1. The Economics of Terrorism Insurance Neil Doherty The Future of Terror Risk Insurance University of Southern California June 2005

2. Performance of Terrorism Insurance versus Performance of Terrorism Risk Management Policy question has been framed as one on terrorism insurance Policy question has been framed as one on terrorism insurance Insurance is only one of number of strategies for managing terrorism risk Insurance is only one of number of strategies for managing terrorism risk “failure” of insurance market does not imply risk not being managed is inefficiently managed “failure” of insurance market does not imply risk not being managed is inefficiently managed Other risk management tools include Other risk management tools include Leverage Leverage Contingent financing Contingent financing Structured debt Structured debt etc etc

3. Terrorism insurance issues Is terrorism an “insurable” risk? Is terrorism an “insurable” risk? Is terrorism insurance market shaped by pre- existing imperfections? Is terrorism insurance market shaped by pre- existing imperfections? Does market clear? Does market clear? Would insurers be able to pay for mega loss? Would insurers be able to pay for mega loss? Would post-loss supply be constrained by capital shortage? Would post-loss supply be constrained by capital shortage?

4. Is terrorism an “insurable” risk? Measurement of risk Measurement of risk Some progress in modeling severity - but Some progress in modeling severity - but Difficulty of estimating frequency risk Difficulty of estimating frequency risk Difficulty of game theoretic modeling Difficulty of game theoretic modeling Federal government has access to sensitive risk information Federal government has access to sensitive risk information Moral hazard Moral hazard Hazard function of government policy Hazard function of government policy Foreign policy- counter-terrorism – homeland security Foreign policy- counter-terrorism – homeland security Does this suggest role of Federal govt. as risk bearer Does this suggest role of Federal govt. as risk bearer Interdependent security Interdependent security Can shift risk to others Can shift risk to others Incentive to under-invest – benefits of security externalized Incentive to under-invest – benefits of security externalized

5. Is terrorism insurance market shaped by pre- existing imperfections? Existing insurance market faces: Existing insurance market faces: Compulsory insurance – Compulsory insurance – workers comp workers comp Fire following some states Fire following some states Rate regulation Rate regulation Double taxation Double taxation Samaritan’s dilemma – expectation of ex post relief Samaritan’s dilemma – expectation of ex post relief Two broad choices for policy debate Two broad choices for policy debate Reduce existing market imperfections Reduce existing market imperfections Bolster insurance market to overcome imperfections Bolster insurance market to overcome imperfections

6. Does market clear? Before TRIA little coverage Before TRIA little coverage Coverage with TRIA Coverage with TRIA 54% AON sample buy some terrorism ins (TRIA/non certified/stand alone) 54% AON sample buy some terrorism ins (TRIA/non certified/stand alone) Varies across industry Varies across industry HIGH -Real estate /transportation/health care HIGH -Real estate /transportation/health care LOW – Pharma/chemicals/construction LOW – Pharma/chemicals/construction Varies across region Varies across region HIGH – south-west-northeast-central-southeast – LOW HIGH – south-west-northeast-central-southeast – LOW Terrorism premium varies Terrorism premium varies HIGH – real estate/basic materials/entertainment HIGH – real estate/basic materials/entertainment LOW – Tedchnology/Consumer goods LOW – Tedchnology/Consumer goods HIGH – northeast –south – central – west – southeast HIGH – northeast –south – central – west – southeast Regression (dependent variable Terrorism limit/insurable value Regression (dependent variable Terrorism limit/insurable value Demand elasticity -0.646 Demand elasticity -0.646 coefficient for property limit/insurable value = 0.879 coefficient for property limit/insurable value = 0.879 Generally rational market but with many choosing non insurance Generally rational market but with many choosing non insurance Incentive for captives – free reinsurance option – low deductible Incentive for captives – free reinsurance option – low deductible

7. Figure 9.2. Terrorism Limit / Property Limit by Insurable Value (478 AON accounts)

8. Figure 9.3. Predicted Probabilities of Purchasing Terrorism Coverage for Accounts at 25 th Percentile, Mean, and 75 th Percentiles of Insurable Value

9. Table 9.5 Regressions of Terrorism Limit / Insurable Value on Terrorism Premium Rate Relative to Property Rate and Control Variables (248 Aon accounts with any terrorism coverage and non-zero premiums) Without Region and Industry With Region and Industry Variable Coeff.p-value Coeff.p-value Constant-1.4990.043-1.2790.055 Terrorism rate / property rate -0.6460.000-0.6640.000 Property limit / insurable value 0.8790.0000.8600.000 Insurable value -0.0670.052-0.0540.078 Central0.3440.003

10. Does market clear? continued Generally rational market but with many choosing non insurance Generally rational market but with many choosing non insurance Many forms absorb loss through capital structure Many forms absorb loss through capital structure Some mitigate (e.g. chemical firms) Some mitigate (e.g. chemical firms) Incentive for captives – Incentive for captives – free reinsurance option – free reinsurance option – low deductible – captives usually have low TRIA premiums low deductible – captives usually have low TRIA premiums Relatively little reinsurance Relatively little reinsurance For non certified For non certified For TRIA deductible For TRIA deductible Difficult to forecast reinsurance in absence of no renewal Difficult to forecast reinsurance in absence of no renewal not obligated under compulsory insurance statutes not obligated under compulsory insurance statutes

11. Would post-loss supply be constrained by capital shortage? Major industry-wide shocks tend to be followed by Major industry-wide shocks tend to be followed by Shortage of capacity in ALL lines Shortage of capacity in ALL lines More severe in line affected (revision of loss expectancy) and lines which demand more capital (long tail) More severe in line affected (revision of loss expectancy) and lines which demand more capital (long tail) Demand surge in affected line Demand surge in affected line Reduced supply and increased demand cause premium spikes Reduced supply and increased demand cause premium spikes Over time new capital enters to exploit hard market Over time new capital enters to exploit hard market High leveraged firms worst hit with some insolvencies High leveraged firms worst hit with some insolvencies This is known as capacity constraint model This is known as capacity constraint model Evidence after Liability Crisis, Northridge, Andrew, 9/11, etc. Evidence after Liability Crisis, Northridge, Andrew, 9/11, etc.

12. Would post-loss supply be constrained by capital shortage? continued Capacity constraint shows markets to be self correcting following major shocks but self correction involves some disruption – hard markets Would market be self correcting (in time) following another major terrorism attack? Depends on climate of uncertainty Depends on climate of uncertainty Depends on whether insurers allowed to service hard market Depends on whether insurers allowed to service hard market