adjustments. The rating agency requires the company to hold capital to a 1-in-100 net aggregate exceedance probability. The CUO notes the ILW is 'cheap capital' and the CFO prefers the bond for accounting reasons.
1. Evaluate the three options against risk transfer, basis risk, capital, earnings volatility and liquidity.
Read the worked answer
Cat bond (a). £150m xs £150m, fully collateralised, industry-loss trigger. Strengths: no counterparty credit risk; multi-year cover; likely to get rating-agency credit; 12% model SCR reduction. Weaknesses: basis risk because the trigger is industry loss, not the company's loss; if the company's loss is idiosyncratic (e.g., a single large account), the bond may not pay. Cost: 6.5% coupon on £150m = £9.75m per year plus issuance costs. Accounting: likely a reinsurance asset, which the CFO prefers, but the economics should drive the decision.
ILW (b). £80m xs £120m, attaches at a 1-in-10 industry loss. Strengths: cheap at 40% of the bond's risk-adjusted cost; can fill a specific layer. Weaknesses: high basis risk; the attachment at 1-in-10 may be too high to protect against a 1-in-50 company event; counterparty risk unless collateralised; may not reduce the rating agency's 1-in-100 net PML. Model SCR reduction of 7% is lower and more uncertain.
Extra XL (c). £100m more traditional cat XL. Strengths: indemnity trigger, no basis risk, 10% SCR reduction, straightforward. Weaknesses: priced at market, may be expensive; reinstatements and limited capacity.
Comparison. On risk transfer, (c) > (a) > (b). On cost per unit of capital relief, (b) looks cheapest, but the relief is illusory if the trigger does not respond. On earnings volatility, (a) and (c) are more reliable; (b) may not pay in the year of a loss. On liquidity, (a) is fully collateralised; (b) may have counterparty risk. On accounting, (a) may be treated as reinsurance; (b) may be a derivative, which could create P&L volatility.
Interaction with the aggregate cover. The £100m aggregate protection already responds to frequency. Adding (c) reduces severity; adding (a) protects the tail; adding (b) may overlap with the aggregate and create basis-risk mismatches.
2. How should model uncertainty and the rating agency requirement affect the comparison?
Read the worked answer
Model uncertainty. The internal model's 12%/7%/10% SCR reductions are point estimates. The ILW's 7% assumes the industry-loss trigger correlates with the company's loss. I would stress this: run the model with a lower correlation, a different industry-loss curve, and an idiosyncratic large-loss scenario. If the ILW fails to pay in a 1-in-50 company event, its capital credit should be reduced or disallowed for the rating agency.
Rating agency. The agency requires capital to a 1-in-100 net aggregate exceedance probability using its own model. It will not give full credit for an ILW if the attachment is above its 1-in-100 loss or if the trigger is not aligned. The cat bond may get more credit because it is collateralised and has a defined trigger, but the agency will still apply basis-risk haircuts. Extra XL gets full credit because it is indemnity.
Calibration. I would ask the broker for the rating agency's view of each option, not just the internal model. I would also test the 1-in-100 net PML under each option using the agency's cat model, and compare the marginal cost per unit of PML reduction, not per unit of SCR.
Parameter risk. The bond coupon and ILW price are market observations; the loss curves are uncertain. Add a margin for model uncertainty in the board paper, and show a range of SCR reductions (e.g., bond 8–12%, ILW 3–7%, XL 8–10%). Do not present a single number as if it were certain.
3. What is your recommendation and what conditions would you attach?
Read the worked answer
Recommendation. Buy the cat bond and a smaller amount of extra XL; do not rely on the ILW as core protection. Specifically: (a) £150m xs £150m bond; (c) £50m xs £250m extra XL; and keep the ILW under review only if it is collateralised and the trigger is at or below the 1-in-20 industry loss. This gives tail protection, indemnity protection for frequency, and avoids overpaying for basis risk.
Conditions.
- Legal review of the bond trigger and the ILW terms.
- Confirm the accounting treatment with the auditor; do not choose the bond merely because it 'looks smoother'.
- Stress the 1-in-100 net PML under the rating agency's model.
- Set a board risk appetite for basis risk and counterparty risk.
- Revisit at renewal if the ILW price falls or the trigger attachment lowers.
What I would tell the CUO. The ILW is cheap, but cheap capacity that does not pay when you need it is not capital relief. The bond is expensive but reliable and collateralised. The extra XL is the most efficient for earnings volatility per unit of basis risk. The final programme should be judged on net risk, not on the lowest premium.