Quantitative Impact Study 1 (QIS1) Summary Report for Belgium 21 March 2006 1
Quantitative Impact Study 1 (QIS1) Summary Report for Belgium INTRODUCTORY REMARKS...4 1. GENERAL OBSERVATIONS...4 1.1. Market share covered by QIS1...4 1.2. Quality and availability of data...5 1.3. Problems encountered by insurers in connection with QIS1...6 2. STRUCTURE OF QIS1...6 2.1. Scope of the study...6 2.3. Segmentation into risk groups...6 2.3. Method of aggregation...8 3. QUANTITATIVE IMPACT ON TECHNICAL PROVISIONS...8 3.1 Impact on Life insurance provisions...8 3.2 Impact on Non-Life insurance provisions...10 4. METHODOLOGICAL ASSESSMENT...13 4.1. Life insurance risk factors...13 4.1.1. Method applied to discounting...13 4.1.2. Method applied to financial guarantees, future bonuses & profit sharing...13 4.1.3. Method applied to surrender & lapse rates...14 4.1.4. Method applied to mortality and morbidity rates...14 4.1.5. Method applied to expenses...14 4.1.6. Risk margin estimates...15 4.2. Non-life insurance risk factors...15 2
4.2.1. Method applied to claim provisions...15 4.2.2. Method applied to premium provisions...16 4.2.3. Risk margin estimates...16 16 3
Introductory remarks CEIOPS has been asked to deliver advice to the EU Commission on the introduction of a new riskbased solvency regime, commonly known as Solvency II. For this purpose CEIOPS will conduct a series of quantitative impact studies (QIS) to support the proposal of a framework directive for Solvency II. In view of such quantitative support, CEIOPS has conducted a first QIS in December 2005 with a focus on testing the level of prudence in technical provisions under several hypotheses. This report summarises the main findings and conclusion to be drawn for future work of Solvency II, based on the QIS1 output received from participating insurers for the Belgian market. For a cross country comparison of QIS1 results for the EEA we refer to the CEIOPS summary report 1. The results are based on individual firm data as at end of 2004 financial year. It should be noted however, that at this early stage of the QIS process the precise figures obtained are less important than their general order of magnitude or the views expressed by firms about the practicability of performing such calculations. 1. General Observations 1.1. Market share covered by QIS1 There are 10 legal entities participating in the QIS1, of which 7 entities are reporting for their Life portfolio and 8 entities reporting for their Non-Life portfolio. Two composite insurers opted to exclude their Life activity for QIS1, as they were not in a position to stochastically model the cash flows of their Life portfolio within the tight timeframe available. These two insurers are therefore accounted for as pure P&C insurers in QIS1. The following table summarizes the characteristics of the participating firms in terms of activity reported (Life, Non-life) and their absolute size (Small, Medium, Large). Number of participating firms Number of firms by size 2 Market share 3 Totals for all participating firms Small Medium Large Total Total % Life insurance 2 1 4 7 59,8% Non-life insurance 3 4 1 8 51,1% Reinsurance 0 0 0 0 0,0% Total number of firms (entities) 3 3 4 10 Table 1: Number of participating firms 1 A public version of the QIS1- Summary report (CEIOPS- FS- 01/06) is available on the CEIOPS website (http://www.ceiops.org/media/files/publications/submissionstotheec/ceiops-fs-0106rev32006-03-17pa.pdf ) 2 For the Life activity the classification is based on the absolute size of gross technical provisions ( 'small <1,000 mio EUR, 1,000 mio EUR < medium < 9,999 mio EUR and large > 10,000 mio EUR). For the Non-life activity the classification is based on absolute size of gross written premium ( 'small <200 mio EUR, 200 mio EUR < medium < 999 mio EUR and large > 1,000 mio EUR) 3 The market share is the ratio of the statutory technical provisions of the participating (life/non-life/reinsurance) firm to the technical provisions of the whole (life/non-life/reinsurance) market. For composite firms only its life technical provisions should be allowed for in the life insurance market share and only its non-life technical provisions in the non-life insurance market share. 4
For entities belonging to the same insurance group (6 participating insurance groups) the data was reported on a legal entity basis. All data relates to the direct domestic insurance activity only. The market coverage in terms of number of participating companies by line of business is fairly limited (6 out of 28 composites, 3 out of 122 Non-life insurers, 1 out of 31 Life insurers). Nevertheless, the participating entities represent 59.8% of the market s statutory technical provisions for Life, and 51.1% of the statutory provisions for Non-Life. However, not all of the participants included 100% of their respective activity in QIS1. Reinsurance companies were excluded from the exercise, as these entities are not regulated by the CBFA. We consider the sample to be adequately representative for large sized firms and to a lesser extent for medium sized firms. Two of the 3 small firms included in the QIS are part of an insurance group, of which the parent company is also participating in the QIS, albeit in a different jurisdiction. The QIS is therefore believed to be not representative for highlighting characteristics of truly SME firms. 1.2. Quality and availability of data For the Life activity, insurance liability calculations were performed only on a gross of reinsurance basis. In addition, not a single firm provided an estimate of the effect of reinsurance default as reinsurance of the life portfolio is said to be immaterial. For the non-life activity, a large proportion of firms provided data on both an undiscounted and discounted basis, gross of reinsurance. For most firms the requested risk margin calculations (75% and 90% confidence levels) were provided on this same basis. Given the complex arrangement of multiple reinsurance treaties depending on the year of claim occurrence (excess of loss, quota share, stop loss) and the use of multi-year and multi-line stop loss treaties, it was difficult to model the cash flows relating to the reinsurer s part in technical provisions and even more difficult to allocate this part across the different risk groups. The following tables provide an overview of the availability of data under the different calculation bases requested in QIS1. The data is expressed as a proportion of firms being able to provide the data under each basis, separately for the Life and Non-life activities. Coverage - Summary of availability of data from firms Best estimate Percentage of firms providing following data 75th Percentile 90th Percentile Standard deviation 60th Percentile Company view Life insurance % % % % % % Total provisions, gross of reinsurance 100,0% 57,1% 71,4% 0,0% 28,6% 14,3% Total provisions, net of reinsurance 0,0% 0,0% 0,0% 0,0% 0,0% Time value of guarantees & options 28,6% Surrender value floor difference 14,3% Contribution of surrender risk 28,6% 42,9% Effect of reinsurance default 0,0% 0,0% 0,0% 0,0% 0,0% Diversification effect on whole portfolio, gross of reinsurance 0,0% 57,1% 71,4% 0,0% 0,0% Table 2: Availability of data for Life insurance 5
Coverage - Summary of availability of data from firms Non-life insurance Best estimate Percentage of firms providing following data 75th Percentile 90th Percentile Standard deviation 60th Percentile Company view Undiscounted total provisions, gross of reinsurance 75,0% 75,0% 75,0% 62,5% 37,5% 25,0% Discounted total provisions, gross of reinsurance 75,0% 75,0% 75,0% 50,0% 37,5% 25,0% Undiscounted total provisions, net of reinsurance 25,0% 12,5% 12,5% 12,5% 0,0% 12,5% Discounted total provisions, net of reinsurance 12,5% 12,5% 12,5% 12,5% 0,0% 12,5% Effect of reinsurance default on discounted provisions 0,0% 0,0% 0,0% 0,0% 0,0% Diversification effect on whole portfolio based on discounted provisions, gross of reinsurance 12,5% 12,5% 12,5% 12,5% 0,0% Table 3: Availability of data for Non-life insurance 1.3. Problems encountered by insurers in connection with QIS1 Time constraints limited the ability of some firms to produce the results in sufficient detail to allow a meaningful comparison across all firms. For the same reason, not all firms provided data which was optional (60% confidence level, company view of risk margin). 2. Structure of QIS1 2.1. Scope of the study The main focus of QIS1 is on a market-consistent valuation of insurance liabilities for each relevant homogenous risk group for both the Life and Non-Life portfolios. The expected present value of projected cash flows relating to insurance contracts should be used in determining the best estimate of liabilities. The modelling of such future cashflows should be based on sound actuarial assumptions and discounted using a risk-free duration dependent term structure which was provided by CEIOPS. By stochastically simulating the variation in cashflows (based on random variation in the underlying insurance risk factors) to determine an appropriate distribution, the best estimate should correspond to the mean of that distribution. To determine a given level of prudence in technical provisions a risk margin approach based on the 75% and 90% confidence intervals of the underlying distributions has been chosen. Alternative approaches to a prudent valuation of technical provisions (cost-of-capital approach, risk neutral value, etc.) have also been included in the scope of QIS1 for those firms that provided their own view on liability valuation. The mechanics of the QIS1 are meant to provide an assessment of a transition towards a marketconsistent value of technical provisions. The impact assessment is therefore based on a comparison of the best estimate of insurance liabilities as a ratio of statutory provisions (as of end 2004) and the best estimate + risk margin valuation as a ratio of statutory provisions. This information is provided for each broad risk group of the aggregated (across firms) Life portfolio, and for each group of Nonlife insurance class as described in Article 63 of the insurance accounting Directive 91/674/EEC. 2.3. Segmentation into risk groups In general, a segmentation of the Life portfolio was performed on the basis of products with same risk characteristics (sensitivity to mortality rates, savings/investment features, distribution channel). The following broadly defined homogenous risk groups were used: 1. Traditional Individual: term assurance policies and annuity contracts for personal lines 6
2. Traditional Group: term and annuity group contracts 3. Universal Life with profit sharing: with profit guaranteed rate investment policies 4. Life without profit sharing: non-profit investment contracts 5. Unit Linked: investment policies where the investment risk is borne by the policyholder Segmentation of life provisions Risk group Share of each risk group in statutory provisions included in QIS1 Traditional Individual Business 29,5% Traditional Group business 22,9% Universal Life business with profit sharing 26,4% Life business without profit sharing 0,4% Unit Linked business 20,7% Table 4: Segmentation of life provisions For the Non-life portfolio the suggested segmentation into groups of classes according to Directive 91/674/EEC was mostly adopted. Where firms provided data on a different basis (internal lines of business), the data was repartitioned to achieve a maximum of uniformity in the reported classes. The segmentation consisted of the following 11 groups of classes: 1. Accident & Health: workmen s compensation, personal accident 2. Motor, 3 rd party liability: 3. Motor, other classes: theft, damage 4. Marine, aviation & transport: damage, liability related to these activities 5. Fire & other property damage: fire, natural perils 6. 3 rd party liability: liability other than covered by classes 2 and 4 7. Credit & suretyship: (included in class 10) 8. Legal expense: coverage of legal expenses 9. Assistance: road, repatriation 10. Miscellaneous: compensation for pecuniary losses, credit & suretyship 11. Reinsurance: inward reinsurance Segmentation of non-life provisions Risk group Share of risk group in statutory provisions included in QIS1 Accident and health 9,1% Motor, third party liability 46,3% Motor, other classes 8,7% Marine, aviation and transport 0,1% Fire and other damage of property 15,0% Third-party liability 17,2% Credit and suretyship excluded Legal expenses 2,4% Assistance 0,03% Miscellaneous non-life insurance 1,1% Reinsurance Table 5: Segmentation of non-life provisions excluded 7
The Accident & Health business is generally excluded from the scope of QIS1. For the firms that excluded Accident & Health the modeling approaches for this line of business are substantially different from the other non-life businesses (includes life risk features related to longevity and mortality) and the results were viewed to be not stable enough to be included in the QIS. The credit & suretyship activity has been excluded from the study as this represents an insignificant line of business. Only one firm provided data on the inward reinsurance activity. For reasons of comparability across different firms this risk group was excluded from the QIS analysis. 2.3. Method of aggregation Based on the segmentation described above, the individual firm results were aggregated into the relevant risk groups on a line by line basis. Aggregation tables were produced corresponding to each cell of table 2, for each relevant risk group. Aggregate best estimate and results were calculated as weighted average ratio s of the corresponding statutory provisions. Maximum, minimum and unweighted average results across individual firm reports were calculated to reflect the range of outcomes in the ratio s. Aggregate results for the totals for the Life or Non-life portfolios are based on a line by line aggregation across the aggregated risk groups, excluding potential diversification effects between different risk groups. For those firms that reported the liability ratio s including diversification effects between risk groups, these ratio s were compared to the ratio s excluding diversification effects for those firm only that reported diversification effects. For the Life portfolio, 4 firms provided diversification effects across the risk groups. Only 1 firm provided an indication of diversification effects on a group basis. For the Non-life portfolio, only 1 firm provided diversification effects across groups of insurance classes. As a general aggregation method, the relevant weighted average ratio s were only calculated for those firms that provided the necessary data. Hence, for a given risk group the calculation of the best estimate ratio and the risk margins are not necessarily based on the same sample of firms. As a result, a comparison of the 60%, 75% and 90% margins might appear to reveal inconsistencies in the order of magnitude. 3. Quantitative Impact on Technical Provisions The effects of the application of a market consistent valuation have been assessed for the insurance liabilities. The following sections describe the comparison to the level of statutory technical provisions as of end 2004. 3.1 Impact on Life insurance provisions In general terms, the market-consistent valuation of liabilities is lower than the current level of liabilities. The average reduction is 9.79% when comparing the best estimate (including provisions for future bonuses) to the current level, a reduction of 7.92% for a 75 th margin and 6.11% for a 90 th margin. This effect is mainly attributable to discounting future liability cash flows at the higher risk-free rates (for the longer spectrum of the term structure) compared to the current technical interest rates for calculating current technical provisions. The following remarks need to be made: The comparisons indicate sufficient prudence in current technical provisions, except for the small portfolio of Life business without profit sharing. 8
The valuation of future bonuses and profit sharing has an important impact on the best estimate of liabilities. The risk margins (75% and 90% confidence levels) seem to add little prudence to the best estimate. The company view on risk margins (based on cost of capital approach) seem to be close to the best estimate of liabilities Diversification effects across risk groups lead to a marginal decrease in the risk margins of liabilities. The following table illustrates the aggregate impact on the different risk groups of the Life portfolio and on the total technical provisions. The aggregation is performed for those firms being able to provide best estimate figures and risk margin calculations. Summary of life insurance provisions gross of reinsurance Current bases Total liabilities (incl. future bonuses) Best estimates 4 Total liabilities (excl. future bonuses) 75 th 90 th Total Liabilities 60 th Company view on risk margins 5 Risk group (millions) % % % % % % Traditional Individual Business 18.983 86,68% 82,84% 87,81% 95,56% 70,50% 80,70% Traditional Group business 14.753 96,91% 89,37% 93,39% 95,84% 79,59% 92,35% Universal Life business with profit sharing 16.970 98,56% 96,29% 98,04% 99,99% 93,03% 98,72% Life business without profit sharing 262 116,02% 116,02% 117,82% 119,47% 100,06% 100,23% Unit Linked business 13.323 90,37% 90,37% 92,77% 91,17% 96,30% 90,22% Total 64.291 90,21% 73,38% 91,64% 93,52% 83,95% 92,49% Total including diversification effects 91,53% 93,30% 83,75% 93,66% Corresponding total for firms, excluding diversification effects 92,08% 93,89% 83,95% Table 6: Summary of life insurance provisions The following table illustrates the range in outcomes for the default 75 th risk margin, in terms of minimum, average and maximum ratio of current technical provisions: 4 Best estimate and liability ratios are calculated as a weigthed average of technical provisions on a current basis, only for those firms that were able to calculate the "relevant" liability (e.g. best estimate, 75%, 90%, 60% etc ) 5 Company view on risk margin calculated on a portfolio basis only, excluding group diversification effects,. 9
Range in outcomes for life insurance provisions gross of reinsurance Total 75 th Percentile Liabilities as % of Current provisions Minimum ratio Average ratio 6 Maximum ratio Standard deviation Risk group % % % % Traditional Individual Business 74,4% 86,2% 92,8% Traditional Group business 84,0% 93,4% 105,0% Universal Life business with profit sharing 85,5% 96,8% 103,3% Life business without profit sharing 101,0% 127,0% 153,0% Unit Linked business 92,0% 94,1% 96,7% Total 86,9% 91,5% 94,5% 4,1% Total, including diversification effects 86,6% 91,4% 94,5% 4,2% Table 7: Range in outcomes for life insurance provisions gross of reinsurance Again, the portfolio of Life business without profit sharing generates the greatest diversity across firms. This might be due to the limited size of the portfolio in general. 3.2 Impact on Non-Life insurance provisions For the Non-life portfolio a split-up is provided into provisions for unearned premiums (premium provisions) and claim provisions. However, the sum of both does not necessarily correspond to the total volume of provisions, as not all firms provided the split-up. The impact assessment consists of comparing the market-consistent valuation of liabilities to the current statutory basis for each type of provision (premium, claim, and total). The table below illustrates the split-up of current provisions across the different Non-life risk groups. Current basis (undiscounted) Summary of non-life insurance provisions gross of reinsurance Premium provisions Current Bases Claim provisions Total 7 Risk group (millions) (millions) (millions) Accident and health 22,4 615,8 695,8 Motor, third party liability 420,0 2.891,2 3.527,1 Motor, other classes 134,1 501,0 662,4 Marine, aviation and transport 0,4 5,3 5,7 Fire and other damage of property 347,4 518,1 1.138,8 Third-party liability 57,4 1.225,5 1.308,8 Legal expenses 25,7 159,0 184,9 Assistance 1,5 0,9 2,5 Miscellaneous non-life insurance 11,7 74,8 86,9 Total 1.020,6 5.991,6 7.612,8 Total equalisation provision 289,70 Percentage of provisions included 8 80,6% Table 8: Summary of non-life insurance provisions gross of reinsurance 6 Unweighted average ratio. 7 Total statutory technical provisions excluding equalisation provisions. 8 Weighted average of provisions included. 10
In general, the market-consistent valuation of liabilities is lower than the current level of liabilities (excluding amounts for equalisation provisions), both for premium provisions and claim provisions. The average reduction is 5.9% points on a best estimate undiscounted basis for premium provisions and 13.6% points for claim provisions. On a total basis the average reduction is 18% points. Before discounting Best estimates before Discounting 9 75th Percentile before Discounting Summary of non-life insurance provisions gross of reinsurance Premium provisions Claim provisions Total Premium provisions Claim provisions Risk group % % % % % % Accident and health 74,8% 88,4% 83,5% 100,0% 103,9% 97,8% Motor, third party liability 99,6% 85,8% 86,3% 102,6% 95,3% 91,0% Motor, other classes 96,2% 85,9% 85,4% 101,2% 90,0% 87,8% Marine, aviation and transport 0,0% 0,0% 93,7% 0,0% 0,0% 0,0% Fire and other damage of property 90,9% 83,7% 62,1% 76,7% 89,6% 65,3% Third-party liability 101,5% 84,1% 88,0% 100,8% 94,4% 91,6% Legal expenses 92,5% 120,3% 98,7% 92,4% 103,6% 104,3% Assistance 87,7% 99,3% 92,1% 0,0% 0,0% 0,0% Miscellaneous non-life insurance 75,6% 99,6% 95,2% 97,1% 108,0% 98,2% Total 94,1% 86,4% 82,0% 97,6% 94,5% 84,7% Table 9: Summary of non-life insurance provisions gross of reinsurance, undiscounted Total The following additional remarks need to be made: Market-consistent provisions for legal expense business are above the statutory reserves, which is typical for the long-tail nature of the business. The effects of discounting appear to be material, although not all firms provided data on a discounted basis. The effect of excluding equalisation provisions from the current basis has a limited impact on the best estimate and ratio s on a discounted basis. The results for the total best estimate and 75 th for each risk group are more reliable than the separate results for premium provisions and claim provision, as not all firms provided this split-up. After discounting Total Liabilities after discounting 10 Summary of non-life insurance provisions gross of reinsurance Best estimate 75 th 90 th Standard deviation 60 th Company view on risk margins Risk group % % % % % % Accident and health 68,5% 73,4% 78,1% 5,5% 0,0% 74,4% Motor, third party liability 70,6% 72,7% 74,8% 2,8% 78,8% 69,6% Motor, other classes 79,5% 81,8% 84,0% 5,5% 62,6% 65,0% 9 Best estimate premium provision and claim provision calculated as a weighted average ratio of the relevant current basis provision of table 4A 10 Best estimate and provisions for each risk group calculated as a weighted average ratio of the relevant current basis provision of table 4A 11
Marine, aviation and transport 63,6% 67,4% 71,5% 0,0% 0,0% 0,0% Fire and other damage of property 59,3% 61,1% 63,7% 6,8% 67,0% 62,5% Third-party liability 71,1% 73,2% 75,2% 3,3% 67,1% 75,9% Legal expenses 95,3% 98,6% 114,7% 6,4% 92,9% 85,9% Assistance 92,1% 95,0% 98,0% 0,0% 0,0% 0,0% Miscellaneous non-life insurance 89,7% 91,6% 93,8% 0,0% 0,0% 0,0% Total 11 74,2% 76,6% 79,1% 3,9% 74,2% 75,4% Comparison of the totals, including equalisation provision on current basis 70,8% 73,1% 75,5% 3,6% 71,9% 73,7% Table 10: Summary of non-life insurance provisions gross of reinsurance, discounted Summary of non-life insurance provisions gross of reinsurance Only claims provisions (after discounting) 12 Risk group Best estimate 75 th 90 th Standard deviation 60 th Company view on risk margins Accident and health 75,0% 78,1% 86,7% 0,0% 81,3% 89,6% Motor, third party liability 59,2% 79,2% 81,2% 4,2% 79,9% 82,2% Motor, other classes 80,1% 82,5% 84,9% 3,1% 41,6% 44,0% Marine, aviation and transport 0,0% 0,0% 0,0% 0,0% 0,0% 0,0% Fire and other damage of property 78,9% 82,1% 87,1% 9,2% 78,3% 91,4% Third-party liability 72,0% 73,7% 75,4% 3,9% 68,1% 77,0% Legal expenses 118,0% 119,6% 122,8% 7,1% 94,4% 90,3% Assistance 0,0% 0,0% 0,0% 0,0% 0,0% 0,0% Miscellaneous non-life insurance 96,1% 97,4% 98,7% 10,1% 0,0% 0,0% Total 76,9% 78,6% 80,8% 4,2% 76,9% 81,1% Table 11: Only claims provisions (after discounting) When comparing best estimate and outcomes for claims provisions only to current levels of technical provisions for claims outstanding, the order of magnitude in the results is similar to the total liability results of table 9. Not all firms modelled claims provisions on an individual risk group basis, but only on a whole portfolio basis. These outcomes were taken into account in the ratio s for total claims provisions. Total 75th Percentile Liabilities as % of current provisions Range of outcomes, Non-life insurance provisions, Standard gross of reinsurance Minimum ratio Average ratio(1) Maximum ratio deviation Non-life risk group % % % % Accident and health 71,8% 97,8% 145,3% Motor, third party liability 80,5% 91,4% 109,2% Motor, other classes 57,3% 78,2% 95,3% Marine, aviation and transport 97,2% 97,2% 97,2% Fire and other damage of property 59,0% 70,5% 81,5% Third-party liability 74,8% 92,8% 117,7% 11 Best estimate and provisions calculated as a weighted average ratio of the current basis provision excluding equalisation provision 12 Best estimate and provisions calculated as a weighted average ratio of the current basis claim provision 12
Legal expenses 92,5% 102,5% 112,4% Assistance 0,0% 0,0% 0,0% Miscellaneous non-life insurance 92,3% 119,7% 164,1% Total 80,2% 85,7% 95,3% 4,5% Table 12: Range of outcomes, Non-life insurance provisions, gross of reinsurance For the default 75 th liability ratio s the range in individual outcomes is substantial, even for the larger portfolio s like motor 3 rd party liability, fire & property, general 3 rd party liability. On an aggregate basis across all risk groups the range is narrower, owing possibly to the larger sample of firms involved. Total 75th Percentile Liabilities as % of current provisions Range of outcomes, Non-life claims provisions, Standard gross of reinsurance Minimum ratio Average ratio(1) Maximum ratio deviation Non-life risk group - only claims provisions Accident and health 62,0% 85,3% 124,5% Motor, third party liability 68,2% 81,6% 98,2% Motor, other classes 26,7% 54,6% 86,0% Marine, aviation and transport 0,0% 0,0% 0,0% Fire and other damage of property 56,4% 78,3% 95,6% Third-party liability 62,2% 77,2% 91,5% Legal expenses 83,8% 110,7% 136,7% Assistance 0,0% 0,0% 0,0% Miscellaneous non-life insurance 79,6% 122,5% 165,3% Total 71,9% 80,9% 90,6% 7,2% Table 13: Range of outcomes, Non-life claims provisions, gross of reinsurance Similarly diverse outcomes are obtained for the 75 th of the claims provisions ratio s. 4. Methodological assessment 4.1. Life insurance risk factors 4.1.1. Method applied to discounting All institutions used a term structure approach for discounting future liability cash flows. Some institutions made use of internally constructed curves based market data on the Euro swap rates at future valuation dates, but up to duration points corresponding to their portfolio. The missing intermediate rates were linearly interpolated between available duration points, and extrapolated beyond available market data points. However, the deviations from the CEIOPS prescribed rates were very marginal (maximum 3 bps). 4.1.2. Method applied to financial guarantees, future bonuses & profit sharing The methodologies used to model future financial guarantees, bonuses and profit sharing arrangements included: 13
For products exposed to financial guarantees (variable guaranteed rates), the liability cash flows are a function of future investment conditions. The value of such guarantees is derived using arbitrage-free pricing models, based on stochastically simulated future investment conditions. Discretionary bonuses are modeled as a function of the calculated investment return of the asset portfolio under simulated economic scenario s. In some cases, simulated investment returns on a benchmark portfolio are used instead of the return on the actual portfolio. Investment income above a certain threshold return is allocated to the contract reserve and translated into future benefits to policyholders. The value of such bonus options is then the net present value of such future benefits, averaged over the different economic scenarios. Some institutions use a more simplified profit sharing formula based on the weighted average of the smoothed 10 year bond yield and a smoothed equity return. However, due to the low interest rate environment the value of such profit sharing options is currently limited. The profit sharing formula reflects possible regime switching due to persistently low interest rates. 4.1.3. Method applied to surrender & lapse rates Assumed best estimate lapse rates are based on historical lapse experience of the institution s own portfolio. In other cases, lapse rates are deterministically modeled to take into account the product type, contract duration and potential tax treatment. More sophisticated stochastic approaches are also used, where surrender and lapse behaviour is modeled as a function of simulated economic scenarios. However, due to tax disincentive effects and severe surrender penalties of exercising surrender options, the rate of take-up of such options is assumed to be quite low. Other policyholder options (right to change contract conditions, conversion options to other insurance forms, etc.) are not modeled. 4.1.4. Method applied to mortality and morbidity rates Best estimate assumptions on mortality and morbidity rates were mostly based on national industry survival tables, but are in most cases adjusted to take into account the historical mortality experience in their own portfolio. Some other institutions adjusted these actuarial rates to take into account a margin for uncertainty in the estimated level and volatility of mortality, future mortality and morbidity trends as well as calamity scenarios of extreme positive fluctuations in mortality risk. 4.1.5. Method applied to expenses Future cash flow modelling takes into account overheads, acquisition costs, operating and investment management expenses in as far as they can be directly or indirectly attributed to the book of contracts. The best estimate expense assumptions are based on actual expense developments and take into account the effect of changing business volume on future expenses. Where inflation assumptions have been made, these are based on market expectations for the near future (up to 2007) and a flat rate of no more than 2% as of 2008. 14
4.1.6. Risk margin estimates The risk margins are derived from the empirical distributions of liability cash flows obtained through variation (Monte Carlo simulations) of the underlying risk factors. Due to time and resource constraints not all firms were able to report the requested margins. In addition, not a single firm favoured a approach on the grounds that a approach is inconsistent with a market valuation approach. In effect, using the same level of prudence for all business lines arbitrarily introduces inconsistencies as some risks are more skewed than others. As some firms did not want to signal a support of the approach, risk margins were calculated using their internal cost of capital approach. Under this approach the market value margin corresponds to the present value of the cost of future economic capital required to run-off all insurance liabilities. To calculate the market value margin (MVM) the following practical approach is widely adopted: Calculate the economic capital required to cover unhedgeable financial and non-financial risks in the current insurance portfolio Project the economic capital for the full run off of the portfolio Discount the stream of economic capital levels at an appropriate term structure to calculate the present value of economic capital Apply a fixed cost of capital charge (spread over risk-free rate) to the present value of economic capital to calculate the MVM Table 14 below provides some additional information on the proportion of firms using a particular methodology for the Life portfolio. Average duration of portfolio Firms applying term approach to discounting Firms applying duration approach to discounting Firms able to assess risk margins by simulation Life insurance Years % % % Traditional Individual Business 10,92 Traditional Group business 22,05 Universal Life business with profit sharing 10,35 Life business without profit sharing 10,28 Unit Linked business Firms able to calculate market value of options and guarantees % Total 100,0% 0,0% 85,7% 85,7% Table 14: Miscellaneous methodological information Life 4.2. Non-life insurance risk factors 4.2.1. Method applied to claim provisions Best estimate valuations of claim provisions were statistically derived from past claims payment or claims incurrence data on an occurrence and development year basis. Run-off triangle methods (chain ladder method) were used to project future payment patterns. The number of years developed in the tails of the triangles depends on the tail characteristics of each line of business. Extreme large claims are typically excluded from the triangles, and are reserved on a case-by-case basis. Future 15
claims volume inflation and claims handling costs are based on historical trends. Where discounting was applied, the risk-free, duration dependent term structure was used. 4.2.2. Method applied to premium provisions The best estimate for the unearned premium provision is deterministically (prorata temporis) calculated as the unconsumed portion of emitted premiums on a case-by-case basis. For current accident year claims exceeding earned premiums for the same accident year, the unexpired risk provision is calculated as the loss ratio (or combined ratio)*unearned premium provision. Such simplified approaches assume that the portion of risk corresponding to the unearned premium has the same run-off pattern as the claims for the earned portion of premium. 4.2.3. Risk margin estimates Bootstrapping techniques are used on paid claims triangles to derive coefficients of variation (CoV) in payments for each historical triangle, assuming a (log)-normal distribution of the claim provisions. The levels of the s (60%, 75%, 90%) are derived from stressing CoV*Best estimate to achieve the corresponding confidence levels of the empirical distribution of claim provisions. Table 15 below provides some additional information on the proportion of firms using a particular methodology for the Non-life portfolio. Average duration of portfolio Firms applying term approach to discounting Firms applying duration approach to discounting Firms assessing risk margins by simulation Non-life insurance Years % % % Accident and health 5,51 Motor, third party liability 4,14 Motor, other classes 3,19 Marine, aviation and transport 0 Fire and other damage of property 2,32 Third-party liability 5,09 Legal expenses 4,01 Assistance 0 Miscellaneous non-life insurance 4,37 Firms calculating market value of options and guarantees % Total 100,0% 0,0% 100,0% 0,0% Table 15:Miscellaneous methodological information Non-life 16