Titel hier presented by John Doe Date here

Titel hier presented by John Doe Date here ASSOCIATION SOUTH AFRICAN INSURANCE

Trade Credit and Surety Non-Life Underwriting Risk (NLUR) parameters for SA QIS 3 13 June 2013 2

WG Structure ICISA S 2 Expert Group Non Life Underwriting Risk Fire Natural Catastrophe Man-Made Motor Liability Credit and surety Marine Premium Risk Reserve Risk NLUWR Workbook Aviation Terrorism SAIA TCS Forum

SCR NLUR overview (TCS) (V prem + V res) * ρ (σ) where ρ(σ)approximates 99. 5 th percentile = approx 3 * σ σ = sqrt((σ prem * V prem)^2 + (2 * 0. 5 * σ prem * σ res * Vprem * Vres) + (σ res V res)^2) / (V prem + V res) LGD % * max (sum 2 largest individual exposures, sum 2 largest group exposures) V prem (gross) * LR Recession Adjust for reinsurance SQRT (SCR Cat indiv ^2 + SCR Cat Recessionary^2) SQRT (SCR premres ^2 + SCR Cat ^2 + 2 * 0. 25 *SCR premres * SCR Cat) Ignoring lapse and counter-party default adjustment 4

History and current developments Compone S 2 QIS 5 nt = SA QIS 1 (June 2011) SA QIS 2 (July 2012) Subsequent S 2 developments (per ICISA) Premium and reserve risk σ prem = 21. 5% σ res = 19% σ prem = 11. 7% σ res = 19% No change (from EIOPA June 2010 calibration, adjusted) (from EIOPA Dec 2011 calibration, excludes catastrophe) 3 largest defaults LGD = 10% 2 largest defaults LGD = 10% PML 14%, recourse rate 28% LGD May 2012 study draft results show lower parameters V prem * 75% (adjust for RI) Also tested 25%, 50%, 75%, 100%, 150%, 200% EIOPA 100% V prem Consensus that dampener approach inappropriate, insufficient justification. (based on EIOPA developments at the time) ICISA <= 75% Underlying formula too complex, open to manipulation EIOPA / ICISA Impact study Oct 2011 showed average firm can only meet SCR NLUR at 75% Cat Individual default (LGD from ICISA PML study 2005 -7) Cat recessionary Complex scenarios with dampener 100 -200% of Vprem (EIOPA) * EIOPA = EIOPA Solvency 2 Cat Task Force Motivation not yet provided by EIOPA Still under discussion 5

Reasonability? CAT Recessionary V prem * 75% • How do you separate out cat frequency event? (Same issue applies to other classes eg Prop Commercial and crop) • CAT Recessionary charge must be considered together with σ prem and σ res! • Else will double count! (ICISA would prefer combined charge) • Look at 1 in 200 year loss ratio implied by proposed charges and then compare to industry experience So the key question is: What is the 1 in 200 year Loss ratio? 6

CAT Recessionary What is the 1 in 200 year Loss ratio? Implied by SA QIS 2 parameters – using ICISA spreadsheet Premium and reserve risk as per SA QIS 2, SCR Cat expressed as fraction of recessiona premium. ry V res / V Prem Sigma lob Assumed Total Break even Implied 1 in SCR NL loss ratio 200 loss ratio SCR pr 25% 11. 7% 43% 75% 95% 70% 165% 50% 12. 3% 54% 75% 103% 70% 173% 75% 12. 9% 66% 75% 112% 70% 182% 100% 13. 4% 80% 75% 122% 70% 192% 125% 13. 9% 93% 75% 134% 70% 204% 150% 14. 3% 107% 75% 145% 70% 215% 7

CAT Recessionary What is the 1 in 200 year Loss ratio? - ICISA study Feb 2012 250% Financial crisis loss ratios versus SA QIS 2 parameters 200% 1 in 200 high (215%) Loss ratio 150% 1 in 200 low (165%) 100% 50% 0% 2007 (ave 42%) 2008 (ave 82%) 2009 (ave 86%) 8

CAT Recessionary What is the 1 in 200 year Loss ratio? - ICISA • ICISA study Feb 2012 – Financial crisis 2008/9 was the worst crisis in the last 75 years (at least a 1 in 75 year event) – Average loss ratios doubled from 40% to 80%, – But recovered very quickly by 2010 (faster than overall economy) – Ability to manage credit limits by reducing or cancelling exposures at short notice (“dynamic exposure management”) – SA has further flexibility because policy terms also allow (trade credit) premium rates to be changed with 30 days notice (European terms generally annual) 9

CAT Recessionary What is the 1 in 200 year Loss ratio? – SA industry • • Distribution fitted to loss ratios past 20 years Internal models Management views Need to take cogniscence of ability to manage down exposures and increase premiums 1 in 200 loss ratio? Considered Trade Credit and Surety separately 10

Cat recessionary What is the 1 in 200 year Loss ratio? – SA industry Trade Credit Surety Combined (weighted average) 1 in 200 LR (lower) 144% 165% 151% 1 in 200 LR (upper bound) 201% 220% 207% Corresponds to cat recessionary charge* (using ICISA calculation tool) 55% 75% 60% Option 1: 55% V prem credit + 75% V prem surety • Better reflection of underlying risk • Suggest do not split premium and reserve risk at this stage, as this reduces credibility of data pool, which is already small. Maybe consider later • Could modify to allow for diversification / correlation Option 2 60% V prem • Simpler 11

Cat recessionary What is the 1 in 200 year Loss ratio? – SA industry Option 3 Shock should be applied to actual loss ratio, not break-even loss ratio (similar to mortality shock) May be tested as part of SA QIS 3 Eg 75% * Actual loss ratio / assumed break-even loss ratio Use at least 10 years loss ratio data (due to TCS underwriting cycle) • Better reflection of underlying risk • But may be distorted by inclusion of IBNR 12

Next steps for NLUR • Premium and reserve risk calibration – Allow for cat? Outliers? Credibility? • Man-made catastrophe – Discussion document has been drafted by NLUR, incorporating SAIA TCS research, update for new feedback • Impact studies – Economic impact – Consider equivalent banking products (market penetration and solvency) 13

References • FSB SA QIS 1 and 2 technical specifications • FSB NLUR Data Request 2012 • ICISA CAT Scenarios for Credit and Suretyship May 2011 (draft) • ICISA Impact Study SCR non-life Nov 2011 • ICISA NLUR for Trade Credit Insurance • ICISA LGD study May 2012 (draft) • Draft Discussion Document - SAM man-made catastrophe sub-working group Credit and Surety module May 2013 • EIOPA Calibration of Premium and reserve risk factors in 14 Standard Formula of S 2, 12 Dec 2011

Titel hier presented by John Doe Date here Thank you!

Additional slides 16

Trade credit vs surety/bond? Trade Credit Surety Insure against buyer default Insure against non-performance Underlying risk exposure is general economy Exposed mainly to construction industry Average loss ratios higher, but less volatile Average loss ratios lower but more volatile (see ICISA slides) Shorter-term guarantees and premiums Longer-term guarantees and premiums 17

Claims ratios: Trade credit vs surety